ip_input.c revision 1.57
1/* $NetBSD: ip_input.c,v 1.57 1998/02/13 18:21:44 tls Exp $ */ 2 3/* 4 * Copyright (c) 1982, 1986, 1988, 1993 5 * The Regents of the University of California. 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. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 36 */ 37 38#include "opt_mrouting.h" 39 40#include <sys/param.h> 41#include <sys/systm.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/domain.h> 45#include <sys/protosw.h> 46#include <sys/socket.h> 47#include <sys/socketvar.h> 48#include <sys/errno.h> 49#include <sys/time.h> 50#include <sys/kernel.h> 51#include <sys/proc.h> 52 53#include <vm/vm.h> 54#include <sys/sysctl.h> 55 56#include <net/if.h> 57#include <net/if_dl.h> 58#include <net/route.h> 59#include <net/pfil.h> 60 61#include <netinet/in.h> 62#include <netinet/in_systm.h> 63#include <netinet/ip.h> 64#include <netinet/in_pcb.h> 65#include <netinet/in_var.h> 66#include <netinet/ip_var.h> 67#include <netinet/ip_icmp.h> 68 69#ifndef IPFORWARDING 70#ifdef GATEWAY 71#define IPFORWARDING 1 /* forward IP packets not for us */ 72#else /* GATEWAY */ 73#define IPFORWARDING 0 /* don't forward IP packets not for us */ 74#endif /* GATEWAY */ 75#endif /* IPFORWARDING */ 76#ifndef IPSENDREDIRECTS 77#define IPSENDREDIRECTS 1 78#endif 79#ifndef IPFORWSRCRT 80#define IPFORWSRCRT 1 /* forward source-routed packets */ 81#endif 82#ifndef IPALLOWSRCRT 83#define IPALLOWSRCRT 1 /* allow source-routed packets */ 84#endif 85#ifndef IPMTUDISC 86#define IPMTUDISC 0 87#endif 88 89/* 90 * Note: DIRECTED_BROADCAST is handled this way so that previous 91 * configuration using this option will Just Work. 92 */ 93#ifndef IPDIRECTEDBCAST 94#ifdef DIRECTED_BROADCAST 95#define IPDIRECTEDBCAST 1 96#else 97#define IPDIRECTEDBCAST 0 98#endif /* DIRECTED_BROADCAST */ 99#endif /* IPDIRECTEDBCAST */ 100int ipforwarding = IPFORWARDING; 101int ipsendredirects = IPSENDREDIRECTS; 102int ip_defttl = IPDEFTTL; 103int ip_forwsrcrt = IPFORWSRCRT; 104int ip_directedbcast = IPDIRECTEDBCAST; 105int ip_allowsrcrt = IPALLOWSRCRT; 106int ip_mtudisc = IPMTUDISC; 107#ifdef DIAGNOSTIC 108int ipprintfs = 0; 109#endif 110 111extern struct domain inetdomain; 112extern struct protosw inetsw[]; 113u_char ip_protox[IPPROTO_MAX]; 114int ipqmaxlen = IFQ_MAXLEN; 115struct in_ifaddrhead in_ifaddr; 116struct in_ifaddrhashhead *in_ifaddrhashtbl; 117struct ifqueue ipintrq; 118 119/* 120 * We need to save the IP options in case a protocol wants to respond 121 * to an incoming packet over the same route if the packet got here 122 * using IP source routing. This allows connection establishment and 123 * maintenance when the remote end is on a network that is not known 124 * to us. 125 */ 126int ip_nhops = 0; 127static struct ip_srcrt { 128 struct in_addr dst; /* final destination */ 129 char nop; /* one NOP to align */ 130 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 131 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 132} ip_srcrt; 133 134static void save_rte __P((u_char *, struct in_addr)); 135 136/* 137 * IP initialization: fill in IP protocol switch table. 138 * All protocols not implemented in kernel go to raw IP protocol handler. 139 */ 140void 141ip_init() 142{ 143 register struct protosw *pr; 144 register int i; 145 146 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 147 if (pr == 0) 148 panic("ip_init"); 149 for (i = 0; i < IPPROTO_MAX; i++) 150 ip_protox[i] = pr - inetsw; 151 for (pr = inetdomain.dom_protosw; 152 pr < inetdomain.dom_protoswNPROTOSW; pr++) 153 if (pr->pr_domain->dom_family == PF_INET && 154 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 155 ip_protox[pr->pr_protocol] = pr - inetsw; 156 LIST_INIT(&ipq); 157 ip_id = time.tv_sec & 0xffff; 158 ipintrq.ifq_maxlen = ipqmaxlen; 159 TAILQ_INIT(&in_ifaddr); 160 in_ifaddrhashtbl = 161 hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); 162} 163 164struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 165struct route ipforward_rt; 166 167/* 168 * Ip input routine. Checksum and byte swap header. If fragmented 169 * try to reassemble. Process options. Pass to next level. 170 */ 171void 172ipintr() 173{ 174 register struct ip *ip = NULL; 175 register struct mbuf *m; 176 register struct ipq *fp; 177 register struct in_ifaddr *ia; 178 register struct ifaddr *ifa; 179 struct ipqent *ipqe; 180 int hlen = 0, mff, len, s; 181#ifdef PFIL_HOOKS 182 struct packet_filter_hook *pfh; 183 struct mbuf *m0; 184 int rv; 185#endif /* PFIL_HOOKS */ 186 187next: 188 /* 189 * Get next datagram off input queue and get IP header 190 * in first mbuf. 191 */ 192 s = splimp(); 193 IF_DEQUEUE(&ipintrq, m); 194 splx(s); 195 if (m == 0) 196 return; 197#ifdef DIAGNOSTIC 198 if ((m->m_flags & M_PKTHDR) == 0) 199 panic("ipintr no HDR"); 200#endif 201 /* 202 * If no IP addresses have been set yet but the interfaces 203 * are receiving, can't do anything with incoming packets yet. 204 */ 205 if (in_ifaddr.tqh_first == 0) 206 goto bad; 207 ipstat.ips_total++; 208 if (m->m_len < sizeof (struct ip) && 209 (m = m_pullup(m, sizeof (struct ip))) == 0) { 210 ipstat.ips_toosmall++; 211 goto next; 212 } 213 ip = mtod(m, struct ip *); 214 if (ip->ip_v != IPVERSION) { 215 ipstat.ips_badvers++; 216 goto bad; 217 } 218 hlen = ip->ip_hl << 2; 219 if (hlen < sizeof(struct ip)) { /* minimum header length */ 220 ipstat.ips_badhlen++; 221 goto bad; 222 } 223 if (hlen > m->m_len) { 224 if ((m = m_pullup(m, hlen)) == 0) { 225 ipstat.ips_badhlen++; 226 goto next; 227 } 228 ip = mtod(m, struct ip *); 229 } 230 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 231 ipstat.ips_badsum++; 232 goto bad; 233 } 234 235 /* 236 * Convert fields to host representation. 237 */ 238 NTOHS(ip->ip_len); 239 NTOHS(ip->ip_id); 240 NTOHS(ip->ip_off); 241 len = ip->ip_len; 242 243 /* 244 * Check that the amount of data in the buffers 245 * is as at least much as the IP header would have us expect. 246 * Trim mbufs if longer than we expect. 247 * Drop packet if shorter than we expect. 248 */ 249 if (m->m_pkthdr.len < len) { 250 ipstat.ips_tooshort++; 251 goto bad; 252 } 253 if (m->m_pkthdr.len > len) { 254 if (m->m_len == m->m_pkthdr.len) { 255 m->m_len = len; 256 m->m_pkthdr.len = len; 257 } else 258 m_adj(m, len - m->m_pkthdr.len); 259 } 260 261#ifdef PFIL_HOOKS 262 /* 263 * Run through list of hooks for input packets. 264 */ 265 m0 = m; 266 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 267 if (pfh->pfil_func) { 268 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 269 if (rv) 270 goto next; 271 ip = mtod(m = m0, struct ip *); 272 } 273#endif /* PFIL_HOOKS */ 274 275 /* 276 * Process options and, if not destined for us, 277 * ship it on. ip_dooptions returns 1 when an 278 * error was detected (causing an icmp message 279 * to be sent and the original packet to be freed). 280 */ 281 ip_nhops = 0; /* for source routed packets */ 282 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 283 goto next; 284 285 /* 286 * Check our list of addresses, to see if the packet is for us. 287 */ 288 INADDR_TO_IA(ip->ip_dst, ia); 289 if (ia != NULL) goto ours; 290 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 291 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 292 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 293 if (ifa->ifa_addr->sa_family != AF_INET) continue; 294 ia = ifatoia(ifa); 295 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 296 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 297 /* 298 * Look for all-0's host part (old broadcast addr), 299 * either for subnet or net. 300 */ 301 ip->ip_dst.s_addr == ia->ia_subnet || 302 ip->ip_dst.s_addr == ia->ia_net) 303 goto ours; 304 /* 305 * An interface with IP address zero accepts 306 * all packets that arrive on that interface. 307 */ 308 if (in_nullhost(ia->ia_addr.sin_addr)) 309 goto ours; 310 } 311 } 312 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 313 struct in_multi *inm; 314#ifdef MROUTING 315 extern struct socket *ip_mrouter; 316 317 if (m->m_flags & M_EXT) { 318 if ((m = m_pullup(m, hlen)) == 0) { 319 ipstat.ips_toosmall++; 320 goto next; 321 } 322 ip = mtod(m, struct ip *); 323 } 324 325 if (ip_mrouter) { 326 /* 327 * If we are acting as a multicast router, all 328 * incoming multicast packets are passed to the 329 * kernel-level multicast forwarding function. 330 * The packet is returned (relatively) intact; if 331 * ip_mforward() returns a non-zero value, the packet 332 * must be discarded, else it may be accepted below. 333 * 334 * (The IP ident field is put in the same byte order 335 * as expected when ip_mforward() is called from 336 * ip_output().) 337 */ 338 ip->ip_id = htons(ip->ip_id); 339 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 340 ipstat.ips_cantforward++; 341 m_freem(m); 342 goto next; 343 } 344 ip->ip_id = ntohs(ip->ip_id); 345 346 /* 347 * The process-level routing demon needs to receive 348 * all multicast IGMP packets, whether or not this 349 * host belongs to their destination groups. 350 */ 351 if (ip->ip_p == IPPROTO_IGMP) 352 goto ours; 353 ipstat.ips_forward++; 354 } 355#endif 356 /* 357 * See if we belong to the destination multicast group on the 358 * arrival interface. 359 */ 360 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 361 if (inm == NULL) { 362 ipstat.ips_cantforward++; 363 m_freem(m); 364 goto next; 365 } 366 goto ours; 367 } 368 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 369 in_nullhost(ip->ip_dst)) 370 goto ours; 371 372 /* 373 * Not for us; forward if possible and desirable. 374 */ 375 if (ipforwarding == 0) { 376 ipstat.ips_cantforward++; 377 m_freem(m); 378 } else 379 ip_forward(m, 0); 380 goto next; 381 382ours: 383 /* 384 * If offset or IP_MF are set, must reassemble. 385 * Otherwise, nothing need be done. 386 * (We could look in the reassembly queue to see 387 * if the packet was previously fragmented, 388 * but it's not worth the time; just let them time out.) 389 */ 390 if (ip->ip_off & ~(IP_DF|IP_RF)) { 391 /* 392 * Look for queue of fragments 393 * of this datagram. 394 */ 395 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 396 if (ip->ip_id == fp->ipq_id && 397 in_hosteq(ip->ip_src, fp->ipq_src) && 398 in_hosteq(ip->ip_dst, fp->ipq_dst) && 399 ip->ip_p == fp->ipq_p) 400 goto found; 401 fp = 0; 402found: 403 404 /* 405 * Adjust ip_len to not reflect header, 406 * set ipqe_mff if more fragments are expected, 407 * convert offset of this to bytes. 408 */ 409 ip->ip_len -= hlen; 410 mff = (ip->ip_off & IP_MF) != 0; 411 if (mff) { 412 /* 413 * Make sure that fragments have a data length 414 * that's a non-zero multiple of 8 bytes. 415 */ 416 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 417 ipstat.ips_badfrags++; 418 goto bad; 419 } 420 } 421 ip->ip_off <<= 3; 422 423 /* 424 * If datagram marked as having more fragments 425 * or if this is not the first fragment, 426 * attempt reassembly; if it succeeds, proceed. 427 */ 428 if (mff || ip->ip_off) { 429 ipstat.ips_fragments++; 430 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 431 M_IPQ, M_NOWAIT); 432 if (ipqe == NULL) { 433 ipstat.ips_rcvmemdrop++; 434 goto bad; 435 } 436 ipqe->ipqe_mff = mff; 437 ipqe->ipqe_m = m; 438 ipqe->ipqe_ip = ip; 439 m = ip_reass(ipqe, fp); 440 if (m == 0) 441 goto next; 442 ipstat.ips_reassembled++; 443 ip = mtod(m, struct ip *); 444 } else 445 if (fp) 446 ip_freef(fp); 447 } else 448 ip->ip_len -= hlen; 449 450 /* 451 * Switch out to protocol's input routine. 452 */ 453 ipstat.ips_delivered++; 454 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 455 goto next; 456bad: 457 m_freem(m); 458 goto next; 459} 460 461/* 462 * Take incoming datagram fragment and try to 463 * reassemble it into whole datagram. If a chain for 464 * reassembly of this datagram already exists, then it 465 * is given as fp; otherwise have to make a chain. 466 */ 467struct mbuf * 468ip_reass(ipqe, fp) 469 register struct ipqent *ipqe; 470 register struct ipq *fp; 471{ 472 register struct mbuf *m = ipqe->ipqe_m; 473 register struct ipqent *nq, *p, *q; 474 struct ip *ip; 475 struct mbuf *t; 476 int hlen = ipqe->ipqe_ip->ip_hl << 2; 477 int i, next; 478 479 /* 480 * Presence of header sizes in mbufs 481 * would confuse code below. 482 */ 483 m->m_data += hlen; 484 m->m_len -= hlen; 485 486 /* 487 * If first fragment to arrive, create a reassembly queue. 488 */ 489 if (fp == 0) { 490 MALLOC(fp, struct ipq *, sizeof (struct ipq), 491 M_FTABLE, M_NOWAIT); 492 if (fp == NULL) 493 goto dropfrag; 494 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 495 fp->ipq_ttl = IPFRAGTTL; 496 fp->ipq_p = ipqe->ipqe_ip->ip_p; 497 fp->ipq_id = ipqe->ipqe_ip->ip_id; 498 LIST_INIT(&fp->ipq_fragq); 499 fp->ipq_src = ipqe->ipqe_ip->ip_src; 500 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 501 p = NULL; 502 goto insert; 503 } 504 505 /* 506 * Find a segment which begins after this one does. 507 */ 508 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 509 p = q, q = q->ipqe_q.le_next) 510 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 511 break; 512 513 /* 514 * If there is a preceding segment, it may provide some of 515 * our data already. If so, drop the data from the incoming 516 * segment. If it provides all of our data, drop us. 517 */ 518 if (p != NULL) { 519 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 520 ipqe->ipqe_ip->ip_off; 521 if (i > 0) { 522 if (i >= ipqe->ipqe_ip->ip_len) 523 goto dropfrag; 524 m_adj(ipqe->ipqe_m, i); 525 ipqe->ipqe_ip->ip_off += i; 526 ipqe->ipqe_ip->ip_len -= i; 527 } 528 } 529 530 /* 531 * While we overlap succeeding segments trim them or, 532 * if they are completely covered, dequeue them. 533 */ 534 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 535 q->ipqe_ip->ip_off; q = nq) { 536 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 537 q->ipqe_ip->ip_off; 538 if (i < q->ipqe_ip->ip_len) { 539 q->ipqe_ip->ip_len -= i; 540 q->ipqe_ip->ip_off += i; 541 m_adj(q->ipqe_m, i); 542 break; 543 } 544 nq = q->ipqe_q.le_next; 545 m_freem(q->ipqe_m); 546 LIST_REMOVE(q, ipqe_q); 547 FREE(q, M_IPQ); 548 } 549 550insert: 551 /* 552 * Stick new segment in its place; 553 * check for complete reassembly. 554 */ 555 if (p == NULL) { 556 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 557 } else { 558 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 559 } 560 next = 0; 561 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 562 p = q, q = q->ipqe_q.le_next) { 563 if (q->ipqe_ip->ip_off != next) 564 return (0); 565 next += q->ipqe_ip->ip_len; 566 } 567 if (p->ipqe_mff) 568 return (0); 569 570 /* 571 * Reassembly is complete. Check for a bogus message size and 572 * concatenate fragments. 573 */ 574 q = fp->ipq_fragq.lh_first; 575 ip = q->ipqe_ip; 576 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 577 ipstat.ips_toolong++; 578 ip_freef(fp); 579 return (0); 580 } 581 m = q->ipqe_m; 582 t = m->m_next; 583 m->m_next = 0; 584 m_cat(m, t); 585 nq = q->ipqe_q.le_next; 586 FREE(q, M_IPQ); 587 for (q = nq; q != NULL; q = nq) { 588 t = q->ipqe_m; 589 nq = q->ipqe_q.le_next; 590 FREE(q, M_IPQ); 591 m_cat(m, t); 592 } 593 594 /* 595 * Create header for new ip packet by 596 * modifying header of first packet; 597 * dequeue and discard fragment reassembly header. 598 * Make header visible. 599 */ 600 ip->ip_len = next; 601 ip->ip_src = fp->ipq_src; 602 ip->ip_dst = fp->ipq_dst; 603 LIST_REMOVE(fp, ipq_q); 604 FREE(fp, M_FTABLE); 605 m->m_len += (ip->ip_hl << 2); 606 m->m_data -= (ip->ip_hl << 2); 607 /* some debugging cruft by sklower, below, will go away soon */ 608 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 609 register int plen = 0; 610 for (t = m; t; t = t->m_next) 611 plen += t->m_len; 612 m->m_pkthdr.len = plen; 613 } 614 return (m); 615 616dropfrag: 617 ipstat.ips_fragdropped++; 618 m_freem(m); 619 FREE(ipqe, M_IPQ); 620 return (0); 621} 622 623/* 624 * Free a fragment reassembly header and all 625 * associated datagrams. 626 */ 627void 628ip_freef(fp) 629 struct ipq *fp; 630{ 631 register struct ipqent *q, *p; 632 633 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 634 p = q->ipqe_q.le_next; 635 m_freem(q->ipqe_m); 636 LIST_REMOVE(q, ipqe_q); 637 FREE(q, M_IPQ); 638 } 639 LIST_REMOVE(fp, ipq_q); 640 FREE(fp, M_FTABLE); 641} 642 643/* 644 * IP timer processing; 645 * if a timer expires on a reassembly 646 * queue, discard it. 647 */ 648void 649ip_slowtimo() 650{ 651 register struct ipq *fp, *nfp; 652 int s = splsoftnet(); 653 654 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 655 nfp = fp->ipq_q.le_next; 656 if (--fp->ipq_ttl == 0) { 657 ipstat.ips_fragtimeout++; 658 ip_freef(fp); 659 } 660 } 661 splx(s); 662} 663 664/* 665 * Drain off all datagram fragments. 666 */ 667void 668ip_drain() 669{ 670 671 while (ipq.lh_first != NULL) { 672 ipstat.ips_fragdropped++; 673 ip_freef(ipq.lh_first); 674 } 675} 676 677/* 678 * Do option processing on a datagram, 679 * possibly discarding it if bad options are encountered, 680 * or forwarding it if source-routed. 681 * Returns 1 if packet has been forwarded/freed, 682 * 0 if the packet should be processed further. 683 */ 684int 685ip_dooptions(m) 686 struct mbuf *m; 687{ 688 register struct ip *ip = mtod(m, struct ip *); 689 register u_char *cp; 690 register struct ip_timestamp *ipt; 691 register struct in_ifaddr *ia; 692 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 693 struct in_addr *sin, dst; 694 n_time ntime; 695 696 dst = ip->ip_dst; 697 cp = (u_char *)(ip + 1); 698 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 699 for (; cnt > 0; cnt -= optlen, cp += optlen) { 700 opt = cp[IPOPT_OPTVAL]; 701 if (opt == IPOPT_EOL) 702 break; 703 if (opt == IPOPT_NOP) 704 optlen = 1; 705 else { 706 optlen = cp[IPOPT_OLEN]; 707 if (optlen <= 0 || optlen > cnt) { 708 code = &cp[IPOPT_OLEN] - (u_char *)ip; 709 goto bad; 710 } 711 } 712 switch (opt) { 713 714 default: 715 break; 716 717 /* 718 * Source routing with record. 719 * Find interface with current destination address. 720 * If none on this machine then drop if strictly routed, 721 * or do nothing if loosely routed. 722 * Record interface address and bring up next address 723 * component. If strictly routed make sure next 724 * address is on directly accessible net. 725 */ 726 case IPOPT_LSRR: 727 case IPOPT_SSRR: 728 if (ip_allowsrcrt == 0) { 729 type = ICMP_UNREACH; 730 code = ICMP_UNREACH_NET_PROHIB; 731 goto bad; 732 } 733 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 734 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 735 goto bad; 736 } 737 ipaddr.sin_addr = ip->ip_dst; 738 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 739 if (ia == 0) { 740 if (opt == IPOPT_SSRR) { 741 type = ICMP_UNREACH; 742 code = ICMP_UNREACH_SRCFAIL; 743 goto bad; 744 } 745 /* 746 * Loose routing, and not at next destination 747 * yet; nothing to do except forward. 748 */ 749 break; 750 } 751 off--; /* 0 origin */ 752 if (off > optlen - sizeof(struct in_addr)) { 753 /* 754 * End of source route. Should be for us. 755 */ 756 save_rte(cp, ip->ip_src); 757 break; 758 } 759 /* 760 * locate outgoing interface 761 */ 762 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 763 sizeof(ipaddr.sin_addr)); 764 if (opt == IPOPT_SSRR) { 765#define INA struct in_ifaddr * 766#define SA struct sockaddr * 767 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 768 } else 769 ia = ip_rtaddr(ipaddr.sin_addr); 770 if (ia == 0) { 771 type = ICMP_UNREACH; 772 code = ICMP_UNREACH_SRCFAIL; 773 goto bad; 774 } 775 ip->ip_dst = ipaddr.sin_addr; 776 bcopy((caddr_t)&ia->ia_addr.sin_addr, 777 (caddr_t)(cp + off), sizeof(struct in_addr)); 778 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 779 /* 780 * Let ip_intr's mcast routing check handle mcast pkts 781 */ 782 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 783 break; 784 785 case IPOPT_RR: 786 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 787 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 788 goto bad; 789 } 790 /* 791 * If no space remains, ignore. 792 */ 793 off--; /* 0 origin */ 794 if (off > optlen - sizeof(struct in_addr)) 795 break; 796 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 797 sizeof(ipaddr.sin_addr)); 798 /* 799 * locate outgoing interface; if we're the destination, 800 * use the incoming interface (should be same). 801 */ 802 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 803 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 804 type = ICMP_UNREACH; 805 code = ICMP_UNREACH_HOST; 806 goto bad; 807 } 808 bcopy((caddr_t)&ia->ia_addr.sin_addr, 809 (caddr_t)(cp + off), sizeof(struct in_addr)); 810 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 811 break; 812 813 case IPOPT_TS: 814 code = cp - (u_char *)ip; 815 ipt = (struct ip_timestamp *)cp; 816 if (ipt->ipt_len < 5) 817 goto bad; 818 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 819 if (++ipt->ipt_oflw == 0) 820 goto bad; 821 break; 822 } 823 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 824 switch (ipt->ipt_flg) { 825 826 case IPOPT_TS_TSONLY: 827 break; 828 829 case IPOPT_TS_TSANDADDR: 830 if (ipt->ipt_ptr + sizeof(n_time) + 831 sizeof(struct in_addr) > ipt->ipt_len) 832 goto bad; 833 ipaddr.sin_addr = dst; 834 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 835 m->m_pkthdr.rcvif); 836 if (ia == 0) 837 continue; 838 bcopy((caddr_t)&ia->ia_addr.sin_addr, 839 (caddr_t)sin, sizeof(struct in_addr)); 840 ipt->ipt_ptr += sizeof(struct in_addr); 841 break; 842 843 case IPOPT_TS_PRESPEC: 844 if (ipt->ipt_ptr + sizeof(n_time) + 845 sizeof(struct in_addr) > ipt->ipt_len) 846 goto bad; 847 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 848 sizeof(struct in_addr)); 849 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 850 continue; 851 ipt->ipt_ptr += sizeof(struct in_addr); 852 break; 853 854 default: 855 goto bad; 856 } 857 ntime = iptime(); 858 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 859 sizeof(n_time)); 860 ipt->ipt_ptr += sizeof(n_time); 861 } 862 } 863 if (forward) { 864 if (ip_forwsrcrt == 0) { 865 type = ICMP_UNREACH; 866 code = ICMP_UNREACH_SRCFAIL; 867 goto bad; 868 } 869 ip_forward(m, 1); 870 return (1); 871 } 872 return (0); 873bad: 874 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 875 icmp_error(m, type, code, 0, 0); 876 ipstat.ips_badoptions++; 877 return (1); 878} 879 880/* 881 * Given address of next destination (final or next hop), 882 * return internet address info of interface to be used to get there. 883 */ 884struct in_ifaddr * 885ip_rtaddr(dst) 886 struct in_addr dst; 887{ 888 register struct sockaddr_in *sin; 889 890 sin = satosin(&ipforward_rt.ro_dst); 891 892 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 893 if (ipforward_rt.ro_rt) { 894 RTFREE(ipforward_rt.ro_rt); 895 ipforward_rt.ro_rt = 0; 896 } 897 sin->sin_family = AF_INET; 898 sin->sin_len = sizeof(*sin); 899 sin->sin_addr = dst; 900 901 rtalloc(&ipforward_rt); 902 } 903 if (ipforward_rt.ro_rt == 0) 904 return ((struct in_ifaddr *)0); 905 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 906} 907 908/* 909 * Save incoming source route for use in replies, 910 * to be picked up later by ip_srcroute if the receiver is interested. 911 */ 912void 913save_rte(option, dst) 914 u_char *option; 915 struct in_addr dst; 916{ 917 unsigned olen; 918 919 olen = option[IPOPT_OLEN]; 920#ifdef DIAGNOSTIC 921 if (ipprintfs) 922 printf("save_rte: olen %d\n", olen); 923#endif 924 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 925 return; 926 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 927 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 928 ip_srcrt.dst = dst; 929} 930 931/* 932 * Retrieve incoming source route for use in replies, 933 * in the same form used by setsockopt. 934 * The first hop is placed before the options, will be removed later. 935 */ 936struct mbuf * 937ip_srcroute() 938{ 939 register struct in_addr *p, *q; 940 register struct mbuf *m; 941 942 if (ip_nhops == 0) 943 return ((struct mbuf *)0); 944 m = m_get(M_DONTWAIT, MT_SOOPTS); 945 if (m == 0) 946 return ((struct mbuf *)0); 947 948#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 949 950 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 951 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 952 OPTSIZ; 953#ifdef DIAGNOSTIC 954 if (ipprintfs) 955 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 956#endif 957 958 /* 959 * First save first hop for return route 960 */ 961 p = &ip_srcrt.route[ip_nhops - 1]; 962 *(mtod(m, struct in_addr *)) = *p--; 963#ifdef DIAGNOSTIC 964 if (ipprintfs) 965 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 966#endif 967 968 /* 969 * Copy option fields and padding (nop) to mbuf. 970 */ 971 ip_srcrt.nop = IPOPT_NOP; 972 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 973 bcopy((caddr_t)&ip_srcrt.nop, 974 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 975 q = (struct in_addr *)(mtod(m, caddr_t) + 976 sizeof(struct in_addr) + OPTSIZ); 977#undef OPTSIZ 978 /* 979 * Record return path as an IP source route, 980 * reversing the path (pointers are now aligned). 981 */ 982 while (p >= ip_srcrt.route) { 983#ifdef DIAGNOSTIC 984 if (ipprintfs) 985 printf(" %x", ntohl(q->s_addr)); 986#endif 987 *q++ = *p--; 988 } 989 /* 990 * Last hop goes to final destination. 991 */ 992 *q = ip_srcrt.dst; 993#ifdef DIAGNOSTIC 994 if (ipprintfs) 995 printf(" %x\n", ntohl(q->s_addr)); 996#endif 997 return (m); 998} 999 1000/* 1001 * Strip out IP options, at higher 1002 * level protocol in the kernel. 1003 * Second argument is buffer to which options 1004 * will be moved, and return value is their length. 1005 * XXX should be deleted; last arg currently ignored. 1006 */ 1007void 1008ip_stripoptions(m, mopt) 1009 register struct mbuf *m; 1010 struct mbuf *mopt; 1011{ 1012 register int i; 1013 struct ip *ip = mtod(m, struct ip *); 1014 register caddr_t opts; 1015 int olen; 1016 1017 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1018 opts = (caddr_t)(ip + 1); 1019 i = m->m_len - (sizeof (struct ip) + olen); 1020 bcopy(opts + olen, opts, (unsigned)i); 1021 m->m_len -= olen; 1022 if (m->m_flags & M_PKTHDR) 1023 m->m_pkthdr.len -= olen; 1024 ip->ip_hl = sizeof(struct ip) >> 2; 1025} 1026 1027int inetctlerrmap[PRC_NCMDS] = { 1028 0, 0, 0, 0, 1029 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1030 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1031 EMSGSIZE, EHOSTUNREACH, 0, 0, 1032 0, 0, 0, 0, 1033 ENOPROTOOPT 1034}; 1035 1036/* 1037 * Forward a packet. If some error occurs return the sender 1038 * an icmp packet. Note we can't always generate a meaningful 1039 * icmp message because icmp doesn't have a large enough repertoire 1040 * of codes and types. 1041 * 1042 * If not forwarding, just drop the packet. This could be confusing 1043 * if ipforwarding was zero but some routing protocol was advancing 1044 * us as a gateway to somewhere. However, we must let the routing 1045 * protocol deal with that. 1046 * 1047 * The srcrt parameter indicates whether the packet is being forwarded 1048 * via a source route. 1049 */ 1050void 1051ip_forward(m, srcrt) 1052 struct mbuf *m; 1053 int srcrt; 1054{ 1055 register struct ip *ip = mtod(m, struct ip *); 1056 register struct sockaddr_in *sin; 1057 register struct rtentry *rt; 1058 int error, type = 0, code = 0; 1059 struct mbuf *mcopy; 1060 n_long dest; 1061 struct ifnet *destifp; 1062 1063 dest = 0; 1064#ifdef DIAGNOSTIC 1065 if (ipprintfs) 1066 printf("forward: src %x dst %x ttl %x\n", 1067 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1068#endif 1069 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1070 ipstat.ips_cantforward++; 1071 m_freem(m); 1072 return; 1073 } 1074 HTONS(ip->ip_id); 1075 if (ip->ip_ttl <= IPTTLDEC) { 1076 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1077 return; 1078 } 1079 ip->ip_ttl -= IPTTLDEC; 1080 1081 sin = satosin(&ipforward_rt.ro_dst); 1082 if ((rt = ipforward_rt.ro_rt) == 0 || 1083 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1084 if (ipforward_rt.ro_rt) { 1085 RTFREE(ipforward_rt.ro_rt); 1086 ipforward_rt.ro_rt = 0; 1087 } 1088 sin->sin_family = AF_INET; 1089 sin->sin_len = sizeof(struct sockaddr_in); 1090 sin->sin_addr = ip->ip_dst; 1091 1092 rtalloc(&ipforward_rt); 1093 if (ipforward_rt.ro_rt == 0) { 1094 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1095 return; 1096 } 1097 rt = ipforward_rt.ro_rt; 1098 } 1099 1100 /* 1101 * Save at most 68 bytes of the packet in case 1102 * we need to generate an ICMP message to the src. 1103 */ 1104 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1105 1106 /* 1107 * If forwarding packet using same interface that it came in on, 1108 * perhaps should send a redirect to sender to shortcut a hop. 1109 * Only send redirect if source is sending directly to us, 1110 * and if packet was not source routed (or has any options). 1111 * Also, don't send redirect if forwarding using a default route 1112 * or a route modified by a redirect. 1113 */ 1114 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1115 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1116 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1117 ipsendredirects && !srcrt) { 1118 if (rt->rt_ifa && 1119 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1120 ifatoia(rt->rt_ifa)->ia_subnet) { 1121 if (rt->rt_flags & RTF_GATEWAY) 1122 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1123 else 1124 dest = ip->ip_dst.s_addr; 1125 /* Router requirements says to only send host redirects */ 1126 type = ICMP_REDIRECT; 1127 code = ICMP_REDIRECT_HOST; 1128#ifdef DIAGNOSTIC 1129 if (ipprintfs) 1130 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1131#endif 1132 } 1133 } 1134 1135 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1136 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1137 if (error) 1138 ipstat.ips_cantforward++; 1139 else { 1140 ipstat.ips_forward++; 1141 if (type) 1142 ipstat.ips_redirectsent++; 1143 else { 1144 if (mcopy) 1145 m_freem(mcopy); 1146 return; 1147 } 1148 } 1149 if (mcopy == NULL) 1150 return; 1151 destifp = NULL; 1152 1153 switch (error) { 1154 1155 case 0: /* forwarded, but need redirect */ 1156 /* type, code set above */ 1157 break; 1158 1159 case ENETUNREACH: /* shouldn't happen, checked above */ 1160 case EHOSTUNREACH: 1161 case ENETDOWN: 1162 case EHOSTDOWN: 1163 default: 1164 type = ICMP_UNREACH; 1165 code = ICMP_UNREACH_HOST; 1166 break; 1167 1168 case EMSGSIZE: 1169 type = ICMP_UNREACH; 1170 code = ICMP_UNREACH_NEEDFRAG; 1171 if (ipforward_rt.ro_rt) 1172 destifp = ipforward_rt.ro_rt->rt_ifp; 1173 ipstat.ips_cantfrag++; 1174 break; 1175 1176 case ENOBUFS: 1177 type = ICMP_SOURCEQUENCH; 1178 code = 0; 1179 break; 1180 } 1181 icmp_error(mcopy, type, code, dest, destifp); 1182} 1183 1184void 1185ip_savecontrol(inp, mp, ip, m) 1186 register struct inpcb *inp; 1187 register struct mbuf **mp; 1188 register struct ip *ip; 1189 register struct mbuf *m; 1190{ 1191 1192 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1193 struct timeval tv; 1194 1195 microtime(&tv); 1196 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1197 SCM_TIMESTAMP, SOL_SOCKET); 1198 if (*mp) 1199 mp = &(*mp)->m_next; 1200 } 1201 if (inp->inp_flags & INP_RECVDSTADDR) { 1202 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1203 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1204 if (*mp) 1205 mp = &(*mp)->m_next; 1206 } 1207#ifdef notyet 1208 /* 1209 * XXX 1210 * Moving these out of udp_input() made them even more broken 1211 * than they already were. 1212 * - fenner@parc.xerox.com 1213 */ 1214 /* options were tossed already */ 1215 if (inp->inp_flags & INP_RECVOPTS) { 1216 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1217 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1218 if (*mp) 1219 mp = &(*mp)->m_next; 1220 } 1221 /* ip_srcroute doesn't do what we want here, need to fix */ 1222 if (inp->inp_flags & INP_RECVRETOPTS) { 1223 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1224 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1225 if (*mp) 1226 mp = &(*mp)->m_next; 1227 } 1228#endif 1229 if (inp->inp_flags & INP_RECVIF) { 1230 struct sockaddr_dl sdl; 1231 1232 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1233 sdl.sdl_family = AF_LINK; 1234 sdl.sdl_index = m->m_pkthdr.rcvif ? 1235 m->m_pkthdr.rcvif->if_index : 0; 1236 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1237 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1238 IP_RECVIF, IPPROTO_IP); 1239 if (*mp) 1240 mp = &(*mp)->m_next; 1241 } 1242} 1243 1244int 1245ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1246 int *name; 1247 u_int namelen; 1248 void *oldp; 1249 size_t *oldlenp; 1250 void *newp; 1251 size_t newlen; 1252{ 1253 extern int subnetsarelocal; 1254 1255 int error, old; 1256 1257 /* All sysctl names at this level are terminal. */ 1258 if (namelen != 1) 1259 return (ENOTDIR); 1260 1261 switch (name[0]) { 1262 case IPCTL_FORWARDING: 1263 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1264 case IPCTL_SENDREDIRECTS: 1265 return (sysctl_int(oldp, oldlenp, newp, newlen, 1266 &ipsendredirects)); 1267 case IPCTL_DEFTTL: 1268 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1269#ifdef notyet 1270 case IPCTL_DEFMTU: 1271 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1272#endif 1273 case IPCTL_FORWSRCRT: 1274 /* Don't allow this to change in a secure environment. */ 1275 if (securelevel > 0) 1276 return (sysctl_rdint(oldp, oldlenp, newp, 1277 ip_forwsrcrt)); 1278 else 1279 return (sysctl_int(oldp, oldlenp, newp, newlen, 1280 &ip_forwsrcrt)); 1281 case IPCTL_DIRECTEDBCAST: 1282 return (sysctl_int(oldp, oldlenp, newp, newlen, 1283 &ip_directedbcast)); 1284 case IPCTL_ALLOWSRCRT: 1285 return (sysctl_int(oldp, oldlenp, newp, newlen, 1286 &ip_allowsrcrt)); 1287 case IPCTL_SUBNETSARELOCAL: 1288 return (sysctl_int(oldp, oldlenp, newp, newlen, 1289 &subnetsarelocal)); 1290 case IPCTL_MTUDISC: 1291 return (sysctl_int(oldp, oldlenp, newp, newlen, 1292 &ip_mtudisc)); 1293 case IPCTL_ANONPORTMIN: 1294 old = anonportmin; 1295 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1296 if (anonportmin >= anonportmax || anonportmin > 65535 1297#ifndef IPNOPRIVPORTS 1298 || anonportmin < IPPORT_RESERVED 1299#endif 1300 ) { 1301 anonportmin = old; 1302 return (EINVAL); 1303 } 1304 return (error); 1305 case IPCTL_ANONPORTMAX: 1306 old = anonportmax; 1307 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1308 if (anonportmin >= anonportmax || anonportmax > 65535 1309#ifndef IPNOPRIVPORTS 1310 || anonportmax < IPPORT_RESERVED 1311#endif 1312 ) { 1313 anonportmax = old; 1314 return (EINVAL); 1315 } 1316 return (error); 1317 default: 1318 return (EOPNOTSUPP); 1319 } 1320 /* NOTREACHED */ 1321} 1322