ip_input.c revision 1.82.2.2.4.3
1 /* $NetBSD: ip_input.c,v 1.82.2.2.4.3 1999/11/30 13:35:31 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" 141 142 #ifdef IPSEC 143 #include <netinet6/ipsec.h> 144 #include <netkey/key.h> 145 #include <netkey/key_debug.h> 146 #endif 147 148 #ifndef IPFORWARDING 149 #ifdef GATEWAY 150 #define IPFORWARDING 1 /* forward IP packets not for us */ 151 #else /* GATEWAY */ 152 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 153 #endif /* GATEWAY */ 154 #endif /* IPFORWARDING */ 155 #ifndef IPSENDREDIRECTS 156 #define IPSENDREDIRECTS 1 157 #endif 158 #ifndef IPFORWSRCRT 159 #define IPFORWSRCRT 1 /* forward source-routed packets */ 160 #endif 161 #ifndef IPALLOWSRCRT 162 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 163 #endif 164 #ifndef IPMTUDISC 165 #define IPMTUDISC 0 166 #endif 167 #ifndef IPMTUDISCTIMEOUT 168 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 169 #endif 170 171 /* 172 * Note: DIRECTED_BROADCAST is handled this way so that previous 173 * configuration using this option will Just Work. 174 */ 175 #ifndef IPDIRECTEDBCAST 176 #ifdef DIRECTED_BROADCAST 177 #define IPDIRECTEDBCAST 1 178 #else 179 #define IPDIRECTEDBCAST 0 180 #endif /* DIRECTED_BROADCAST */ 181 #endif /* IPDIRECTEDBCAST */ 182 int ipforwarding = IPFORWARDING; 183 int ipsendredirects = IPSENDREDIRECTS; 184 int ip_defttl = IPDEFTTL; 185 int ip_forwsrcrt = IPFORWSRCRT; 186 int ip_directedbcast = IPDIRECTEDBCAST; 187 int ip_allowsrcrt = IPALLOWSRCRT; 188 int ip_mtudisc = IPMTUDISC; 189 u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 190 #ifdef DIAGNOSTIC 191 int ipprintfs = 0; 192 #endif 193 194 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 195 196 #ifdef ALTQ 197 int (*altq_input) __P((struct mbuf *, int)) = NULL; 198 #endif 199 200 extern struct domain inetdomain; 201 extern struct protosw inetsw[]; 202 u_char ip_protox[IPPROTO_MAX]; 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 int ip_defttl; 210 211 struct ipqhead ipq; 212 int ipq_locked; 213 214 static __inline int ipq_lock_try __P((void)); 215 static __inline void ipq_unlock __P((void)); 216 217 static __inline int 218 ipq_lock_try() 219 { 220 int s; 221 222 s = splimp(); 223 if (ipq_locked) { 224 splx(s); 225 return (0); 226 } 227 ipq_locked = 1; 228 splx(s); 229 return (1); 230 } 231 232 static __inline void 233 ipq_unlock() 234 { 235 int s; 236 237 s = splimp(); 238 ipq_locked = 0; 239 splx(s); 240 } 241 242 #ifdef DIAGNOSTIC 243 #define IPQ_LOCK() \ 244 do { \ 245 if (ipq_lock_try() == 0) { \ 246 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 247 panic("ipq_lock"); \ 248 } \ 249 } while (0) 250 #define IPQ_LOCK_CHECK() \ 251 do { \ 252 if (ipq_locked == 0) { \ 253 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 254 panic("ipq lock check"); \ 255 } \ 256 } while (0) 257 #else 258 #define IPQ_LOCK() (void) ipq_lock_try() 259 #define IPQ_LOCK_CHECK() /* nothing */ 260 #endif 261 262 #define IPQ_UNLOCK() ipq_unlock() 263 264 struct pool ipqent_pool; 265 266 /* 267 * We need to save the IP options in case a protocol wants to respond 268 * to an incoming packet over the same route if the packet got here 269 * using IP source routing. This allows connection establishment and 270 * maintenance when the remote end is on a network that is not known 271 * to us. 272 */ 273 int ip_nhops = 0; 274 static struct ip_srcrt { 275 struct in_addr dst; /* final destination */ 276 char nop; /* one NOP to align */ 277 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 278 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 279 } ip_srcrt; 280 281 static void save_rte __P((u_char *, struct in_addr)); 282 283 /* 284 * IP initialization: fill in IP protocol switch table. 285 * All protocols not implemented in kernel go to raw IP protocol handler. 286 */ 287 void 288 ip_init() 289 { 290 register struct protosw *pr; 291 register int i; 292 293 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 294 0, NULL, NULL, M_IPQ); 295 296 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 297 if (pr == 0) 298 panic("ip_init"); 299 for (i = 0; i < IPPROTO_MAX; i++) 300 ip_protox[i] = pr - inetsw; 301 for (pr = inetdomain.dom_protosw; 302 pr < inetdomain.dom_protoswNPROTOSW; pr++) 303 if (pr->pr_domain->dom_family == PF_INET && 304 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 305 ip_protox[pr->pr_protocol] = pr - inetsw; 306 LIST_INIT(&ipq); 307 ip_id = time.tv_sec & 0xffff; 308 ipintrq.ifq_maxlen = ipqmaxlen; 309 TAILQ_INIT(&in_ifaddr); 310 in_ifaddrhashtbl = 311 hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); 312 if (ip_mtudisc != 0) 313 ip_mtudisc_timeout_q = 314 rt_timer_queue_create(ip_mtudisc_timeout); 315 #ifdef GATEWAY 316 ipflow_init(); 317 #endif 318 } 319 320 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 321 struct route ipforward_rt; 322 323 /* 324 * IP software interrupt routine 325 */ 326 void 327 ipintr() 328 { 329 int s; 330 struct mbuf *m; 331 332 while (1) { 333 s = splimp(); 334 IF_DEQUEUE(&ipintrq, m); 335 splx(s); 336 if (m == 0) 337 return; 338 ip_input(m); 339 } 340 } 341 342 /* 343 * Ip input routine. Checksum and byte swap header. If fragmented 344 * try to reassemble. Process options. Pass to next level. 345 */ 346 void 347 ip_input(struct mbuf *m) 348 { 349 register struct ip *ip = NULL; 350 register struct ipq *fp; 351 register struct in_ifaddr *ia; 352 register struct ifaddr *ifa; 353 struct ipqent *ipqe; 354 int hlen = 0, mff, len; 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 if (in_cksum(m, hlen) != 0) { 405 ipstat.ips_badsum++; 406 goto bad; 407 } 408 409 #ifdef ALTQ 410 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) 411 /* packet is dropped by traffic conditioner */ 412 return; 413 #endif 414 /* 415 * Convert fields to host representation. 416 */ 417 NTOHS(ip->ip_len); 418 NTOHS(ip->ip_off); 419 len = ip->ip_len; 420 421 /* 422 * Check for additional length bogosity 423 */ 424 if (len < hlen) 425 { 426 ipstat.ips_badlen++; 427 goto bad; 428 } 429 430 /* 431 * Check that the amount of data in the buffers 432 * is as at least much as the IP header would have us expect. 433 * Trim mbufs if longer than we expect. 434 * Drop packet if shorter than we expect. 435 */ 436 if (m->m_pkthdr.len < len) { 437 ipstat.ips_tooshort++; 438 goto bad; 439 } 440 if (m->m_pkthdr.len > len) { 441 if (m->m_len == m->m_pkthdr.len) { 442 m->m_len = len; 443 m->m_pkthdr.len = len; 444 } else 445 m_adj(m, len - m->m_pkthdr.len); 446 } 447 448 #ifdef IPSEC 449 /* ipflow (IP fast fowarding) is not compatible with IPsec. */ 450 m->m_flags &= ~M_CANFASTFWD; 451 #else 452 /* 453 * Assume that we can create a fast-forward IP flow entry 454 * based on this packet. 455 */ 456 m->m_flags |= M_CANFASTFWD; 457 #endif 458 459 #ifdef PFIL_HOOKS 460 /* 461 * Run through list of hooks for input packets. If there are any 462 * filters which require that additional packets in the flow are 463 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 464 * Note that filters must _never_ set this flag, as another filter 465 * in the list may have previously cleared it. 466 */ 467 m0 = m; 468 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next) 469 if (pfh->pfil_func) { 470 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 471 if (rv) 472 return; 473 m = m0; 474 if (m == NULL) 475 return; 476 ip = mtod(m, struct ip *); 477 } 478 #endif /* PFIL_HOOKS */ 479 480 /* 481 * Process options and, if not destined for us, 482 * ship it on. ip_dooptions returns 1 when an 483 * error was detected (causing an icmp message 484 * to be sent and the original packet to be freed). 485 */ 486 ip_nhops = 0; /* for source routed packets */ 487 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 488 return; 489 490 /* 491 * Check our list of addresses, to see if the packet is for us. 492 */ 493 INADDR_TO_IA(ip->ip_dst, ia); 494 if (ia != NULL) 495 goto ours; 496 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 497 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 498 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 499 if (ifa->ifa_addr->sa_family != AF_INET) continue; 500 ia = ifatoia(ifa); 501 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 502 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 503 /* 504 * Look for all-0's host part (old broadcast addr), 505 * either for subnet or net. 506 */ 507 ip->ip_dst.s_addr == ia->ia_subnet || 508 ip->ip_dst.s_addr == ia->ia_net) 509 goto ours; 510 /* 511 * An interface with IP address zero accepts 512 * all packets that arrive on that interface. 513 */ 514 if (in_nullhost(ia->ia_addr.sin_addr)) 515 goto ours; 516 } 517 } 518 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 519 struct in_multi *inm; 520 #ifdef MROUTING 521 extern struct socket *ip_mrouter; 522 523 if (m->m_flags & M_EXT) { 524 if ((m = m_pullup(m, hlen)) == 0) { 525 ipstat.ips_toosmall++; 526 return; 527 } 528 ip = mtod(m, struct ip *); 529 } 530 531 if (ip_mrouter) { 532 /* 533 * If we are acting as a multicast router, all 534 * incoming multicast packets are passed to the 535 * kernel-level multicast forwarding function. 536 * The packet is returned (relatively) intact; if 537 * ip_mforward() returns a non-zero value, the packet 538 * must be discarded, else it may be accepted below. 539 * 540 * (The IP ident field is put in the same byte order 541 * as expected when ip_mforward() is called from 542 * ip_output().) 543 */ 544 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 545 ipstat.ips_cantforward++; 546 m_freem(m); 547 return; 548 } 549 550 /* 551 * The process-level routing demon needs to receive 552 * all multicast IGMP packets, whether or not this 553 * host belongs to their destination groups. 554 */ 555 if (ip->ip_p == IPPROTO_IGMP) 556 goto ours; 557 ipstat.ips_forward++; 558 } 559 #endif 560 /* 561 * See if we belong to the destination multicast group on the 562 * arrival interface. 563 */ 564 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 565 if (inm == NULL) { 566 ipstat.ips_cantforward++; 567 m_freem(m); 568 return; 569 } 570 goto ours; 571 } 572 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 573 in_nullhost(ip->ip_dst)) 574 goto ours; 575 576 /* 577 * Not for us; forward if possible and desirable. 578 */ 579 if (ipforwarding == 0) { 580 ipstat.ips_cantforward++; 581 m_freem(m); 582 } else 583 ip_forward(m, 0); 584 return; 585 586 ours: 587 /* 588 * If offset or IP_MF are set, must reassemble. 589 * Otherwise, nothing need be done. 590 * (We could look in the reassembly queue to see 591 * if the packet was previously fragmented, 592 * but it's not worth the time; just let them time out.) 593 */ 594 if (ip->ip_off & ~(IP_DF|IP_RF)) { 595 /* 596 * Look for queue of fragments 597 * of this datagram. 598 */ 599 IPQ_LOCK(); 600 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 601 if (ip->ip_id == fp->ipq_id && 602 in_hosteq(ip->ip_src, fp->ipq_src) && 603 in_hosteq(ip->ip_dst, fp->ipq_dst) && 604 ip->ip_p == fp->ipq_p) 605 goto found; 606 fp = 0; 607 found: 608 609 /* 610 * Adjust ip_len to not reflect header, 611 * set ipqe_mff if more fragments are expected, 612 * convert offset of this to bytes. 613 */ 614 ip->ip_len -= hlen; 615 mff = (ip->ip_off & IP_MF) != 0; 616 if (mff) { 617 /* 618 * Make sure that fragments have a data length 619 * that's a non-zero multiple of 8 bytes. 620 */ 621 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 622 ipstat.ips_badfrags++; 623 IPQ_UNLOCK(); 624 goto bad; 625 } 626 } 627 ip->ip_off <<= 3; 628 629 /* 630 * If datagram marked as having more fragments 631 * or if this is not the first fragment, 632 * attempt reassembly; if it succeeds, proceed. 633 */ 634 if (mff || ip->ip_off) { 635 ipstat.ips_fragments++; 636 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 637 if (ipqe == NULL) { 638 ipstat.ips_rcvmemdrop++; 639 IPQ_UNLOCK(); 640 goto bad; 641 } 642 ipqe->ipqe_mff = mff; 643 ipqe->ipqe_m = m; 644 ipqe->ipqe_ip = ip; 645 m = ip_reass(ipqe, fp); 646 if (m == 0) { 647 IPQ_UNLOCK(); 648 return; 649 } 650 ipstat.ips_reassembled++; 651 ip = mtod(m, struct ip *); 652 hlen = ip->ip_hl << 2; 653 ip->ip_len += hlen; 654 } else 655 if (fp) 656 ip_freef(fp); 657 IPQ_UNLOCK(); 658 } 659 660 /* 661 * Switch out to protocol's input routine. 662 */ 663 #if IFA_STATS 664 ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; 665 #endif 666 ipstat.ips_delivered++; 667 { 668 int off = hlen, nh = ip->ip_p; 669 670 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 671 return; 672 } 673 bad: 674 m_freem(m); 675 } 676 677 /* 678 * Take incoming datagram fragment and try to 679 * reassemble it into whole datagram. If a chain for 680 * reassembly of this datagram already exists, then it 681 * is given as fp; otherwise have to make a chain. 682 */ 683 struct mbuf * 684 ip_reass(ipqe, fp) 685 register struct ipqent *ipqe; 686 register struct ipq *fp; 687 { 688 register struct mbuf *m = ipqe->ipqe_m; 689 register struct ipqent *nq, *p, *q; 690 struct ip *ip; 691 struct mbuf *t; 692 int hlen = ipqe->ipqe_ip->ip_hl << 2; 693 int i, next; 694 695 IPQ_LOCK_CHECK(); 696 697 /* 698 * Presence of header sizes in mbufs 699 * would confuse code below. 700 */ 701 m->m_data += hlen; 702 m->m_len -= hlen; 703 704 /* 705 * If first fragment to arrive, create a reassembly queue. 706 */ 707 if (fp == 0) { 708 MALLOC(fp, struct ipq *, sizeof (struct ipq), 709 M_FTABLE, M_NOWAIT); 710 if (fp == NULL) 711 goto dropfrag; 712 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 713 fp->ipq_ttl = IPFRAGTTL; 714 fp->ipq_p = ipqe->ipqe_ip->ip_p; 715 fp->ipq_id = ipqe->ipqe_ip->ip_id; 716 LIST_INIT(&fp->ipq_fragq); 717 fp->ipq_src = ipqe->ipqe_ip->ip_src; 718 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 719 p = NULL; 720 goto insert; 721 } 722 723 /* 724 * Find a segment which begins after this one does. 725 */ 726 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 727 p = q, q = q->ipqe_q.le_next) 728 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 729 break; 730 731 /* 732 * If there is a preceding segment, it may provide some of 733 * our data already. If so, drop the data from the incoming 734 * segment. If it provides all of our data, drop us. 735 */ 736 if (p != NULL) { 737 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 738 ipqe->ipqe_ip->ip_off; 739 if (i > 0) { 740 if (i >= ipqe->ipqe_ip->ip_len) 741 goto dropfrag; 742 m_adj(ipqe->ipqe_m, i); 743 ipqe->ipqe_ip->ip_off += i; 744 ipqe->ipqe_ip->ip_len -= i; 745 } 746 } 747 748 /* 749 * While we overlap succeeding segments trim them or, 750 * if they are completely covered, dequeue them. 751 */ 752 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 753 q->ipqe_ip->ip_off; q = nq) { 754 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 755 q->ipqe_ip->ip_off; 756 if (i < q->ipqe_ip->ip_len) { 757 q->ipqe_ip->ip_len -= i; 758 q->ipqe_ip->ip_off += i; 759 m_adj(q->ipqe_m, i); 760 break; 761 } 762 nq = q->ipqe_q.le_next; 763 m_freem(q->ipqe_m); 764 LIST_REMOVE(q, ipqe_q); 765 pool_put(&ipqent_pool, q); 766 } 767 768 insert: 769 /* 770 * Stick new segment in its place; 771 * check for complete reassembly. 772 */ 773 if (p == NULL) { 774 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 775 } else { 776 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 777 } 778 next = 0; 779 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 780 p = q, q = q->ipqe_q.le_next) { 781 if (q->ipqe_ip->ip_off != next) 782 return (0); 783 next += q->ipqe_ip->ip_len; 784 } 785 if (p->ipqe_mff) 786 return (0); 787 788 /* 789 * Reassembly is complete. Check for a bogus message size and 790 * concatenate fragments. 791 */ 792 q = fp->ipq_fragq.lh_first; 793 ip = q->ipqe_ip; 794 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 795 ipstat.ips_toolong++; 796 ip_freef(fp); 797 return (0); 798 } 799 m = q->ipqe_m; 800 t = m->m_next; 801 m->m_next = 0; 802 m_cat(m, t); 803 nq = q->ipqe_q.le_next; 804 pool_put(&ipqent_pool, q); 805 for (q = nq; q != NULL; q = nq) { 806 t = q->ipqe_m; 807 nq = q->ipqe_q.le_next; 808 pool_put(&ipqent_pool, q); 809 m_cat(m, t); 810 } 811 812 /* 813 * Create header for new ip packet by 814 * modifying header of first packet; 815 * dequeue and discard fragment reassembly header. 816 * Make header visible. 817 */ 818 ip->ip_len = next; 819 ip->ip_ttl = 0; /* xxx */ 820 ip->ip_sum = 0; 821 ip->ip_src = fp->ipq_src; 822 ip->ip_dst = fp->ipq_dst; 823 LIST_REMOVE(fp, ipq_q); 824 FREE(fp, M_FTABLE); 825 m->m_len += (ip->ip_hl << 2); 826 m->m_data -= (ip->ip_hl << 2); 827 /* some debugging cruft by sklower, below, will go away soon */ 828 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 829 register int plen = 0; 830 for (t = m; t; t = t->m_next) 831 plen += t->m_len; 832 m->m_pkthdr.len = plen; 833 } 834 return (m); 835 836 dropfrag: 837 ipstat.ips_fragdropped++; 838 m_freem(m); 839 pool_put(&ipqent_pool, ipqe); 840 return (0); 841 } 842 843 /* 844 * Free a fragment reassembly header and all 845 * associated datagrams. 846 */ 847 void 848 ip_freef(fp) 849 struct ipq *fp; 850 { 851 register struct ipqent *q, *p; 852 853 IPQ_LOCK_CHECK(); 854 855 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 856 p = q->ipqe_q.le_next; 857 m_freem(q->ipqe_m); 858 LIST_REMOVE(q, ipqe_q); 859 pool_put(&ipqent_pool, q); 860 } 861 LIST_REMOVE(fp, ipq_q); 862 FREE(fp, M_FTABLE); 863 } 864 865 /* 866 * IP timer processing; 867 * if a timer expires on a reassembly 868 * queue, discard it. 869 */ 870 void 871 ip_slowtimo() 872 { 873 register struct ipq *fp, *nfp; 874 int s = splsoftnet(); 875 876 IPQ_LOCK(); 877 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 878 nfp = fp->ipq_q.le_next; 879 if (--fp->ipq_ttl == 0) { 880 ipstat.ips_fragtimeout++; 881 ip_freef(fp); 882 } 883 } 884 IPQ_UNLOCK(); 885 #ifdef GATEWAY 886 ipflow_slowtimo(); 887 #endif 888 splx(s); 889 } 890 891 /* 892 * Drain off all datagram fragments. 893 */ 894 void 895 ip_drain() 896 { 897 898 /* 899 * We may be called from a device's interrupt context. If 900 * the ipq is already busy, just bail out now. 901 */ 902 if (ipq_lock_try() == 0) 903 return; 904 905 while (ipq.lh_first != NULL) { 906 ipstat.ips_fragdropped++; 907 ip_freef(ipq.lh_first); 908 } 909 910 IPQ_UNLOCK(); 911 } 912 913 /* 914 * Do option processing on a datagram, 915 * possibly discarding it if bad options are encountered, 916 * or forwarding it if source-routed. 917 * Returns 1 if packet has been forwarded/freed, 918 * 0 if the packet should be processed further. 919 */ 920 int 921 ip_dooptions(m) 922 struct mbuf *m; 923 { 924 register struct ip *ip = mtod(m, struct ip *); 925 register u_char *cp; 926 register struct ip_timestamp *ipt; 927 register struct in_ifaddr *ia; 928 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 929 struct in_addr *sin, dst; 930 n_time ntime; 931 932 dst = ip->ip_dst; 933 cp = (u_char *)(ip + 1); 934 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 935 for (; cnt > 0; cnt -= optlen, cp += optlen) { 936 opt = cp[IPOPT_OPTVAL]; 937 if (opt == IPOPT_EOL) 938 break; 939 if (opt == IPOPT_NOP) 940 optlen = 1; 941 else { 942 optlen = cp[IPOPT_OLEN]; 943 if (optlen <= 0 || optlen > cnt) { 944 code = &cp[IPOPT_OLEN] - (u_char *)ip; 945 goto bad; 946 } 947 } 948 switch (opt) { 949 950 default: 951 break; 952 953 /* 954 * Source routing with record. 955 * Find interface with current destination address. 956 * If none on this machine then drop if strictly routed, 957 * or do nothing if loosely routed. 958 * Record interface address and bring up next address 959 * component. If strictly routed make sure next 960 * address is on directly accessible net. 961 */ 962 case IPOPT_LSRR: 963 case IPOPT_SSRR: 964 if (ip_allowsrcrt == 0) { 965 type = ICMP_UNREACH; 966 code = ICMP_UNREACH_NET_PROHIB; 967 goto bad; 968 } 969 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 970 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 971 goto bad; 972 } 973 ipaddr.sin_addr = ip->ip_dst; 974 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 975 if (ia == 0) { 976 if (opt == IPOPT_SSRR) { 977 type = ICMP_UNREACH; 978 code = ICMP_UNREACH_SRCFAIL; 979 goto bad; 980 } 981 /* 982 * Loose routing, and not at next destination 983 * yet; nothing to do except forward. 984 */ 985 break; 986 } 987 off--; /* 0 origin */ 988 if (off > optlen - sizeof(struct in_addr)) { 989 /* 990 * End of source route. Should be for us. 991 */ 992 save_rte(cp, ip->ip_src); 993 break; 994 } 995 /* 996 * locate outgoing interface 997 */ 998 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 999 sizeof(ipaddr.sin_addr)); 1000 if (opt == IPOPT_SSRR) { 1001 #define INA struct in_ifaddr * 1002 #define SA struct sockaddr * 1003 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 1004 } else 1005 ia = ip_rtaddr(ipaddr.sin_addr); 1006 if (ia == 0) { 1007 type = ICMP_UNREACH; 1008 code = ICMP_UNREACH_SRCFAIL; 1009 goto bad; 1010 } 1011 ip->ip_dst = ipaddr.sin_addr; 1012 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1013 (caddr_t)(cp + off), sizeof(struct in_addr)); 1014 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1015 /* 1016 * Let ip_intr's mcast routing check handle mcast pkts 1017 */ 1018 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1019 break; 1020 1021 case IPOPT_RR: 1022 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1023 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1024 goto bad; 1025 } 1026 /* 1027 * If no space remains, ignore. 1028 */ 1029 off--; /* 0 origin */ 1030 if (off > optlen - sizeof(struct in_addr)) 1031 break; 1032 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1033 sizeof(ipaddr.sin_addr)); 1034 /* 1035 * locate outgoing interface; if we're the destination, 1036 * use the incoming interface (should be same). 1037 */ 1038 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 1039 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 1040 type = ICMP_UNREACH; 1041 code = ICMP_UNREACH_HOST; 1042 goto bad; 1043 } 1044 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1045 (caddr_t)(cp + off), sizeof(struct in_addr)); 1046 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1047 break; 1048 1049 case IPOPT_TS: 1050 code = cp - (u_char *)ip; 1051 ipt = (struct ip_timestamp *)cp; 1052 if (ipt->ipt_len < 5) 1053 goto bad; 1054 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1055 if (++ipt->ipt_oflw == 0) 1056 goto bad; 1057 break; 1058 } 1059 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 1060 switch (ipt->ipt_flg) { 1061 1062 case IPOPT_TS_TSONLY: 1063 break; 1064 1065 case IPOPT_TS_TSANDADDR: 1066 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1067 sizeof(struct in_addr) > ipt->ipt_len) 1068 goto bad; 1069 ipaddr.sin_addr = dst; 1070 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 1071 m->m_pkthdr.rcvif); 1072 if (ia == 0) 1073 continue; 1074 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1075 (caddr_t)sin, sizeof(struct in_addr)); 1076 ipt->ipt_ptr += sizeof(struct in_addr); 1077 break; 1078 1079 case IPOPT_TS_PRESPEC: 1080 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1081 sizeof(struct in_addr) > ipt->ipt_len) 1082 goto bad; 1083 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 1084 sizeof(struct in_addr)); 1085 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 1086 continue; 1087 ipt->ipt_ptr += sizeof(struct in_addr); 1088 break; 1089 1090 default: 1091 goto bad; 1092 } 1093 ntime = iptime(); 1094 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 1095 sizeof(n_time)); 1096 ipt->ipt_ptr += sizeof(n_time); 1097 } 1098 } 1099 if (forward) { 1100 if (ip_forwsrcrt == 0) { 1101 type = ICMP_UNREACH; 1102 code = ICMP_UNREACH_SRCFAIL; 1103 goto bad; 1104 } 1105 ip_forward(m, 1); 1106 return (1); 1107 } 1108 return (0); 1109 bad: 1110 icmp_error(m, type, code, 0, 0); 1111 ipstat.ips_badoptions++; 1112 return (1); 1113 } 1114 1115 /* 1116 * Given address of next destination (final or next hop), 1117 * return internet address info of interface to be used to get there. 1118 */ 1119 struct in_ifaddr * 1120 ip_rtaddr(dst) 1121 struct in_addr dst; 1122 { 1123 register struct sockaddr_in *sin; 1124 1125 sin = satosin(&ipforward_rt.ro_dst); 1126 1127 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1128 if (ipforward_rt.ro_rt) { 1129 RTFREE(ipforward_rt.ro_rt); 1130 ipforward_rt.ro_rt = 0; 1131 } 1132 sin->sin_family = AF_INET; 1133 sin->sin_len = sizeof(*sin); 1134 sin->sin_addr = dst; 1135 1136 rtalloc(&ipforward_rt); 1137 } 1138 if (ipforward_rt.ro_rt == 0) 1139 return ((struct in_ifaddr *)0); 1140 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1141 } 1142 1143 /* 1144 * Save incoming source route for use in replies, 1145 * to be picked up later by ip_srcroute if the receiver is interested. 1146 */ 1147 void 1148 save_rte(option, dst) 1149 u_char *option; 1150 struct in_addr dst; 1151 { 1152 unsigned olen; 1153 1154 olen = option[IPOPT_OLEN]; 1155 #ifdef DIAGNOSTIC 1156 if (ipprintfs) 1157 printf("save_rte: olen %d\n", olen); 1158 #endif /* 0 */ 1159 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1160 return; 1161 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1162 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1163 ip_srcrt.dst = dst; 1164 } 1165 1166 /* 1167 * Retrieve incoming source route for use in replies, 1168 * in the same form used by setsockopt. 1169 * The first hop is placed before the options, will be removed later. 1170 */ 1171 struct mbuf * 1172 ip_srcroute() 1173 { 1174 register struct in_addr *p, *q; 1175 register struct mbuf *m; 1176 1177 if (ip_nhops == 0) 1178 return ((struct mbuf *)0); 1179 m = m_get(M_DONTWAIT, MT_SOOPTS); 1180 if (m == 0) 1181 return ((struct mbuf *)0); 1182 1183 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1184 1185 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1186 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1187 OPTSIZ; 1188 #ifdef DIAGNOSTIC 1189 if (ipprintfs) 1190 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1191 #endif 1192 1193 /* 1194 * First save first hop for return route 1195 */ 1196 p = &ip_srcrt.route[ip_nhops - 1]; 1197 *(mtod(m, struct in_addr *)) = *p--; 1198 #ifdef DIAGNOSTIC 1199 if (ipprintfs) 1200 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1201 #endif 1202 1203 /* 1204 * Copy option fields and padding (nop) to mbuf. 1205 */ 1206 ip_srcrt.nop = IPOPT_NOP; 1207 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1208 bcopy((caddr_t)&ip_srcrt.nop, 1209 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1210 q = (struct in_addr *)(mtod(m, caddr_t) + 1211 sizeof(struct in_addr) + OPTSIZ); 1212 #undef OPTSIZ 1213 /* 1214 * Record return path as an IP source route, 1215 * reversing the path (pointers are now aligned). 1216 */ 1217 while (p >= ip_srcrt.route) { 1218 #ifdef DIAGNOSTIC 1219 if (ipprintfs) 1220 printf(" %x", ntohl(q->s_addr)); 1221 #endif 1222 *q++ = *p--; 1223 } 1224 /* 1225 * Last hop goes to final destination. 1226 */ 1227 *q = ip_srcrt.dst; 1228 #ifdef DIAGNOSTIC 1229 if (ipprintfs) 1230 printf(" %x\n", ntohl(q->s_addr)); 1231 #endif 1232 return (m); 1233 } 1234 1235 /* 1236 * Strip out IP options, at higher 1237 * level protocol in the kernel. 1238 * Second argument is buffer to which options 1239 * will be moved, and return value is their length. 1240 * XXX should be deleted; last arg currently ignored. 1241 */ 1242 void 1243 ip_stripoptions(m, mopt) 1244 register struct mbuf *m; 1245 struct mbuf *mopt; 1246 { 1247 register int i; 1248 struct ip *ip = mtod(m, struct ip *); 1249 register caddr_t opts; 1250 int olen; 1251 1252 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1253 opts = (caddr_t)(ip + 1); 1254 i = m->m_len - (sizeof (struct ip) + olen); 1255 bcopy(opts + olen, opts, (unsigned)i); 1256 m->m_len -= olen; 1257 if (m->m_flags & M_PKTHDR) 1258 m->m_pkthdr.len -= olen; 1259 ip->ip_len -= olen; 1260 ip->ip_hl = sizeof (struct ip) >> 2; 1261 } 1262 1263 int inetctlerrmap[PRC_NCMDS] = { 1264 0, 0, 0, 0, 1265 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1266 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1267 EMSGSIZE, EHOSTUNREACH, 0, 0, 1268 0, 0, 0, 0, 1269 ENOPROTOOPT 1270 }; 1271 1272 /* 1273 * Forward a packet. If some error occurs return the sender 1274 * an icmp packet. Note we can't always generate a meaningful 1275 * icmp message because icmp doesn't have a large enough repertoire 1276 * of codes and types. 1277 * 1278 * If not forwarding, just drop the packet. This could be confusing 1279 * if ipforwarding was zero but some routing protocol was advancing 1280 * us as a gateway to somewhere. However, we must let the routing 1281 * protocol deal with that. 1282 * 1283 * The srcrt parameter indicates whether the packet is being forwarded 1284 * via a source route. 1285 */ 1286 void 1287 ip_forward(m, srcrt) 1288 struct mbuf *m; 1289 int srcrt; 1290 { 1291 register struct ip *ip = mtod(m, struct ip *); 1292 register struct sockaddr_in *sin; 1293 register struct rtentry *rt; 1294 int error, type = 0, code = 0; 1295 struct mbuf *mcopy; 1296 n_long dest; 1297 struct ifnet *destifp; 1298 #ifdef IPSEC 1299 struct ifnet dummyifp; 1300 #endif 1301 1302 dest = 0; 1303 #ifdef DIAGNOSTIC 1304 if (ipprintfs) 1305 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1306 ntohl(ip->ip_src.s_addr), 1307 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1308 #endif 1309 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1310 ipstat.ips_cantforward++; 1311 m_freem(m); 1312 return; 1313 } 1314 if (ip->ip_ttl <= IPTTLDEC) { 1315 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1316 return; 1317 } 1318 ip->ip_ttl -= IPTTLDEC; 1319 1320 sin = satosin(&ipforward_rt.ro_dst); 1321 if ((rt = ipforward_rt.ro_rt) == 0 || 1322 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1323 if (ipforward_rt.ro_rt) { 1324 RTFREE(ipforward_rt.ro_rt); 1325 ipforward_rt.ro_rt = 0; 1326 } 1327 sin->sin_family = AF_INET; 1328 sin->sin_len = sizeof(struct sockaddr_in); 1329 sin->sin_addr = ip->ip_dst; 1330 1331 rtalloc(&ipforward_rt); 1332 if (ipforward_rt.ro_rt == 0) { 1333 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1334 return; 1335 } 1336 rt = ipforward_rt.ro_rt; 1337 } 1338 1339 /* 1340 * Save at most 68 bytes of the packet in case 1341 * we need to generate an ICMP message to the src. 1342 */ 1343 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1344 1345 /* 1346 * If forwarding packet using same interface that it came in on, 1347 * perhaps should send a redirect to sender to shortcut a hop. 1348 * Only send redirect if source is sending directly to us, 1349 * and if packet was not source routed (or has any options). 1350 * Also, don't send redirect if forwarding using a default route 1351 * or a route modified by a redirect. 1352 */ 1353 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1354 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1355 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1356 ipsendredirects && !srcrt) { 1357 if (rt->rt_ifa && 1358 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1359 ifatoia(rt->rt_ifa)->ia_subnet) { 1360 if (rt->rt_flags & RTF_GATEWAY) 1361 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1362 else 1363 dest = ip->ip_dst.s_addr; 1364 /* 1365 * Router requirements says to only send host 1366 * redirects. 1367 */ 1368 type = ICMP_REDIRECT; 1369 code = ICMP_REDIRECT_HOST; 1370 #ifdef DIAGNOSTIC 1371 if (ipprintfs) 1372 printf("redirect (%d) to %x\n", code, 1373 (u_int32_t)dest); 1374 #endif 1375 } 1376 } 1377 1378 #ifdef IPSEC 1379 m->m_pkthdr.rcvif = NULL; 1380 #endif /*IPSEC*/ 1381 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1382 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1383 if (error) 1384 ipstat.ips_cantforward++; 1385 else { 1386 ipstat.ips_forward++; 1387 if (type) 1388 ipstat.ips_redirectsent++; 1389 else { 1390 if (mcopy) { 1391 #ifdef GATEWAY 1392 if (mcopy->m_flags & M_CANFASTFWD) 1393 ipflow_create(&ipforward_rt, mcopy); 1394 #endif 1395 m_freem(mcopy); 1396 } 1397 return; 1398 } 1399 } 1400 if (mcopy == NULL) 1401 return; 1402 destifp = NULL; 1403 1404 switch (error) { 1405 1406 case 0: /* forwarded, but need redirect */ 1407 /* type, code set above */ 1408 break; 1409 1410 case ENETUNREACH: /* shouldn't happen, checked above */ 1411 case EHOSTUNREACH: 1412 case ENETDOWN: 1413 case EHOSTDOWN: 1414 default: 1415 type = ICMP_UNREACH; 1416 code = ICMP_UNREACH_HOST; 1417 break; 1418 1419 case EMSGSIZE: 1420 type = ICMP_UNREACH; 1421 code = ICMP_UNREACH_NEEDFRAG; 1422 #ifndef IPSEC 1423 if (ipforward_rt.ro_rt) 1424 destifp = ipforward_rt.ro_rt->rt_ifp; 1425 #else 1426 /* 1427 * If the packet is routed over IPsec tunnel, tell the 1428 * originator the tunnel MTU. 1429 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1430 * XXX quickhack!!! 1431 */ 1432 if (ipforward_rt.ro_rt) { 1433 struct secpolicy *sp; 1434 int ipsecerror; 1435 int ipsechdr; 1436 struct route *ro; 1437 1438 sp = ipsec4_getpolicybyaddr(mcopy, 1439 IPSEC_DIR_OUTBOUND, 1440 IP_FORWARDING, 1441 &ipsecerror); 1442 1443 if (sp == NULL) 1444 destifp = ipforward_rt.ro_rt->rt_ifp; 1445 else { 1446 /* count IPsec header size */ 1447 ipsechdr = ipsec4_hdrsiz(mcopy, 1448 IPSEC_DIR_OUTBOUND, 1449 NULL); 1450 1451 /* 1452 * find the correct route for outer IPv4 1453 * header, compute tunnel MTU. 1454 * 1455 * XXX BUG ALERT 1456 * The "dummyifp" code relies upon the fact 1457 * that icmp_error() touches only ifp->if_mtu. 1458 */ 1459 /*XXX*/ 1460 destifp = NULL; 1461 if (sp->req != NULL 1462 && sp->req->sav != NULL 1463 && sp->req->sav->sah != NULL) { 1464 ro = &sp->req->sav->sah->sa_route; 1465 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1466 dummyifp.if_mtu = 1467 ro->ro_rt->rt_ifp->if_mtu; 1468 dummyifp.if_mtu -= ipsechdr; 1469 destifp = &dummyifp; 1470 } 1471 } 1472 1473 key_freesp(sp); 1474 } 1475 } 1476 #endif /*IPSEC*/ 1477 ipstat.ips_cantfrag++; 1478 break; 1479 1480 case ENOBUFS: 1481 type = ICMP_SOURCEQUENCH; 1482 code = 0; 1483 break; 1484 } 1485 icmp_error(mcopy, type, code, dest, destifp); 1486 } 1487 1488 void 1489 ip_savecontrol(inp, mp, ip, m) 1490 register struct inpcb *inp; 1491 register struct mbuf **mp; 1492 register struct ip *ip; 1493 register struct mbuf *m; 1494 { 1495 1496 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1497 struct timeval tv; 1498 1499 microtime(&tv); 1500 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1501 SCM_TIMESTAMP, SOL_SOCKET); 1502 if (*mp) 1503 mp = &(*mp)->m_next; 1504 } 1505 if (inp->inp_flags & INP_RECVDSTADDR) { 1506 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1507 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1508 if (*mp) 1509 mp = &(*mp)->m_next; 1510 } 1511 #ifdef notyet 1512 /* 1513 * XXX 1514 * Moving these out of udp_input() made them even more broken 1515 * than they already were. 1516 * - fenner (at) parc.xerox.com 1517 */ 1518 /* options were tossed already */ 1519 if (inp->inp_flags & INP_RECVOPTS) { 1520 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1521 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1522 if (*mp) 1523 mp = &(*mp)->m_next; 1524 } 1525 /* ip_srcroute doesn't do what we want here, need to fix */ 1526 if (inp->inp_flags & INP_RECVRETOPTS) { 1527 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1528 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1529 if (*mp) 1530 mp = &(*mp)->m_next; 1531 } 1532 #endif 1533 if (inp->inp_flags & INP_RECVIF) { 1534 struct sockaddr_dl sdl; 1535 1536 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1537 sdl.sdl_family = AF_LINK; 1538 sdl.sdl_index = m->m_pkthdr.rcvif ? 1539 m->m_pkthdr.rcvif->if_index : 0; 1540 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1541 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1542 IP_RECVIF, IPPROTO_IP); 1543 if (*mp) 1544 mp = &(*mp)->m_next; 1545 } 1546 } 1547 1548 int 1549 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1550 int *name; 1551 u_int namelen; 1552 void *oldp; 1553 size_t *oldlenp; 1554 void *newp; 1555 size_t newlen; 1556 { 1557 extern int subnetsarelocal; 1558 1559 int error, old; 1560 1561 /* All sysctl names at this level are terminal. */ 1562 if (namelen != 1) 1563 return (ENOTDIR); 1564 1565 switch (name[0]) { 1566 case IPCTL_FORWARDING: 1567 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1568 case IPCTL_SENDREDIRECTS: 1569 return (sysctl_int(oldp, oldlenp, newp, newlen, 1570 &ipsendredirects)); 1571 case IPCTL_DEFTTL: 1572 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1573 #ifdef notyet 1574 case IPCTL_DEFMTU: 1575 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1576 #endif 1577 case IPCTL_FORWSRCRT: 1578 /* Don't allow this to change in a secure environment. */ 1579 if (securelevel > 0) 1580 return (sysctl_rdint(oldp, oldlenp, newp, 1581 ip_forwsrcrt)); 1582 else 1583 return (sysctl_int(oldp, oldlenp, newp, newlen, 1584 &ip_forwsrcrt)); 1585 case IPCTL_DIRECTEDBCAST: 1586 return (sysctl_int(oldp, oldlenp, newp, newlen, 1587 &ip_directedbcast)); 1588 case IPCTL_ALLOWSRCRT: 1589 return (sysctl_int(oldp, oldlenp, newp, newlen, 1590 &ip_allowsrcrt)); 1591 case IPCTL_SUBNETSARELOCAL: 1592 return (sysctl_int(oldp, oldlenp, newp, newlen, 1593 &subnetsarelocal)); 1594 case IPCTL_MTUDISC: 1595 error = sysctl_int(oldp, oldlenp, newp, newlen, 1596 &ip_mtudisc); 1597 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1598 ip_mtudisc_timeout_q = 1599 rt_timer_queue_create(ip_mtudisc_timeout); 1600 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1601 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1602 ip_mtudisc_timeout_q = NULL; 1603 } 1604 return error; 1605 case IPCTL_ANONPORTMIN: 1606 old = anonportmin; 1607 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1608 if (anonportmin >= anonportmax || anonportmin > 65535 1609 #ifndef IPNOPRIVPORTS 1610 || anonportmin < IPPORT_RESERVED 1611 #endif 1612 ) { 1613 anonportmin = old; 1614 return (EINVAL); 1615 } 1616 return (error); 1617 case IPCTL_ANONPORTMAX: 1618 old = anonportmax; 1619 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1620 if (anonportmin >= anonportmax || anonportmax > 65535 1621 #ifndef IPNOPRIVPORTS 1622 || anonportmax < IPPORT_RESERVED 1623 #endif 1624 ) { 1625 anonportmax = old; 1626 return (EINVAL); 1627 } 1628 return (error); 1629 case IPCTL_MTUDISCTIMEOUT: 1630 error = sysctl_int(oldp, oldlenp, newp, newlen, 1631 &ip_mtudisc_timeout); 1632 if (ip_mtudisc_timeout_q != NULL) 1633 rt_timer_queue_change(ip_mtudisc_timeout_q, 1634 ip_mtudisc_timeout); 1635 return (error); 1636 #ifdef GATEWAY 1637 case IPCTL_MAXFLOWS: 1638 { 1639 int s; 1640 1641 error = sysctl_int(oldp, oldlenp, newp, newlen, 1642 &ip_maxflows); 1643 s = splsoftnet(); 1644 ipflow_reap(0); 1645 splx(s); 1646 return (error); 1647 } 1648 #endif 1649 #if NGIF > 0 1650 case IPCTL_GIF_TTL: 1651 return(sysctl_int(oldp, oldlenp, newp, newlen, 1652 &ip_gif_ttl)); 1653 #endif 1654 default: 1655 return (EOPNOTSUPP); 1656 } 1657 /* NOTREACHED */ 1658 } 1659
Indexes created Wed Oct 15 16:09:53 GMT 2025