ip_input.c revision 1.81
1/* $NetBSD: ip_input.c,v 1.81 1999/03/26 08:51:36 proff Exp $ */ 2 3/*- 4 * Copyright (c) 1998 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Public Access Networks Corporation ("Panix"). It was developed under 9 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the NetBSD 22 * Foundation, Inc. and its contributors. 23 * 4. Neither the name of The NetBSD Foundation nor the names of its 24 * contributors may be used to endorse or promote products derived 25 * from this software without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40/* 41 * Copyright (c) 1982, 1986, 1988, 1993 42 * The Regents of the University of California. All rights reserved. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 3. All advertising materials mentioning features or use of this software 53 * must display the following acknowledgement: 54 * This product includes software developed by the University of 55 * California, Berkeley and its contributors. 56 * 4. Neither the name of the University nor the names of its contributors 57 * may be used to endorse or promote products derived from this software 58 * without specific prior written permission. 59 * 60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 63 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 70 * SUCH DAMAGE. 71 * 72 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 73 */ 74 75#include "opt_gateway.h" 76#include "opt_pfil_hooks.h" 77#include "opt_mrouting.h" 78 79#include <sys/param.h> 80#include <sys/systm.h> 81#include <sys/malloc.h> 82#include <sys/mbuf.h> 83#include <sys/domain.h> 84#include <sys/protosw.h> 85#include <sys/socket.h> 86#include <sys/socketvar.h> 87#include <sys/errno.h> 88#include <sys/time.h> 89#include <sys/kernel.h> 90#include <sys/proc.h> 91#include <sys/pool.h> 92 93#include <vm/vm.h> 94#include <sys/sysctl.h> 95 96#include <net/if.h> 97#include <net/if_dl.h> 98#include <net/route.h> 99#include <net/pfil.h> 100 101#include <netinet/in.h> 102#include <netinet/in_systm.h> 103#include <netinet/ip.h> 104#include <netinet/in_pcb.h> 105#include <netinet/in_var.h> 106#include <netinet/ip_var.h> 107#include <netinet/ip_icmp.h> 108 109#ifndef IPFORWARDING 110#ifdef GATEWAY 111#define IPFORWARDING 1 /* forward IP packets not for us */ 112#else /* GATEWAY */ 113#define IPFORWARDING 0 /* don't forward IP packets not for us */ 114#endif /* GATEWAY */ 115#endif /* IPFORWARDING */ 116#ifndef IPSENDREDIRECTS 117#define IPSENDREDIRECTS 1 118#endif 119#ifndef IPFORWSRCRT 120#define IPFORWSRCRT 1 /* forward source-routed packets */ 121#endif 122#ifndef IPALLOWSRCRT 123#define IPALLOWSRCRT 1 /* allow source-routed packets */ 124#endif 125#ifndef IPMTUDISC 126#define IPMTUDISC 0 127#endif 128#ifndef IPMTUDISCTIMEOUT 129#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 130#endif 131 132/* 133 * Note: DIRECTED_BROADCAST is handled this way so that previous 134 * configuration using this option will Just Work. 135 */ 136#ifndef IPDIRECTEDBCAST 137#ifdef DIRECTED_BROADCAST 138#define IPDIRECTEDBCAST 1 139#else 140#define IPDIRECTEDBCAST 0 141#endif /* DIRECTED_BROADCAST */ 142#endif /* IPDIRECTEDBCAST */ 143int ipforwarding = IPFORWARDING; 144int ipsendredirects = IPSENDREDIRECTS; 145int ip_defttl = IPDEFTTL; 146int ip_forwsrcrt = IPFORWSRCRT; 147int ip_directedbcast = IPDIRECTEDBCAST; 148int ip_allowsrcrt = IPALLOWSRCRT; 149int ip_mtudisc = IPMTUDISC; 150u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 151#ifdef DIAGNOSTIC 152int ipprintfs = 0; 153#endif 154 155struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 156 157extern struct domain inetdomain; 158extern struct protosw inetsw[]; 159u_char ip_protox[IPPROTO_MAX]; 160int ipqmaxlen = IFQ_MAXLEN; 161struct in_ifaddrhead in_ifaddr; 162struct in_ifaddrhashhead *in_ifaddrhashtbl; 163struct ifqueue ipintrq; 164struct ipstat ipstat; 165u_int16_t ip_id; 166int ip_defttl; 167 168struct ipqhead ipq; 169int ipq_locked; 170 171static __inline int ipq_lock_try __P((void)); 172static __inline void ipq_unlock __P((void)); 173 174static __inline int 175ipq_lock_try() 176{ 177 int s; 178 179 s = splimp(); 180 if (ipq_locked) { 181 splx(s); 182 return (0); 183 } 184 ipq_locked = 1; 185 splx(s); 186 return (1); 187} 188 189static __inline void 190ipq_unlock() 191{ 192 int s; 193 194 s = splimp(); 195 ipq_locked = 0; 196 splx(s); 197} 198 199#ifdef DIAGNOSTIC 200#define IPQ_LOCK() \ 201do { \ 202 if (ipq_lock_try() == 0) { \ 203 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 204 panic("ipq_lock"); \ 205 } \ 206} while (0) 207#define IPQ_LOCK_CHECK() \ 208do { \ 209 if (ipq_locked == 0) { \ 210 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 211 panic("ipq lock check"); \ 212 } \ 213} while (0) 214#else 215#define IPQ_LOCK() (void) ipq_lock_try() 216#define IPQ_LOCK_CHECK() /* nothing */ 217#endif 218 219#define IPQ_UNLOCK() ipq_unlock() 220 221struct pool ipqent_pool; 222 223/* 224 * We need to save the IP options in case a protocol wants to respond 225 * to an incoming packet over the same route if the packet got here 226 * using IP source routing. This allows connection establishment and 227 * maintenance when the remote end is on a network that is not known 228 * to us. 229 */ 230int ip_nhops = 0; 231static struct ip_srcrt { 232 struct in_addr dst; /* final destination */ 233 char nop; /* one NOP to align */ 234 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 235 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 236} ip_srcrt; 237 238static void save_rte __P((u_char *, struct in_addr)); 239 240/* 241 * IP initialization: fill in IP protocol switch table. 242 * All protocols not implemented in kernel go to raw IP protocol handler. 243 */ 244void 245ip_init() 246{ 247 register struct protosw *pr; 248 register int i; 249 250 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 251 0, NULL, NULL, M_IPQ); 252 253 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 254 if (pr == 0) 255 panic("ip_init"); 256 for (i = 0; i < IPPROTO_MAX; i++) 257 ip_protox[i] = pr - inetsw; 258 for (pr = inetdomain.dom_protosw; 259 pr < inetdomain.dom_protoswNPROTOSW; pr++) 260 if (pr->pr_domain->dom_family == PF_INET && 261 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 262 ip_protox[pr->pr_protocol] = pr - inetsw; 263 LIST_INIT(&ipq); 264 ip_id = time.tv_sec & 0xffff; 265 ipintrq.ifq_maxlen = ipqmaxlen; 266 TAILQ_INIT(&in_ifaddr); 267 in_ifaddrhashtbl = 268 hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); 269 if (ip_mtudisc != 0) 270 ip_mtudisc_timeout_q = 271 rt_timer_queue_create(ip_mtudisc_timeout); 272#ifdef GATEWAY 273 ipflow_init(); 274#endif 275} 276 277struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 278struct route ipforward_rt; 279 280/* 281 * Ip input routine. Checksum and byte swap header. If fragmented 282 * try to reassemble. Process options. Pass to next level. 283 */ 284void 285ipintr() 286{ 287 register struct ip *ip = NULL; 288 register struct mbuf *m; 289 register struct ipq *fp; 290 register struct in_ifaddr *ia; 291 register struct ifaddr *ifa; 292 struct ipqent *ipqe; 293 int hlen = 0, mff, len, s; 294#ifdef PFIL_HOOKS 295 struct packet_filter_hook *pfh; 296 struct mbuf *m0; 297 int rv; 298#endif /* PFIL_HOOKS */ 299 300next: 301 /* 302 * Get next datagram off input queue and get IP header 303 * in first mbuf. 304 */ 305 s = splimp(); 306 IF_DEQUEUE(&ipintrq, m); 307 splx(s); 308 if (m == 0) 309 return; 310#ifdef DIAGNOSTIC 311 if ((m->m_flags & M_PKTHDR) == 0) 312 panic("ipintr no HDR"); 313#endif 314 /* 315 * If no IP addresses have been set yet but the interfaces 316 * are receiving, can't do anything with incoming packets yet. 317 */ 318 if (in_ifaddr.tqh_first == 0) 319 goto bad; 320 ipstat.ips_total++; 321 if (m->m_len < sizeof (struct ip) && 322 (m = m_pullup(m, sizeof (struct ip))) == 0) { 323 ipstat.ips_toosmall++; 324 goto next; 325 } 326 ip = mtod(m, struct ip *); 327 if (ip->ip_v != IPVERSION) { 328 ipstat.ips_badvers++; 329 goto bad; 330 } 331 hlen = ip->ip_hl << 2; 332 if (hlen < sizeof(struct ip)) { /* minimum header length */ 333 ipstat.ips_badhlen++; 334 goto bad; 335 } 336 if (hlen > m->m_len) { 337 if ((m = m_pullup(m, hlen)) == 0) { 338 ipstat.ips_badhlen++; 339 goto next; 340 } 341 ip = mtod(m, struct ip *); 342 } 343 if (in_cksum(m, hlen) != 0) { 344 ipstat.ips_badsum++; 345 goto bad; 346 } 347 348 /* 349 * Convert fields to host representation. 350 */ 351 NTOHS(ip->ip_len); 352 NTOHS(ip->ip_off); 353 len = ip->ip_len; 354 355 /* 356 * Check for additional length bogosity 357 */ 358 if (len < hlen) 359 { 360 ipstat.ips_badlen++; 361 goto bad; 362 } 363 364 /* 365 * Check that the amount of data in the buffers 366 * is as at least much as the IP header would have us expect. 367 * Trim mbufs if longer than we expect. 368 * Drop packet if shorter than we expect. 369 */ 370 if (m->m_pkthdr.len < len) { 371 ipstat.ips_tooshort++; 372 goto bad; 373 } 374 if (m->m_pkthdr.len > len) { 375 if (m->m_len == m->m_pkthdr.len) { 376 m->m_len = len; 377 m->m_pkthdr.len = len; 378 } else 379 m_adj(m, len - m->m_pkthdr.len); 380 } 381 382 /* 383 * Assume that we can create a fast-forward IP flow entry 384 * based on this packet. 385 */ 386 m->m_flags |= M_CANFASTFWD; 387 388#ifdef PFIL_HOOKS 389 /* 390 * Run through list of hooks for input packets. If there are any 391 * filters which require that additional packets in the flow are 392 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 393 * Note that filters must _never_ set this flag, as another filter 394 * in the list may have previously cleared it. 395 */ 396 m0 = m; 397 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next) 398 if (pfh->pfil_func) { 399 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 400 if (rv) 401 goto next; 402 m = m0; 403 if (m == NULL) 404 goto next; 405 ip = mtod(m, struct ip *); 406 } 407#endif /* PFIL_HOOKS */ 408 409 /* 410 * Process options and, if not destined for us, 411 * ship it on. ip_dooptions returns 1 when an 412 * error was detected (causing an icmp message 413 * to be sent and the original packet to be freed). 414 */ 415 ip_nhops = 0; /* for source routed packets */ 416 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 417 goto next; 418 419 /* 420 * Check our list of addresses, to see if the packet is for us. 421 */ 422 INADDR_TO_IA(ip->ip_dst, ia); 423 if (ia != NULL) goto ours; 424 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 425 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 426 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 427 if (ifa->ifa_addr->sa_family != AF_INET) continue; 428 ia = ifatoia(ifa); 429 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 430 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 431 /* 432 * Look for all-0's host part (old broadcast addr), 433 * either for subnet or net. 434 */ 435 ip->ip_dst.s_addr == ia->ia_subnet || 436 ip->ip_dst.s_addr == ia->ia_net) 437 goto ours; 438 /* 439 * An interface with IP address zero accepts 440 * all packets that arrive on that interface. 441 */ 442 if (in_nullhost(ia->ia_addr.sin_addr)) 443 goto ours; 444 } 445 } 446 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 447 struct in_multi *inm; 448#ifdef MROUTING 449 extern struct socket *ip_mrouter; 450 451 if (m->m_flags & M_EXT) { 452 if ((m = m_pullup(m, hlen)) == 0) { 453 ipstat.ips_toosmall++; 454 goto next; 455 } 456 ip = mtod(m, struct ip *); 457 } 458 459 if (ip_mrouter) { 460 /* 461 * If we are acting as a multicast router, all 462 * incoming multicast packets are passed to the 463 * kernel-level multicast forwarding function. 464 * The packet is returned (relatively) intact; if 465 * ip_mforward() returns a non-zero value, the packet 466 * must be discarded, else it may be accepted below. 467 * 468 * (The IP ident field is put in the same byte order 469 * as expected when ip_mforward() is called from 470 * ip_output().) 471 */ 472 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 473 ipstat.ips_cantforward++; 474 m_freem(m); 475 goto next; 476 } 477 478 /* 479 * The process-level routing demon needs to receive 480 * all multicast IGMP packets, whether or not this 481 * host belongs to their destination groups. 482 */ 483 if (ip->ip_p == IPPROTO_IGMP) 484 goto ours; 485 ipstat.ips_forward++; 486 } 487#endif 488 /* 489 * See if we belong to the destination multicast group on the 490 * arrival interface. 491 */ 492 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 493 if (inm == NULL) { 494 ipstat.ips_cantforward++; 495 m_freem(m); 496 goto next; 497 } 498 goto ours; 499 } 500 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 501 in_nullhost(ip->ip_dst)) 502 goto ours; 503 504 /* 505 * Not for us; forward if possible and desirable. 506 */ 507 if (ipforwarding == 0) { 508 ipstat.ips_cantforward++; 509 m_freem(m); 510 } else 511 ip_forward(m, 0); 512 goto next; 513 514ours: 515 /* 516 * If offset or IP_MF are set, must reassemble. 517 * Otherwise, nothing need be done. 518 * (We could look in the reassembly queue to see 519 * if the packet was previously fragmented, 520 * but it's not worth the time; just let them time out.) 521 */ 522 if (ip->ip_off & ~(IP_DF|IP_RF)) { 523 /* 524 * Look for queue of fragments 525 * of this datagram. 526 */ 527 IPQ_LOCK(); 528 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 529 if (ip->ip_id == fp->ipq_id && 530 in_hosteq(ip->ip_src, fp->ipq_src) && 531 in_hosteq(ip->ip_dst, fp->ipq_dst) && 532 ip->ip_p == fp->ipq_p) 533 goto found; 534 fp = 0; 535found: 536 537 /* 538 * Adjust ip_len to not reflect header, 539 * set ipqe_mff if more fragments are expected, 540 * convert offset of this to bytes. 541 */ 542 ip->ip_len -= hlen; 543 mff = (ip->ip_off & IP_MF) != 0; 544 if (mff) { 545 /* 546 * Make sure that fragments have a data length 547 * that's a non-zero multiple of 8 bytes. 548 */ 549 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 550 ipstat.ips_badfrags++; 551 IPQ_UNLOCK(); 552 goto bad; 553 } 554 } 555 ip->ip_off <<= 3; 556 557 /* 558 * If datagram marked as having more fragments 559 * or if this is not the first fragment, 560 * attempt reassembly; if it succeeds, proceed. 561 */ 562 if (mff || ip->ip_off) { 563 ipstat.ips_fragments++; 564 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 565 if (ipqe == NULL) { 566 ipstat.ips_rcvmemdrop++; 567 IPQ_UNLOCK(); 568 goto bad; 569 } 570 ipqe->ipqe_mff = mff; 571 ipqe->ipqe_m = m; 572 ipqe->ipqe_ip = ip; 573 m = ip_reass(ipqe, fp); 574 if (m == 0) { 575 IPQ_UNLOCK(); 576 goto next; 577 } 578 ipstat.ips_reassembled++; 579 ip = mtod(m, struct ip *); 580 hlen = ip->ip_hl << 2; 581 ip->ip_len += hlen; 582 } else 583 if (fp) 584 ip_freef(fp); 585 IPQ_UNLOCK(); 586 } 587 588 /* 589 * Switch out to protocol's input routine. 590 */ 591 ipstat.ips_delivered++; 592 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 593 goto next; 594bad: 595 m_freem(m); 596 goto next; 597} 598 599/* 600 * Take incoming datagram fragment and try to 601 * reassemble it into whole datagram. If a chain for 602 * reassembly of this datagram already exists, then it 603 * is given as fp; otherwise have to make a chain. 604 */ 605struct mbuf * 606ip_reass(ipqe, fp) 607 register struct ipqent *ipqe; 608 register struct ipq *fp; 609{ 610 register struct mbuf *m = ipqe->ipqe_m; 611 register struct ipqent *nq, *p, *q; 612 struct ip *ip; 613 struct mbuf *t; 614 int hlen = ipqe->ipqe_ip->ip_hl << 2; 615 int i, next; 616 617 IPQ_LOCK_CHECK(); 618 619 /* 620 * Presence of header sizes in mbufs 621 * would confuse code below. 622 */ 623 m->m_data += hlen; 624 m->m_len -= hlen; 625 626 /* 627 * If first fragment to arrive, create a reassembly queue. 628 */ 629 if (fp == 0) { 630 MALLOC(fp, struct ipq *, sizeof (struct ipq), 631 M_FTABLE, M_NOWAIT); 632 if (fp == NULL) 633 goto dropfrag; 634 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 635 fp->ipq_ttl = IPFRAGTTL; 636 fp->ipq_p = ipqe->ipqe_ip->ip_p; 637 fp->ipq_id = ipqe->ipqe_ip->ip_id; 638 LIST_INIT(&fp->ipq_fragq); 639 fp->ipq_src = ipqe->ipqe_ip->ip_src; 640 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 641 p = NULL; 642 goto insert; 643 } 644 645 /* 646 * Find a segment which begins after this one does. 647 */ 648 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 649 p = q, q = q->ipqe_q.le_next) 650 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 651 break; 652 653 /* 654 * If there is a preceding segment, it may provide some of 655 * our data already. If so, drop the data from the incoming 656 * segment. If it provides all of our data, drop us. 657 */ 658 if (p != NULL) { 659 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 660 ipqe->ipqe_ip->ip_off; 661 if (i > 0) { 662 if (i >= ipqe->ipqe_ip->ip_len) 663 goto dropfrag; 664 m_adj(ipqe->ipqe_m, i); 665 ipqe->ipqe_ip->ip_off += i; 666 ipqe->ipqe_ip->ip_len -= i; 667 } 668 } 669 670 /* 671 * While we overlap succeeding segments trim them or, 672 * if they are completely covered, dequeue them. 673 */ 674 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 675 q->ipqe_ip->ip_off; q = nq) { 676 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 677 q->ipqe_ip->ip_off; 678 if (i < q->ipqe_ip->ip_len) { 679 q->ipqe_ip->ip_len -= i; 680 q->ipqe_ip->ip_off += i; 681 m_adj(q->ipqe_m, i); 682 break; 683 } 684 nq = q->ipqe_q.le_next; 685 m_freem(q->ipqe_m); 686 LIST_REMOVE(q, ipqe_q); 687 pool_put(&ipqent_pool, q); 688 } 689 690insert: 691 /* 692 * Stick new segment in its place; 693 * check for complete reassembly. 694 */ 695 if (p == NULL) { 696 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 697 } else { 698 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 699 } 700 next = 0; 701 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 702 p = q, q = q->ipqe_q.le_next) { 703 if (q->ipqe_ip->ip_off != next) 704 return (0); 705 next += q->ipqe_ip->ip_len; 706 } 707 if (p->ipqe_mff) 708 return (0); 709 710 /* 711 * Reassembly is complete. Check for a bogus message size and 712 * concatenate fragments. 713 */ 714 q = fp->ipq_fragq.lh_first; 715 ip = q->ipqe_ip; 716 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 717 ipstat.ips_toolong++; 718 ip_freef(fp); 719 return (0); 720 } 721 m = q->ipqe_m; 722 t = m->m_next; 723 m->m_next = 0; 724 m_cat(m, t); 725 nq = q->ipqe_q.le_next; 726 pool_put(&ipqent_pool, q); 727 for (q = nq; q != NULL; q = nq) { 728 t = q->ipqe_m; 729 nq = q->ipqe_q.le_next; 730 pool_put(&ipqent_pool, q); 731 m_cat(m, t); 732 } 733 734 /* 735 * Create header for new ip packet by 736 * modifying header of first packet; 737 * dequeue and discard fragment reassembly header. 738 * Make header visible. 739 */ 740 ip->ip_len = next; 741 ip->ip_src = fp->ipq_src; 742 ip->ip_dst = fp->ipq_dst; 743 LIST_REMOVE(fp, ipq_q); 744 FREE(fp, M_FTABLE); 745 m->m_len += (ip->ip_hl << 2); 746 m->m_data -= (ip->ip_hl << 2); 747 /* some debugging cruft by sklower, below, will go away soon */ 748 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 749 register int plen = 0; 750 for (t = m; t; t = t->m_next) 751 plen += t->m_len; 752 m->m_pkthdr.len = plen; 753 } 754 return (m); 755 756dropfrag: 757 ipstat.ips_fragdropped++; 758 m_freem(m); 759 pool_put(&ipqent_pool, ipqe); 760 return (0); 761} 762 763/* 764 * Free a fragment reassembly header and all 765 * associated datagrams. 766 */ 767void 768ip_freef(fp) 769 struct ipq *fp; 770{ 771 register struct ipqent *q, *p; 772 773 IPQ_LOCK_CHECK(); 774 775 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 776 p = q->ipqe_q.le_next; 777 m_freem(q->ipqe_m); 778 LIST_REMOVE(q, ipqe_q); 779 pool_put(&ipqent_pool, q); 780 } 781 LIST_REMOVE(fp, ipq_q); 782 FREE(fp, M_FTABLE); 783} 784 785/* 786 * IP timer processing; 787 * if a timer expires on a reassembly 788 * queue, discard it. 789 */ 790void 791ip_slowtimo() 792{ 793 register struct ipq *fp, *nfp; 794 int s = splsoftnet(); 795 796 IPQ_LOCK(); 797 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 798 nfp = fp->ipq_q.le_next; 799 if (--fp->ipq_ttl == 0) { 800 ipstat.ips_fragtimeout++; 801 ip_freef(fp); 802 } 803 } 804 IPQ_UNLOCK(); 805#ifdef GATEWAY 806 ipflow_slowtimo(); 807#endif 808 splx(s); 809} 810 811/* 812 * Drain off all datagram fragments. 813 */ 814void 815ip_drain() 816{ 817 818 /* 819 * We may be called from a device's interrupt context. If 820 * the ipq is already busy, just bail out now. 821 */ 822 if (ipq_lock_try() == 0) 823 return; 824 825 while (ipq.lh_first != NULL) { 826 ipstat.ips_fragdropped++; 827 ip_freef(ipq.lh_first); 828 } 829 830 IPQ_UNLOCK(); 831} 832 833/* 834 * Do option processing on a datagram, 835 * possibly discarding it if bad options are encountered, 836 * or forwarding it if source-routed. 837 * Returns 1 if packet has been forwarded/freed, 838 * 0 if the packet should be processed further. 839 */ 840int 841ip_dooptions(m) 842 struct mbuf *m; 843{ 844 register struct ip *ip = mtod(m, struct ip *); 845 register u_char *cp; 846 register struct ip_timestamp *ipt; 847 register struct in_ifaddr *ia; 848 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 849 struct in_addr *sin, dst; 850 n_time ntime; 851 852 dst = ip->ip_dst; 853 cp = (u_char *)(ip + 1); 854 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 855 for (; cnt > 0; cnt -= optlen, cp += optlen) { 856 opt = cp[IPOPT_OPTVAL]; 857 if (opt == IPOPT_EOL) 858 break; 859 if (opt == IPOPT_NOP) 860 optlen = 1; 861 else { 862 optlen = cp[IPOPT_OLEN]; 863 if (optlen <= 0 || optlen > cnt) { 864 code = &cp[IPOPT_OLEN] - (u_char *)ip; 865 goto bad; 866 } 867 } 868 switch (opt) { 869 870 default: 871 break; 872 873 /* 874 * Source routing with record. 875 * Find interface with current destination address. 876 * If none on this machine then drop if strictly routed, 877 * or do nothing if loosely routed. 878 * Record interface address and bring up next address 879 * component. If strictly routed make sure next 880 * address is on directly accessible net. 881 */ 882 case IPOPT_LSRR: 883 case IPOPT_SSRR: 884 if (ip_allowsrcrt == 0) { 885 type = ICMP_UNREACH; 886 code = ICMP_UNREACH_NET_PROHIB; 887 goto bad; 888 } 889 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 890 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 891 goto bad; 892 } 893 ipaddr.sin_addr = ip->ip_dst; 894 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 895 if (ia == 0) { 896 if (opt == IPOPT_SSRR) { 897 type = ICMP_UNREACH; 898 code = ICMP_UNREACH_SRCFAIL; 899 goto bad; 900 } 901 /* 902 * Loose routing, and not at next destination 903 * yet; nothing to do except forward. 904 */ 905 break; 906 } 907 off--; /* 0 origin */ 908 if (off > optlen - sizeof(struct in_addr)) { 909 /* 910 * End of source route. Should be for us. 911 */ 912 save_rte(cp, ip->ip_src); 913 break; 914 } 915 /* 916 * locate outgoing interface 917 */ 918 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 919 sizeof(ipaddr.sin_addr)); 920 if (opt == IPOPT_SSRR) { 921#define INA struct in_ifaddr * 922#define SA struct sockaddr * 923 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 924 } else 925 ia = ip_rtaddr(ipaddr.sin_addr); 926 if (ia == 0) { 927 type = ICMP_UNREACH; 928 code = ICMP_UNREACH_SRCFAIL; 929 goto bad; 930 } 931 ip->ip_dst = ipaddr.sin_addr; 932 bcopy((caddr_t)&ia->ia_addr.sin_addr, 933 (caddr_t)(cp + off), sizeof(struct in_addr)); 934 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 935 /* 936 * Let ip_intr's mcast routing check handle mcast pkts 937 */ 938 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 939 break; 940 941 case IPOPT_RR: 942 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 943 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 944 goto bad; 945 } 946 /* 947 * If no space remains, ignore. 948 */ 949 off--; /* 0 origin */ 950 if (off > optlen - sizeof(struct in_addr)) 951 break; 952 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 953 sizeof(ipaddr.sin_addr)); 954 /* 955 * locate outgoing interface; if we're the destination, 956 * use the incoming interface (should be same). 957 */ 958 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 959 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 960 type = ICMP_UNREACH; 961 code = ICMP_UNREACH_HOST; 962 goto bad; 963 } 964 bcopy((caddr_t)&ia->ia_addr.sin_addr, 965 (caddr_t)(cp + off), sizeof(struct in_addr)); 966 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 967 break; 968 969 case IPOPT_TS: 970 code = cp - (u_char *)ip; 971 ipt = (struct ip_timestamp *)cp; 972 if (ipt->ipt_len < 5) 973 goto bad; 974 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 975 if (++ipt->ipt_oflw == 0) 976 goto bad; 977 break; 978 } 979 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 980 switch (ipt->ipt_flg) { 981 982 case IPOPT_TS_TSONLY: 983 break; 984 985 case IPOPT_TS_TSANDADDR: 986 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 987 sizeof(struct in_addr) > ipt->ipt_len) 988 goto bad; 989 ipaddr.sin_addr = dst; 990 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 991 m->m_pkthdr.rcvif); 992 if (ia == 0) 993 continue; 994 bcopy((caddr_t)&ia->ia_addr.sin_addr, 995 (caddr_t)sin, sizeof(struct in_addr)); 996 ipt->ipt_ptr += sizeof(struct in_addr); 997 break; 998 999 case IPOPT_TS_PRESPEC: 1000 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1001 sizeof(struct in_addr) > ipt->ipt_len) 1002 goto bad; 1003 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 1004 sizeof(struct in_addr)); 1005 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 1006 continue; 1007 ipt->ipt_ptr += sizeof(struct in_addr); 1008 break; 1009 1010 default: 1011 goto bad; 1012 } 1013 ntime = iptime(); 1014 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 1015 sizeof(n_time)); 1016 ipt->ipt_ptr += sizeof(n_time); 1017 } 1018 } 1019 if (forward) { 1020 if (ip_forwsrcrt == 0) { 1021 type = ICMP_UNREACH; 1022 code = ICMP_UNREACH_SRCFAIL; 1023 goto bad; 1024 } 1025 ip_forward(m, 1); 1026 return (1); 1027 } 1028 return (0); 1029bad: 1030 icmp_error(m, type, code, 0, 0); 1031 ipstat.ips_badoptions++; 1032 return (1); 1033} 1034 1035/* 1036 * Given address of next destination (final or next hop), 1037 * return internet address info of interface to be used to get there. 1038 */ 1039struct in_ifaddr * 1040ip_rtaddr(dst) 1041 struct in_addr dst; 1042{ 1043 register struct sockaddr_in *sin; 1044 1045 sin = satosin(&ipforward_rt.ro_dst); 1046 1047 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1048 if (ipforward_rt.ro_rt) { 1049 RTFREE(ipforward_rt.ro_rt); 1050 ipforward_rt.ro_rt = 0; 1051 } 1052 sin->sin_family = AF_INET; 1053 sin->sin_len = sizeof(*sin); 1054 sin->sin_addr = dst; 1055 1056 rtalloc(&ipforward_rt); 1057 } 1058 if (ipforward_rt.ro_rt == 0) 1059 return ((struct in_ifaddr *)0); 1060 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1061} 1062 1063/* 1064 * Save incoming source route for use in replies, 1065 * to be picked up later by ip_srcroute if the receiver is interested. 1066 */ 1067void 1068save_rte(option, dst) 1069 u_char *option; 1070 struct in_addr dst; 1071{ 1072 unsigned olen; 1073 1074 olen = option[IPOPT_OLEN]; 1075#ifdef DIAGNOSTIC 1076 if (ipprintfs) 1077 printf("save_rte: olen %d\n", olen); 1078#endif 1079 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1080 return; 1081 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1082 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1083 ip_srcrt.dst = dst; 1084} 1085 1086/* 1087 * Retrieve incoming source route for use in replies, 1088 * in the same form used by setsockopt. 1089 * The first hop is placed before the options, will be removed later. 1090 */ 1091struct mbuf * 1092ip_srcroute() 1093{ 1094 register struct in_addr *p, *q; 1095 register struct mbuf *m; 1096 1097 if (ip_nhops == 0) 1098 return ((struct mbuf *)0); 1099 m = m_get(M_DONTWAIT, MT_SOOPTS); 1100 if (m == 0) 1101 return ((struct mbuf *)0); 1102 1103#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1104 1105 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1106 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1107 OPTSIZ; 1108#ifdef DIAGNOSTIC 1109 if (ipprintfs) 1110 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1111#endif 1112 1113 /* 1114 * First save first hop for return route 1115 */ 1116 p = &ip_srcrt.route[ip_nhops - 1]; 1117 *(mtod(m, struct in_addr *)) = *p--; 1118#ifdef DIAGNOSTIC 1119 if (ipprintfs) 1120 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1121#endif 1122 1123 /* 1124 * Copy option fields and padding (nop) to mbuf. 1125 */ 1126 ip_srcrt.nop = IPOPT_NOP; 1127 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1128 bcopy((caddr_t)&ip_srcrt.nop, 1129 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1130 q = (struct in_addr *)(mtod(m, caddr_t) + 1131 sizeof(struct in_addr) + OPTSIZ); 1132#undef OPTSIZ 1133 /* 1134 * Record return path as an IP source route, 1135 * reversing the path (pointers are now aligned). 1136 */ 1137 while (p >= ip_srcrt.route) { 1138#ifdef DIAGNOSTIC 1139 if (ipprintfs) 1140 printf(" %x", ntohl(q->s_addr)); 1141#endif 1142 *q++ = *p--; 1143 } 1144 /* 1145 * Last hop goes to final destination. 1146 */ 1147 *q = ip_srcrt.dst; 1148#ifdef DIAGNOSTIC 1149 if (ipprintfs) 1150 printf(" %x\n", ntohl(q->s_addr)); 1151#endif 1152 return (m); 1153} 1154 1155/* 1156 * Strip out IP options, at higher 1157 * level protocol in the kernel. 1158 * Second argument is buffer to which options 1159 * will be moved, and return value is their length. 1160 * XXX should be deleted; last arg currently ignored. 1161 */ 1162void 1163ip_stripoptions(m, mopt) 1164 register struct mbuf *m; 1165 struct mbuf *mopt; 1166{ 1167 register int i; 1168 struct ip *ip = mtod(m, struct ip *); 1169 register caddr_t opts; 1170 int olen; 1171 1172 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1173 opts = (caddr_t)(ip + 1); 1174 i = m->m_len - (sizeof (struct ip) + olen); 1175 bcopy(opts + olen, opts, (unsigned)i); 1176 m->m_len -= olen; 1177 if (m->m_flags & M_PKTHDR) 1178 m->m_pkthdr.len -= olen; 1179 ip->ip_len -= olen; 1180 ip->ip_hl = sizeof (struct ip) >> 2; 1181} 1182 1183int inetctlerrmap[PRC_NCMDS] = { 1184 0, 0, 0, 0, 1185 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1186 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1187 EMSGSIZE, EHOSTUNREACH, 0, 0, 1188 0, 0, 0, 0, 1189 ENOPROTOOPT 1190}; 1191 1192/* 1193 * Forward a packet. If some error occurs return the sender 1194 * an icmp packet. Note we can't always generate a meaningful 1195 * icmp message because icmp doesn't have a large enough repertoire 1196 * of codes and types. 1197 * 1198 * If not forwarding, just drop the packet. This could be confusing 1199 * if ipforwarding was zero but some routing protocol was advancing 1200 * us as a gateway to somewhere. However, we must let the routing 1201 * protocol deal with that. 1202 * 1203 * The srcrt parameter indicates whether the packet is being forwarded 1204 * via a source route. 1205 */ 1206void 1207ip_forward(m, srcrt) 1208 struct mbuf *m; 1209 int srcrt; 1210{ 1211 register struct ip *ip = mtod(m, struct ip *); 1212 register struct sockaddr_in *sin; 1213 register struct rtentry *rt; 1214 int error, type = 0, code = 0; 1215 struct mbuf *mcopy; 1216 n_long dest; 1217 struct ifnet *destifp; 1218 1219 dest = 0; 1220#ifdef DIAGNOSTIC 1221 if (ipprintfs) 1222 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1223 ntohl(ip->ip_src.s_addr), 1224 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1225#endif 1226 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1227 ipstat.ips_cantforward++; 1228 m_freem(m); 1229 return; 1230 } 1231 if (ip->ip_ttl <= IPTTLDEC) { 1232 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1233 return; 1234 } 1235 ip->ip_ttl -= IPTTLDEC; 1236 1237 sin = satosin(&ipforward_rt.ro_dst); 1238 if ((rt = ipforward_rt.ro_rt) == 0 || 1239 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1240 if (ipforward_rt.ro_rt) { 1241 RTFREE(ipforward_rt.ro_rt); 1242 ipforward_rt.ro_rt = 0; 1243 } 1244 sin->sin_family = AF_INET; 1245 sin->sin_len = sizeof(struct sockaddr_in); 1246 sin->sin_addr = ip->ip_dst; 1247 1248 rtalloc(&ipforward_rt); 1249 if (ipforward_rt.ro_rt == 0) { 1250 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1251 return; 1252 } 1253 rt = ipforward_rt.ro_rt; 1254 } 1255 1256 /* 1257 * Save at most 68 bytes of the packet in case 1258 * we need to generate an ICMP message to the src. 1259 */ 1260 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1261 1262 /* 1263 * If forwarding packet using same interface that it came in on, 1264 * perhaps should send a redirect to sender to shortcut a hop. 1265 * Only send redirect if source is sending directly to us, 1266 * and if packet was not source routed (or has any options). 1267 * Also, don't send redirect if forwarding using a default route 1268 * or a route modified by a redirect. 1269 */ 1270 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1271 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1272 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1273 ipsendredirects && !srcrt) { 1274 if (rt->rt_ifa && 1275 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1276 ifatoia(rt->rt_ifa)->ia_subnet) { 1277 if (rt->rt_flags & RTF_GATEWAY) 1278 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1279 else 1280 dest = ip->ip_dst.s_addr; 1281 /* 1282 * Router requirements says to only send host 1283 * redirects. 1284 */ 1285 type = ICMP_REDIRECT; 1286 code = ICMP_REDIRECT_HOST; 1287#ifdef DIAGNOSTIC 1288 if (ipprintfs) 1289 printf("redirect (%d) to %x\n", code, 1290 (u_int32_t)dest); 1291#endif 1292 } 1293 } 1294 1295 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1296 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1297 if (error) 1298 ipstat.ips_cantforward++; 1299 else { 1300 ipstat.ips_forward++; 1301 if (type) 1302 ipstat.ips_redirectsent++; 1303 else { 1304 if (mcopy) { 1305#ifdef GATEWAY 1306 if (mcopy->m_flags & M_CANFASTFWD) 1307 ipflow_create(&ipforward_rt, mcopy); 1308#endif 1309 m_freem(mcopy); 1310 } 1311 return; 1312 } 1313 } 1314 if (mcopy == NULL) 1315 return; 1316 destifp = NULL; 1317 1318 switch (error) { 1319 1320 case 0: /* forwarded, but need redirect */ 1321 /* type, code set above */ 1322 break; 1323 1324 case ENETUNREACH: /* shouldn't happen, checked above */ 1325 case EHOSTUNREACH: 1326 case ENETDOWN: 1327 case EHOSTDOWN: 1328 default: 1329 type = ICMP_UNREACH; 1330 code = ICMP_UNREACH_HOST; 1331 break; 1332 1333 case EMSGSIZE: 1334 type = ICMP_UNREACH; 1335 code = ICMP_UNREACH_NEEDFRAG; 1336 if (ipforward_rt.ro_rt) 1337 destifp = ipforward_rt.ro_rt->rt_ifp; 1338 ipstat.ips_cantfrag++; 1339 break; 1340 1341 case ENOBUFS: 1342 type = ICMP_SOURCEQUENCH; 1343 code = 0; 1344 break; 1345 } 1346 icmp_error(mcopy, type, code, dest, destifp); 1347} 1348 1349void 1350ip_savecontrol(inp, mp, ip, m) 1351 register struct inpcb *inp; 1352 register struct mbuf **mp; 1353 register struct ip *ip; 1354 register struct mbuf *m; 1355{ 1356 1357 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1358 struct timeval tv; 1359 1360 microtime(&tv); 1361 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1362 SCM_TIMESTAMP, SOL_SOCKET); 1363 if (*mp) 1364 mp = &(*mp)->m_next; 1365 } 1366 if (inp->inp_flags & INP_RECVDSTADDR) { 1367 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1368 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1369 if (*mp) 1370 mp = &(*mp)->m_next; 1371 } 1372#ifdef notyet 1373 /* 1374 * XXX 1375 * Moving these out of udp_input() made them even more broken 1376 * than they already were. 1377 * - fenner@parc.xerox.com 1378 */ 1379 /* options were tossed already */ 1380 if (inp->inp_flags & INP_RECVOPTS) { 1381 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1382 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1383 if (*mp) 1384 mp = &(*mp)->m_next; 1385 } 1386 /* ip_srcroute doesn't do what we want here, need to fix */ 1387 if (inp->inp_flags & INP_RECVRETOPTS) { 1388 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1389 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1390 if (*mp) 1391 mp = &(*mp)->m_next; 1392 } 1393#endif 1394 if (inp->inp_flags & INP_RECVIF) { 1395 struct sockaddr_dl sdl; 1396 1397 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1398 sdl.sdl_family = AF_LINK; 1399 sdl.sdl_index = m->m_pkthdr.rcvif ? 1400 m->m_pkthdr.rcvif->if_index : 0; 1401 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1402 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1403 IP_RECVIF, IPPROTO_IP); 1404 if (*mp) 1405 mp = &(*mp)->m_next; 1406 } 1407} 1408 1409int 1410ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1411 int *name; 1412 u_int namelen; 1413 void *oldp; 1414 size_t *oldlenp; 1415 void *newp; 1416 size_t newlen; 1417{ 1418 extern int subnetsarelocal; 1419 1420 int error, old; 1421 1422 /* All sysctl names at this level are terminal. */ 1423 if (namelen != 1) 1424 return (ENOTDIR); 1425 1426 switch (name[0]) { 1427 case IPCTL_FORWARDING: 1428 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1429 case IPCTL_SENDREDIRECTS: 1430 return (sysctl_int(oldp, oldlenp, newp, newlen, 1431 &ipsendredirects)); 1432 case IPCTL_DEFTTL: 1433 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1434#ifdef notyet 1435 case IPCTL_DEFMTU: 1436 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1437#endif 1438 case IPCTL_FORWSRCRT: 1439 /* Don't allow this to change in a secure environment. */ 1440 if (securelevel > 0) 1441 return (sysctl_rdint(oldp, oldlenp, newp, 1442 ip_forwsrcrt)); 1443 else 1444 return (sysctl_int(oldp, oldlenp, newp, newlen, 1445 &ip_forwsrcrt)); 1446 case IPCTL_DIRECTEDBCAST: 1447 return (sysctl_int(oldp, oldlenp, newp, newlen, 1448 &ip_directedbcast)); 1449 case IPCTL_ALLOWSRCRT: 1450 return (sysctl_int(oldp, oldlenp, newp, newlen, 1451 &ip_allowsrcrt)); 1452 case IPCTL_SUBNETSARELOCAL: 1453 return (sysctl_int(oldp, oldlenp, newp, newlen, 1454 &subnetsarelocal)); 1455 case IPCTL_MTUDISC: 1456 error = sysctl_int(oldp, oldlenp, newp, newlen, 1457 &ip_mtudisc); 1458 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1459 ip_mtudisc_timeout_q = 1460 rt_timer_queue_create(ip_mtudisc_timeout); 1461 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1462 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1463 ip_mtudisc_timeout_q = NULL; 1464 } 1465 return error; 1466 case IPCTL_ANONPORTMIN: 1467 old = anonportmin; 1468 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1469 if (anonportmin >= anonportmax || anonportmin > 65535 1470#ifndef IPNOPRIVPORTS 1471 || anonportmin < IPPORT_RESERVED 1472#endif 1473 ) { 1474 anonportmin = old; 1475 return (EINVAL); 1476 } 1477 return (error); 1478 case IPCTL_ANONPORTMAX: 1479 old = anonportmax; 1480 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1481 if (anonportmin >= anonportmax || anonportmax > 65535 1482#ifndef IPNOPRIVPORTS 1483 || anonportmax < IPPORT_RESERVED 1484#endif 1485 ) { 1486 anonportmax = old; 1487 return (EINVAL); 1488 } 1489 return (error); 1490 case IPCTL_MTUDISCTIMEOUT: 1491 error = sysctl_int(oldp, oldlenp, newp, newlen, 1492 &ip_mtudisc_timeout); 1493 if (ip_mtudisc_timeout_q != NULL) 1494 rt_timer_queue_change(ip_mtudisc_timeout_q, 1495 ip_mtudisc_timeout); 1496 return (error); 1497#ifdef GATEWAY 1498 case IPCTL_MAXFLOWS: 1499 { 1500 int s; 1501 1502 error = sysctl_int(oldp, oldlenp, newp, newlen, 1503 &ip_maxflows); 1504 s = splsoftnet(); 1505 ipflow_reap(0); 1506 splx(s); 1507 return (error); 1508 } 1509#endif 1510 default: 1511 return (EOPNOTSUPP); 1512 } 1513 /* NOTREACHED */ 1514} 1515