ip_input.c revision 1.114.4.7
1 /* $NetBSD: ip_input.c,v 1.114.4.7 2002/02/26 20:57:03 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 ip_forward(m, 0); 628 } 629 return; 630 631 ours: 632 /* 633 * If offset or IP_MF are set, must reassemble. 634 * Otherwise, nothing need be done. 635 * (We could look in the reassembly queue to see 636 * if the packet was previously fragmented, 637 * but it's not worth the time; just let them time out.) 638 */ 639 if (ip->ip_off & ~(IP_DF|IP_RF)) { 640 /* 641 * Look for queue of fragments 642 * of this datagram. 643 */ 644 IPQ_LOCK(); 645 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 646 if (ip->ip_id == fp->ipq_id && 647 in_hosteq(ip->ip_src, fp->ipq_src) && 648 in_hosteq(ip->ip_dst, fp->ipq_dst) && 649 ip->ip_p == fp->ipq_p) 650 goto found; 651 fp = 0; 652 found: 653 654 /* 655 * Adjust ip_len to not reflect header, 656 * set ipqe_mff if more fragments are expected, 657 * convert offset of this to bytes. 658 */ 659 ip->ip_len -= hlen; 660 mff = (ip->ip_off & IP_MF) != 0; 661 if (mff) { 662 /* 663 * Make sure that fragments have a data length 664 * that's a non-zero multiple of 8 bytes. 665 */ 666 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 667 ipstat.ips_badfrags++; 668 IPQ_UNLOCK(); 669 goto bad; 670 } 671 } 672 ip->ip_off <<= 3; 673 674 /* 675 * If datagram marked as having more fragments 676 * or if this is not the first fragment, 677 * attempt reassembly; if it succeeds, proceed. 678 */ 679 if (mff || ip->ip_off) { 680 ipstat.ips_fragments++; 681 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 682 if (ipqe == NULL) { 683 ipstat.ips_rcvmemdrop++; 684 IPQ_UNLOCK(); 685 goto bad; 686 } 687 ipqe->ipqe_mff = mff; 688 ipqe->ipqe_m = m; 689 ipqe->ipqe_ip = ip; 690 m = ip_reass(ipqe, fp); 691 if (m == 0) { 692 IPQ_UNLOCK(); 693 return; 694 } 695 ipstat.ips_reassembled++; 696 ip = mtod(m, struct ip *); 697 hlen = ip->ip_hl << 2; 698 ip->ip_len += hlen; 699 } else 700 if (fp) 701 ip_freef(fp); 702 IPQ_UNLOCK(); 703 } 704 705 #ifdef IPSEC 706 /* 707 * enforce IPsec policy checking if we are seeing last header. 708 * note that we do not visit this with protocols with pcb layer 709 * code - like udp/tcp/raw ip. 710 */ 711 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 712 ipsec4_in_reject(m, NULL)) { 713 ipsecstat.in_polvio++; 714 goto bad; 715 } 716 #endif 717 718 /* 719 * Switch out to protocol's input routine. 720 */ 721 #if IFA_STATS 722 ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; 723 #endif 724 ipstat.ips_delivered++; 725 { 726 int off = hlen, nh = ip->ip_p; 727 728 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 729 return; 730 } 731 bad: 732 m_freem(m); 733 } 734 735 /* 736 * Take incoming datagram fragment and try to 737 * reassemble it into whole datagram. If a chain for 738 * reassembly of this datagram already exists, then it 739 * is given as fp; otherwise have to make a chain. 740 */ 741 struct mbuf * 742 ip_reass(ipqe, fp) 743 struct ipqent *ipqe; 744 struct ipq *fp; 745 { 746 struct mbuf *m = ipqe->ipqe_m; 747 struct ipqent *nq, *p, *q; 748 struct ip *ip; 749 struct mbuf *t; 750 int hlen = ipqe->ipqe_ip->ip_hl << 2; 751 int i, next; 752 753 IPQ_LOCK_CHECK(); 754 755 /* 756 * Presence of header sizes in mbufs 757 * would confuse code below. 758 */ 759 m->m_data += hlen; 760 m->m_len -= hlen; 761 762 /* 763 * If first fragment to arrive, create a reassembly queue. 764 */ 765 if (fp == 0) { 766 /* 767 * Enforce upper bound on number of fragmented packets 768 * for which we attempt reassembly; 769 * If maxfrag is 0, never accept fragments. 770 * If maxfrag is -1, accept all fragments without limitation. 771 */ 772 if (ip_maxfragpackets < 0) 773 ; 774 else if (ip_nfragpackets >= ip_maxfragpackets) 775 goto dropfrag; 776 ip_nfragpackets++; 777 MALLOC(fp, struct ipq *, sizeof (struct ipq), 778 M_FTABLE, M_NOWAIT); 779 if (fp == NULL) 780 goto dropfrag; 781 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 782 fp->ipq_ttl = IPFRAGTTL; 783 fp->ipq_p = ipqe->ipqe_ip->ip_p; 784 fp->ipq_id = ipqe->ipqe_ip->ip_id; 785 LIST_INIT(&fp->ipq_fragq); 786 fp->ipq_src = ipqe->ipqe_ip->ip_src; 787 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 788 p = NULL; 789 goto insert; 790 } 791 792 /* 793 * Find a segment which begins after this one does. 794 */ 795 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 796 p = q, q = q->ipqe_q.le_next) 797 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 798 break; 799 800 /* 801 * If there is a preceding segment, it may provide some of 802 * our data already. If so, drop the data from the incoming 803 * segment. If it provides all of our data, drop us. 804 */ 805 if (p != NULL) { 806 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 807 ipqe->ipqe_ip->ip_off; 808 if (i > 0) { 809 if (i >= ipqe->ipqe_ip->ip_len) 810 goto dropfrag; 811 m_adj(ipqe->ipqe_m, i); 812 ipqe->ipqe_ip->ip_off += i; 813 ipqe->ipqe_ip->ip_len -= i; 814 } 815 } 816 817 /* 818 * While we overlap succeeding segments trim them or, 819 * if they are completely covered, dequeue them. 820 */ 821 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 822 q->ipqe_ip->ip_off; q = nq) { 823 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 824 q->ipqe_ip->ip_off; 825 if (i < q->ipqe_ip->ip_len) { 826 q->ipqe_ip->ip_len -= i; 827 q->ipqe_ip->ip_off += i; 828 m_adj(q->ipqe_m, i); 829 break; 830 } 831 nq = q->ipqe_q.le_next; 832 m_freem(q->ipqe_m); 833 LIST_REMOVE(q, ipqe_q); 834 pool_put(&ipqent_pool, q); 835 } 836 837 insert: 838 /* 839 * Stick new segment in its place; 840 * check for complete reassembly. 841 */ 842 if (p == NULL) { 843 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 844 } else { 845 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 846 } 847 next = 0; 848 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 849 p = q, q = q->ipqe_q.le_next) { 850 if (q->ipqe_ip->ip_off != next) 851 return (0); 852 next += q->ipqe_ip->ip_len; 853 } 854 if (p->ipqe_mff) 855 return (0); 856 857 /* 858 * Reassembly is complete. Check for a bogus message size and 859 * concatenate fragments. 860 */ 861 q = fp->ipq_fragq.lh_first; 862 ip = q->ipqe_ip; 863 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 864 ipstat.ips_toolong++; 865 ip_freef(fp); 866 return (0); 867 } 868 m = q->ipqe_m; 869 t = m->m_next; 870 m->m_next = 0; 871 m_cat(m, t); 872 nq = q->ipqe_q.le_next; 873 pool_put(&ipqent_pool, q); 874 for (q = nq; q != NULL; q = nq) { 875 t = q->ipqe_m; 876 nq = q->ipqe_q.le_next; 877 pool_put(&ipqent_pool, q); 878 m_cat(m, t); 879 } 880 881 /* 882 * Create header for new ip packet by 883 * modifying header of first packet; 884 * dequeue and discard fragment reassembly header. 885 * Make header visible. 886 */ 887 ip->ip_len = next; 888 ip->ip_src = fp->ipq_src; 889 ip->ip_dst = fp->ipq_dst; 890 LIST_REMOVE(fp, ipq_q); 891 FREE(fp, M_FTABLE); 892 ip_nfragpackets--; 893 m->m_len += (ip->ip_hl << 2); 894 m->m_data -= (ip->ip_hl << 2); 895 /* some debugging cruft by sklower, below, will go away soon */ 896 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 897 int plen = 0; 898 for (t = m; t; t = t->m_next) 899 plen += t->m_len; 900 m->m_pkthdr.len = plen; 901 } 902 return (m); 903 904 dropfrag: 905 ipstat.ips_fragdropped++; 906 m_freem(m); 907 pool_put(&ipqent_pool, ipqe); 908 return (0); 909 } 910 911 /* 912 * Free a fragment reassembly header and all 913 * associated datagrams. 914 */ 915 void 916 ip_freef(fp) 917 struct ipq *fp; 918 { 919 struct ipqent *q, *p; 920 921 IPQ_LOCK_CHECK(); 922 923 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 924 p = q->ipqe_q.le_next; 925 m_freem(q->ipqe_m); 926 LIST_REMOVE(q, ipqe_q); 927 pool_put(&ipqent_pool, q); 928 } 929 LIST_REMOVE(fp, ipq_q); 930 FREE(fp, M_FTABLE); 931 ip_nfragpackets--; 932 } 933 934 /* 935 * IP timer processing; 936 * if a timer expires on a reassembly 937 * queue, discard it. 938 */ 939 void 940 ip_slowtimo() 941 { 942 struct ipq *fp, *nfp; 943 int s = splsoftnet(); 944 945 IPQ_LOCK(); 946 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 947 nfp = fp->ipq_q.le_next; 948 if (--fp->ipq_ttl == 0) { 949 ipstat.ips_fragtimeout++; 950 ip_freef(fp); 951 } 952 } 953 /* 954 * If we are over the maximum number of fragments 955 * (due to the limit being lowered), drain off 956 * enough to get down to the new limit. 957 */ 958 if (ip_maxfragpackets < 0) 959 ; 960 else { 961 while (ip_nfragpackets > ip_maxfragpackets && ipq.lh_first) 962 ip_freef(ipq.lh_first); 963 } 964 IPQ_UNLOCK(); 965 #ifdef GATEWAY 966 ipflow_slowtimo(); 967 #endif 968 splx(s); 969 } 970 971 /* 972 * Drain off all datagram fragments. 973 */ 974 void 975 ip_drain() 976 { 977 978 /* 979 * We may be called from a device's interrupt context. If 980 * the ipq is already busy, just bail out now. 981 */ 982 if (ipq_lock_try() == 0) 983 return; 984 985 while (ipq.lh_first != NULL) { 986 ipstat.ips_fragdropped++; 987 ip_freef(ipq.lh_first); 988 } 989 990 IPQ_UNLOCK(); 991 } 992 993 /* 994 * Do option processing on a datagram, 995 * possibly discarding it if bad options are encountered, 996 * or forwarding it if source-routed. 997 * Returns 1 if packet has been forwarded/freed, 998 * 0 if the packet should be processed further. 999 */ 1000 int 1001 ip_dooptions(m) 1002 struct mbuf *m; 1003 { 1004 struct ip *ip = mtod(m, struct ip *); 1005 u_char *cp, *cp0; 1006 struct ip_timestamp *ipt; 1007 struct in_ifaddr *ia; 1008 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1009 struct in_addr dst; 1010 n_time ntime; 1011 1012 dst = ip->ip_dst; 1013 cp = (u_char *)(ip + 1); 1014 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1015 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1016 opt = cp[IPOPT_OPTVAL]; 1017 if (opt == IPOPT_EOL) 1018 break; 1019 if (opt == IPOPT_NOP) 1020 optlen = 1; 1021 else { 1022 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1023 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1024 goto bad; 1025 } 1026 optlen = cp[IPOPT_OLEN]; 1027 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1028 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1029 goto bad; 1030 } 1031 } 1032 switch (opt) { 1033 1034 default: 1035 break; 1036 1037 /* 1038 * Source routing with record. 1039 * Find interface with current destination address. 1040 * If none on this machine then drop if strictly routed, 1041 * or do nothing if loosely routed. 1042 * Record interface address and bring up next address 1043 * component. If strictly routed make sure next 1044 * address is on directly accessible net. 1045 */ 1046 case IPOPT_LSRR: 1047 case IPOPT_SSRR: 1048 if (ip_allowsrcrt == 0) { 1049 type = ICMP_UNREACH; 1050 code = ICMP_UNREACH_NET_PROHIB; 1051 goto bad; 1052 } 1053 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1054 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1055 goto bad; 1056 } 1057 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1058 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1059 goto bad; 1060 } 1061 ipaddr.sin_addr = ip->ip_dst; 1062 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1063 if (ia == 0) { 1064 if (opt == IPOPT_SSRR) { 1065 type = ICMP_UNREACH; 1066 code = ICMP_UNREACH_SRCFAIL; 1067 goto bad; 1068 } 1069 /* 1070 * Loose routing, and not at next destination 1071 * yet; nothing to do except forward. 1072 */ 1073 break; 1074 } 1075 off--; /* 0 origin */ 1076 if ((off + sizeof(struct in_addr)) > optlen) { 1077 /* 1078 * End of source route. Should be for us. 1079 */ 1080 save_rte(cp, ip->ip_src); 1081 break; 1082 } 1083 /* 1084 * locate outgoing interface 1085 */ 1086 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1087 sizeof(ipaddr.sin_addr)); 1088 if (opt == IPOPT_SSRR) 1089 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1090 else 1091 ia = ip_rtaddr(ipaddr.sin_addr); 1092 if (ia == 0) { 1093 type = ICMP_UNREACH; 1094 code = ICMP_UNREACH_SRCFAIL; 1095 goto bad; 1096 } 1097 ip->ip_dst = ipaddr.sin_addr; 1098 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1099 (caddr_t)(cp + off), sizeof(struct in_addr)); 1100 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1101 /* 1102 * Let ip_intr's mcast routing check handle mcast pkts 1103 */ 1104 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1105 break; 1106 1107 case IPOPT_RR: 1108 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1109 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1110 goto bad; 1111 } 1112 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1113 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1114 goto bad; 1115 } 1116 /* 1117 * If no space remains, ignore. 1118 */ 1119 off--; /* 0 origin */ 1120 if ((off + sizeof(struct in_addr)) > optlen) 1121 break; 1122 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1123 sizeof(ipaddr.sin_addr)); 1124 /* 1125 * locate outgoing interface; if we're the destination, 1126 * use the incoming interface (should be same). 1127 */ 1128 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1129 == NULL && 1130 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1131 type = ICMP_UNREACH; 1132 code = ICMP_UNREACH_HOST; 1133 goto bad; 1134 } 1135 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1136 (caddr_t)(cp + off), sizeof(struct in_addr)); 1137 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1138 break; 1139 1140 case IPOPT_TS: 1141 code = cp - (u_char *)ip; 1142 ipt = (struct ip_timestamp *)cp; 1143 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1144 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1145 goto bad; 1146 } 1147 if (ipt->ipt_ptr < 5) { 1148 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1149 goto bad; 1150 } 1151 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1152 if (++ipt->ipt_oflw == 0) { 1153 code = (u_char *)&ipt->ipt_ptr - 1154 (u_char *)ip; 1155 goto bad; 1156 } 1157 break; 1158 } 1159 cp0 = (cp + ipt->ipt_ptr - 1); 1160 switch (ipt->ipt_flg) { 1161 1162 case IPOPT_TS_TSONLY: 1163 break; 1164 1165 case IPOPT_TS_TSANDADDR: 1166 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1167 sizeof(struct in_addr) > ipt->ipt_len) { 1168 code = (u_char *)&ipt->ipt_ptr - 1169 (u_char *)ip; 1170 goto bad; 1171 } 1172 ipaddr.sin_addr = dst; 1173 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1174 m->m_pkthdr.rcvif)); 1175 if (ia == 0) 1176 continue; 1177 bcopy(&ia->ia_addr.sin_addr, 1178 cp0, sizeof(struct in_addr)); 1179 ipt->ipt_ptr += sizeof(struct in_addr); 1180 break; 1181 1182 case IPOPT_TS_PRESPEC: 1183 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1184 sizeof(struct in_addr) > ipt->ipt_len) { 1185 code = (u_char *)&ipt->ipt_ptr - 1186 (u_char *)ip; 1187 goto bad; 1188 } 1189 bcopy(cp0, &ipaddr.sin_addr, 1190 sizeof(struct in_addr)); 1191 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1192 == NULL) 1193 continue; 1194 ipt->ipt_ptr += sizeof(struct in_addr); 1195 break; 1196 1197 default: 1198 /* XXX can't take &ipt->ipt_flg */ 1199 code = (u_char *)&ipt->ipt_ptr - 1200 (u_char *)ip + 1; 1201 goto bad; 1202 } 1203 ntime = iptime(); 1204 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1205 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1206 sizeof(n_time)); 1207 ipt->ipt_ptr += sizeof(n_time); 1208 } 1209 } 1210 if (forward) { 1211 if (ip_forwsrcrt == 0) { 1212 type = ICMP_UNREACH; 1213 code = ICMP_UNREACH_SRCFAIL; 1214 goto bad; 1215 } 1216 ip_forward(m, 1); 1217 return (1); 1218 } 1219 return (0); 1220 bad: 1221 icmp_error(m, type, code, 0, 0); 1222 ipstat.ips_badoptions++; 1223 return (1); 1224 } 1225 1226 /* 1227 * Given address of next destination (final or next hop), 1228 * return internet address info of interface to be used to get there. 1229 */ 1230 struct in_ifaddr * 1231 ip_rtaddr(dst) 1232 struct in_addr dst; 1233 { 1234 struct sockaddr_in *sin; 1235 1236 sin = satosin(&ipforward_rt.ro_dst); 1237 1238 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1239 if (ipforward_rt.ro_rt) { 1240 RTFREE(ipforward_rt.ro_rt); 1241 ipforward_rt.ro_rt = 0; 1242 } 1243 sin->sin_family = AF_INET; 1244 sin->sin_len = sizeof(*sin); 1245 sin->sin_addr = dst; 1246 1247 rtalloc(&ipforward_rt); 1248 } 1249 if (ipforward_rt.ro_rt == 0) 1250 return ((struct in_ifaddr *)0); 1251 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1252 } 1253 1254 /* 1255 * Save incoming source route for use in replies, 1256 * to be picked up later by ip_srcroute if the receiver is interested. 1257 */ 1258 void 1259 save_rte(option, dst) 1260 u_char *option; 1261 struct in_addr dst; 1262 { 1263 unsigned olen; 1264 1265 olen = option[IPOPT_OLEN]; 1266 #ifdef DIAGNOSTIC 1267 if (ipprintfs) 1268 printf("save_rte: olen %d\n", olen); 1269 #endif /* 0 */ 1270 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1271 return; 1272 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1273 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1274 ip_srcrt.dst = dst; 1275 } 1276 1277 /* 1278 * Retrieve incoming source route for use in replies, 1279 * in the same form used by setsockopt. 1280 * The first hop is placed before the options, will be removed later. 1281 */ 1282 struct mbuf * 1283 ip_srcroute() 1284 { 1285 struct in_addr *p, *q; 1286 struct mbuf *m; 1287 1288 if (ip_nhops == 0) 1289 return ((struct mbuf *)0); 1290 m = m_get(M_DONTWAIT, MT_SOOPTS); 1291 if (m == 0) 1292 return ((struct mbuf *)0); 1293 1294 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1295 1296 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1297 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1298 OPTSIZ; 1299 #ifdef DIAGNOSTIC 1300 if (ipprintfs) 1301 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1302 #endif 1303 1304 /* 1305 * First save first hop for return route 1306 */ 1307 p = &ip_srcrt.route[ip_nhops - 1]; 1308 *(mtod(m, struct in_addr *)) = *p--; 1309 #ifdef DIAGNOSTIC 1310 if (ipprintfs) 1311 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1312 #endif 1313 1314 /* 1315 * Copy option fields and padding (nop) to mbuf. 1316 */ 1317 ip_srcrt.nop = IPOPT_NOP; 1318 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1319 bcopy((caddr_t)&ip_srcrt.nop, 1320 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1321 q = (struct in_addr *)(mtod(m, caddr_t) + 1322 sizeof(struct in_addr) + OPTSIZ); 1323 #undef OPTSIZ 1324 /* 1325 * Record return path as an IP source route, 1326 * reversing the path (pointers are now aligned). 1327 */ 1328 while (p >= ip_srcrt.route) { 1329 #ifdef DIAGNOSTIC 1330 if (ipprintfs) 1331 printf(" %x", ntohl(q->s_addr)); 1332 #endif 1333 *q++ = *p--; 1334 } 1335 /* 1336 * Last hop goes to final destination. 1337 */ 1338 *q = ip_srcrt.dst; 1339 #ifdef DIAGNOSTIC 1340 if (ipprintfs) 1341 printf(" %x\n", ntohl(q->s_addr)); 1342 #endif 1343 return (m); 1344 } 1345 1346 /* 1347 * Strip out IP options, at higher 1348 * level protocol in the kernel. 1349 * Second argument is buffer to which options 1350 * will be moved, and return value is their length. 1351 * XXX should be deleted; last arg currently ignored. 1352 */ 1353 void 1354 ip_stripoptions(m, mopt) 1355 struct mbuf *m; 1356 struct mbuf *mopt; 1357 { 1358 int i; 1359 struct ip *ip = mtod(m, struct ip *); 1360 caddr_t opts; 1361 int olen; 1362 1363 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1364 opts = (caddr_t)(ip + 1); 1365 i = m->m_len - (sizeof (struct ip) + olen); 1366 bcopy(opts + olen, opts, (unsigned)i); 1367 m->m_len -= olen; 1368 if (m->m_flags & M_PKTHDR) 1369 m->m_pkthdr.len -= olen; 1370 ip->ip_len -= olen; 1371 ip->ip_hl = sizeof (struct ip) >> 2; 1372 } 1373 1374 int inetctlerrmap[PRC_NCMDS] = { 1375 0, 0, 0, 0, 1376 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1377 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1378 EMSGSIZE, EHOSTUNREACH, 0, 0, 1379 0, 0, 0, 0, 1380 ENOPROTOOPT 1381 }; 1382 1383 /* 1384 * Forward a packet. If some error occurs return the sender 1385 * an icmp packet. Note we can't always generate a meaningful 1386 * icmp message because icmp doesn't have a large enough repertoire 1387 * of codes and types. 1388 * 1389 * If not forwarding, just drop the packet. This could be confusing 1390 * if ipforwarding was zero but some routing protocol was advancing 1391 * us as a gateway to somewhere. However, we must let the routing 1392 * protocol deal with that. 1393 * 1394 * The srcrt parameter indicates whether the packet is being forwarded 1395 * via a source route. 1396 */ 1397 void 1398 ip_forward(m, srcrt) 1399 struct mbuf *m; 1400 int srcrt; 1401 { 1402 struct ip *ip = mtod(m, struct ip *); 1403 struct sockaddr_in *sin; 1404 struct rtentry *rt; 1405 int error, type = 0, code = 0; 1406 struct mbuf *mcopy; 1407 n_long dest; 1408 struct ifnet *destifp; 1409 #ifdef IPSEC 1410 struct ifnet dummyifp; 1411 #endif 1412 1413 dest = 0; 1414 #ifdef DIAGNOSTIC 1415 if (ipprintfs) 1416 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1417 ntohl(ip->ip_src.s_addr), 1418 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1419 #endif 1420 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1421 ipstat.ips_cantforward++; 1422 m_freem(m); 1423 return; 1424 } 1425 if (ip->ip_ttl <= IPTTLDEC) { 1426 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1427 return; 1428 } 1429 ip->ip_ttl -= IPTTLDEC; 1430 1431 sin = satosin(&ipforward_rt.ro_dst); 1432 if ((rt = ipforward_rt.ro_rt) == 0 || 1433 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1434 if (ipforward_rt.ro_rt) { 1435 RTFREE(ipforward_rt.ro_rt); 1436 ipforward_rt.ro_rt = 0; 1437 } 1438 sin->sin_family = AF_INET; 1439 sin->sin_len = sizeof(struct sockaddr_in); 1440 sin->sin_addr = ip->ip_dst; 1441 1442 rtalloc(&ipforward_rt); 1443 if (ipforward_rt.ro_rt == 0) { 1444 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1445 return; 1446 } 1447 rt = ipforward_rt.ro_rt; 1448 } 1449 1450 /* 1451 * Save at most 68 bytes of the packet in case 1452 * we need to generate an ICMP message to the src. 1453 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1454 */ 1455 mcopy = m_copym(m, 0, imin((int)ip->ip_len, 68), M_DONTWAIT); 1456 if (mcopy) 1457 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1458 1459 /* 1460 * If forwarding packet using same interface that it came in on, 1461 * perhaps should send a redirect to sender to shortcut a hop. 1462 * Only send redirect if source is sending directly to us, 1463 * and if packet was not source routed (or has any options). 1464 * Also, don't send redirect if forwarding using a default route 1465 * or a route modified by a redirect. 1466 */ 1467 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1468 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1469 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1470 ipsendredirects && !srcrt) { 1471 if (rt->rt_ifa && 1472 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1473 ifatoia(rt->rt_ifa)->ia_subnet) { 1474 if (rt->rt_flags & RTF_GATEWAY) 1475 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1476 else 1477 dest = ip->ip_dst.s_addr; 1478 /* 1479 * Router requirements says to only send host 1480 * redirects. 1481 */ 1482 type = ICMP_REDIRECT; 1483 code = ICMP_REDIRECT_HOST; 1484 #ifdef DIAGNOSTIC 1485 if (ipprintfs) 1486 printf("redirect (%d) to %x\n", code, 1487 (u_int32_t)dest); 1488 #endif 1489 } 1490 } 1491 1492 #ifdef IPSEC 1493 /* Don't lookup socket in forwading case */ 1494 (void)ipsec_setsocket(m, NULL); 1495 #endif 1496 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1497 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1498 if (error) 1499 ipstat.ips_cantforward++; 1500 else { 1501 ipstat.ips_forward++; 1502 if (type) 1503 ipstat.ips_redirectsent++; 1504 else { 1505 if (mcopy) { 1506 #ifdef GATEWAY 1507 if (mcopy->m_flags & M_CANFASTFWD) 1508 ipflow_create(&ipforward_rt, mcopy); 1509 #endif 1510 m_freem(mcopy); 1511 } 1512 return; 1513 } 1514 } 1515 if (mcopy == NULL) 1516 return; 1517 destifp = NULL; 1518 1519 switch (error) { 1520 1521 case 0: /* forwarded, but need redirect */ 1522 /* type, code set above */ 1523 break; 1524 1525 case ENETUNREACH: /* shouldn't happen, checked above */ 1526 case EHOSTUNREACH: 1527 case ENETDOWN: 1528 case EHOSTDOWN: 1529 default: 1530 type = ICMP_UNREACH; 1531 code = ICMP_UNREACH_HOST; 1532 break; 1533 1534 case EMSGSIZE: 1535 type = ICMP_UNREACH; 1536 code = ICMP_UNREACH_NEEDFRAG; 1537 #ifndef IPSEC 1538 if (ipforward_rt.ro_rt) 1539 destifp = ipforward_rt.ro_rt->rt_ifp; 1540 #else 1541 /* 1542 * If the packet is routed over IPsec tunnel, tell the 1543 * originator the tunnel MTU. 1544 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1545 * XXX quickhack!!! 1546 */ 1547 if (ipforward_rt.ro_rt) { 1548 struct secpolicy *sp; 1549 int ipsecerror; 1550 size_t ipsechdr; 1551 struct route *ro; 1552 1553 sp = ipsec4_getpolicybyaddr(mcopy, 1554 IPSEC_DIR_OUTBOUND, 1555 IP_FORWARDING, 1556 &ipsecerror); 1557 1558 if (sp == NULL) 1559 destifp = ipforward_rt.ro_rt->rt_ifp; 1560 else { 1561 /* count IPsec header size */ 1562 ipsechdr = ipsec4_hdrsiz(mcopy, 1563 IPSEC_DIR_OUTBOUND, 1564 NULL); 1565 1566 /* 1567 * find the correct route for outer IPv4 1568 * header, compute tunnel MTU. 1569 * 1570 * XXX BUG ALERT 1571 * The "dummyifp" code relies upon the fact 1572 * that icmp_error() touches only ifp->if_mtu. 1573 */ 1574 /*XXX*/ 1575 destifp = NULL; 1576 if (sp->req != NULL 1577 && sp->req->sav != NULL 1578 && sp->req->sav->sah != NULL) { 1579 ro = &sp->req->sav->sah->sa_route; 1580 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1581 dummyifp.if_mtu = 1582 ro->ro_rt->rt_ifp->if_mtu; 1583 dummyifp.if_mtu -= ipsechdr; 1584 destifp = &dummyifp; 1585 } 1586 } 1587 1588 key_freesp(sp); 1589 } 1590 } 1591 #endif /*IPSEC*/ 1592 ipstat.ips_cantfrag++; 1593 break; 1594 1595 case ENOBUFS: 1596 type = ICMP_SOURCEQUENCH; 1597 code = 0; 1598 break; 1599 } 1600 icmp_error(mcopy, type, code, dest, destifp); 1601 } 1602 1603 void 1604 ip_savecontrol(inp, mp, ip, m) 1605 struct inpcb *inp; 1606 struct mbuf **mp; 1607 struct ip *ip; 1608 struct mbuf *m; 1609 { 1610 1611 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1612 struct timeval tv; 1613 1614 microtime(&tv); 1615 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1616 SCM_TIMESTAMP, SOL_SOCKET); 1617 if (*mp) 1618 mp = &(*mp)->m_next; 1619 } 1620 if (inp->inp_flags & INP_RECVDSTADDR) { 1621 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1622 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1623 if (*mp) 1624 mp = &(*mp)->m_next; 1625 } 1626 #ifdef notyet 1627 /* 1628 * XXX 1629 * Moving these out of udp_input() made them even more broken 1630 * than they already were. 1631 * - fenner (at) parc.xerox.com 1632 */ 1633 /* options were tossed already */ 1634 if (inp->inp_flags & INP_RECVOPTS) { 1635 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1636 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1637 if (*mp) 1638 mp = &(*mp)->m_next; 1639 } 1640 /* ip_srcroute doesn't do what we want here, need to fix */ 1641 if (inp->inp_flags & INP_RECVRETOPTS) { 1642 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1643 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1644 if (*mp) 1645 mp = &(*mp)->m_next; 1646 } 1647 #endif 1648 if (inp->inp_flags & INP_RECVIF) { 1649 struct sockaddr_dl sdl; 1650 1651 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1652 sdl.sdl_family = AF_LINK; 1653 sdl.sdl_index = m->m_pkthdr.rcvif ? 1654 m->m_pkthdr.rcvif->if_index : 0; 1655 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1656 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1657 IP_RECVIF, IPPROTO_IP); 1658 if (*mp) 1659 mp = &(*mp)->m_next; 1660 } 1661 } 1662 1663 int 1664 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1665 int *name; 1666 u_int namelen; 1667 void *oldp; 1668 size_t *oldlenp; 1669 void *newp; 1670 size_t newlen; 1671 { 1672 extern int subnetsarelocal, hostzeroisbroadcast; 1673 1674 int error, old; 1675 1676 /* All sysctl names at this level are terminal. */ 1677 if (namelen != 1) 1678 return (ENOTDIR); 1679 1680 switch (name[0]) { 1681 case IPCTL_FORWARDING: 1682 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1683 case IPCTL_SENDREDIRECTS: 1684 return (sysctl_int(oldp, oldlenp, newp, newlen, 1685 &ipsendredirects)); 1686 case IPCTL_DEFTTL: 1687 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1688 #ifdef notyet 1689 case IPCTL_DEFMTU: 1690 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1691 #endif 1692 case IPCTL_FORWSRCRT: 1693 /* Don't allow this to change in a secure environment. */ 1694 if (securelevel > 0) 1695 return (sysctl_rdint(oldp, oldlenp, newp, 1696 ip_forwsrcrt)); 1697 else 1698 return (sysctl_int(oldp, oldlenp, newp, newlen, 1699 &ip_forwsrcrt)); 1700 case IPCTL_DIRECTEDBCAST: 1701 return (sysctl_int(oldp, oldlenp, newp, newlen, 1702 &ip_directedbcast)); 1703 case IPCTL_ALLOWSRCRT: 1704 return (sysctl_int(oldp, oldlenp, newp, newlen, 1705 &ip_allowsrcrt)); 1706 case IPCTL_SUBNETSARELOCAL: 1707 return (sysctl_int(oldp, oldlenp, newp, newlen, 1708 &subnetsarelocal)); 1709 case IPCTL_MTUDISC: 1710 error = sysctl_int(oldp, oldlenp, newp, newlen, 1711 &ip_mtudisc); 1712 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1713 ip_mtudisc_timeout_q = 1714 rt_timer_queue_create(ip_mtudisc_timeout); 1715 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1716 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1717 ip_mtudisc_timeout_q = NULL; 1718 } 1719 return error; 1720 case IPCTL_ANONPORTMIN: 1721 old = anonportmin; 1722 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1723 if (anonportmin >= anonportmax || anonportmin < 0 1724 || anonportmin > 65535 1725 #ifndef IPNOPRIVPORTS 1726 || anonportmin < IPPORT_RESERVED 1727 #endif 1728 ) { 1729 anonportmin = old; 1730 return (EINVAL); 1731 } 1732 return (error); 1733 case IPCTL_ANONPORTMAX: 1734 old = anonportmax; 1735 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1736 if (anonportmin >= anonportmax || anonportmax < 0 1737 || anonportmax > 65535 1738 #ifndef IPNOPRIVPORTS 1739 || anonportmax < IPPORT_RESERVED 1740 #endif 1741 ) { 1742 anonportmax = old; 1743 return (EINVAL); 1744 } 1745 return (error); 1746 case IPCTL_MTUDISCTIMEOUT: 1747 error = sysctl_int(oldp, oldlenp, newp, newlen, 1748 &ip_mtudisc_timeout); 1749 if (ip_mtudisc_timeout_q != NULL) 1750 rt_timer_queue_change(ip_mtudisc_timeout_q, 1751 ip_mtudisc_timeout); 1752 return (error); 1753 #ifdef GATEWAY 1754 case IPCTL_MAXFLOWS: 1755 { 1756 int s; 1757 1758 error = sysctl_int(oldp, oldlenp, newp, newlen, 1759 &ip_maxflows); 1760 s = splsoftnet(); 1761 ipflow_reap(0); 1762 splx(s); 1763 return (error); 1764 } 1765 #endif 1766 case IPCTL_HOSTZEROBROADCAST: 1767 return (sysctl_int(oldp, oldlenp, newp, newlen, 1768 &hostzeroisbroadcast)); 1769 #if NGIF > 0 1770 case IPCTL_GIF_TTL: 1771 return(sysctl_int(oldp, oldlenp, newp, newlen, 1772 &ip_gif_ttl)); 1773 #endif 1774 1775 #if NGRE > 0 1776 case IPCTL_GRE_TTL: 1777 return(sysctl_int(oldp, oldlenp, newp, newlen, 1778 &ip_gre_ttl)); 1779 #endif 1780 1781 #ifndef IPNOPRIVPORTS 1782 case IPCTL_LOWPORTMIN: 1783 old = lowportmin; 1784 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 1785 if (lowportmin >= lowportmax 1786 || lowportmin > IPPORT_RESERVEDMAX 1787 || lowportmin < IPPORT_RESERVEDMIN 1788 ) { 1789 lowportmin = old; 1790 return (EINVAL); 1791 } 1792 return (error); 1793 case IPCTL_LOWPORTMAX: 1794 old = lowportmax; 1795 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 1796 if (lowportmin >= lowportmax 1797 || lowportmax > IPPORT_RESERVEDMAX 1798 || lowportmax < IPPORT_RESERVEDMIN 1799 ) { 1800 lowportmax = old; 1801 return (EINVAL); 1802 } 1803 return (error); 1804 #endif 1805 1806 case IPCTL_MAXFRAGPACKETS: 1807 return (sysctl_int(oldp, oldlenp, newp, newlen, 1808 &ip_maxfragpackets)); 1809 1810 default: 1811 return (EOPNOTSUPP); 1812 } 1813 /* NOTREACHED */ 1814 } 1815
Indexes created Fri Oct 17 17:09:57 GMT 2025