ip_input.c revision 1.35
1/* $NetBSD: ip_input.c,v 1.35 1996/09/09 14:51:16 mycroft 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 <sys/param.h> 39#include <sys/systm.h> 40#include <sys/malloc.h> 41#include <sys/mbuf.h> 42#include <sys/domain.h> 43#include <sys/protosw.h> 44#include <sys/socket.h> 45#include <sys/errno.h> 46#include <sys/time.h> 47#include <sys/kernel.h> 48#include <sys/proc.h> 49 50#include <vm/vm.h> 51#include <sys/sysctl.h> 52 53#include <net/if.h> 54#include <net/route.h> 55 56#include <netinet/in.h> 57#include <netinet/in_systm.h> 58#include <netinet/ip.h> 59#include <netinet/in_pcb.h> 60#include <netinet/in_var.h> 61#include <netinet/ip_var.h> 62#include <netinet/ip_icmp.h> 63 64#ifndef IPFORWARDING 65#ifdef GATEWAY 66#define IPFORWARDING 1 /* forward IP packets not for us */ 67#else /* GATEWAY */ 68#define IPFORWARDING 0 /* don't forward IP packets not for us */ 69#endif /* GATEWAY */ 70#endif /* IPFORWARDING */ 71#ifndef IPSENDREDIRECTS 72#define IPSENDREDIRECTS 1 73#endif 74#ifndef IPFORWSRCRT 75#define IPFORWSRCRT 1 /* allow source-routed packets */ 76#endif 77/* 78 * Note: DIRECTED_BROADCAST is handled this way so that previous 79 * configuration using this option will Just Work. 80 */ 81#ifndef IPDIRECTEDBCAST 82#ifdef DIRECTED_BROADCAST 83#define IPDIRECTEDBCAST 1 84#else 85#define IPDIRECTEDBCAST 0 86#endif /* DIRECTED_BROADCAST */ 87#endif /* IPDIRECTEDBCAST */ 88int ipforwarding = IPFORWARDING; 89int ipsendredirects = IPSENDREDIRECTS; 90int ip_defttl = IPDEFTTL; 91int ip_forwsrcrt = IPFORWSRCRT; 92int ip_directedbcast = IPDIRECTEDBCAST; 93#ifdef DIAGNOSTIC 94int ipprintfs = 0; 95#endif 96 97extern struct domain inetdomain; 98extern struct protosw inetsw[]; 99u_char ip_protox[IPPROTO_MAX]; 100int ipqmaxlen = IFQ_MAXLEN; 101struct in_ifaddrhead in_ifaddr; 102struct ifqueue ipintrq; 103 104/* 105 * We need to save the IP options in case a protocol wants to respond 106 * to an incoming packet over the same route if the packet got here 107 * using IP source routing. This allows connection establishment and 108 * maintenance when the remote end is on a network that is not known 109 * to us. 110 */ 111int ip_nhops = 0; 112static struct ip_srcrt { 113 struct in_addr dst; /* final destination */ 114 char nop; /* one NOP to align */ 115 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 116 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 117} ip_srcrt; 118 119static void save_rte __P((u_char *, struct in_addr)); 120 121/* 122 * IP initialization: fill in IP protocol switch table. 123 * All protocols not implemented in kernel go to raw IP protocol handler. 124 */ 125void 126ip_init() 127{ 128 register struct protosw *pr; 129 register int i; 130 131 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 132 if (pr == 0) 133 panic("ip_init"); 134 for (i = 0; i < IPPROTO_MAX; i++) 135 ip_protox[i] = pr - inetsw; 136 for (pr = inetdomain.dom_protosw; 137 pr < inetdomain.dom_protoswNPROTOSW; pr++) 138 if (pr->pr_domain->dom_family == PF_INET && 139 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 140 ip_protox[pr->pr_protocol] = pr - inetsw; 141 LIST_INIT(&ipq); 142 ip_id = time.tv_sec & 0xffff; 143 ipintrq.ifq_maxlen = ipqmaxlen; 144 TAILQ_INIT(&in_ifaddr); 145} 146 147struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 148struct route ipforward_rt; 149 150/* 151 * Ip input routine. Checksum and byte swap header. If fragmented 152 * try to reassemble. Process options. Pass to next level. 153 */ 154void 155ipintr() 156{ 157 register struct ip *ip = NULL; 158 register struct mbuf *m; 159 register struct ipq *fp; 160 register struct in_ifaddr *ia; 161 struct ipqent *ipqe; 162 int hlen = 0, mff, len, s; 163#ifdef PACKET_FILTER 164 struct packet_filter_hook *pfh; 165 struct mbuf *m0; 166#endif /* PACKET_FILTER */ 167 168next: 169 /* 170 * Get next datagram off input queue and get IP header 171 * in first mbuf. 172 */ 173 s = splimp(); 174 IF_DEQUEUE(&ipintrq, m); 175 splx(s); 176 if (m == 0) 177 return; 178#ifdef DIAGNOSTIC 179 if ((m->m_flags & M_PKTHDR) == 0) 180 panic("ipintr no HDR"); 181#endif 182 /* 183 * If no IP addresses have been set yet but the interfaces 184 * are receiving, can't do anything with incoming packets yet. 185 */ 186 if (in_ifaddr.tqh_first == 0) 187 goto bad; 188 ipstat.ips_total++; 189 if (m->m_len < sizeof (struct ip) && 190 (m = m_pullup(m, sizeof (struct ip))) == 0) { 191 ipstat.ips_toosmall++; 192 goto next; 193 } 194 ip = mtod(m, struct ip *); 195 if (ip->ip_v != IPVERSION) { 196 ipstat.ips_badvers++; 197 goto bad; 198 } 199 hlen = ip->ip_hl << 2; 200 if (hlen < sizeof(struct ip)) { /* minimum header length */ 201 ipstat.ips_badhlen++; 202 goto bad; 203 } 204 if (hlen > m->m_len) { 205 if ((m = m_pullup(m, hlen)) == 0) { 206 ipstat.ips_badhlen++; 207 goto next; 208 } 209 ip = mtod(m, struct ip *); 210 } 211 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 212 ipstat.ips_badsum++; 213 goto bad; 214 } 215 216 /* 217 * Convert fields to host representation. 218 */ 219 NTOHS(ip->ip_len); 220 NTOHS(ip->ip_id); 221 NTOHS(ip->ip_off); 222 len = ip->ip_len; 223 224 /* 225 * Check that the amount of data in the buffers 226 * is as at least much as the IP header would have us expect. 227 * Trim mbufs if longer than we expect. 228 * Drop packet if shorter than we expect. 229 */ 230 if (m->m_pkthdr.len < len) { 231 ipstat.ips_tooshort++; 232 goto bad; 233 } 234 if (m->m_pkthdr.len > len) { 235 if (m->m_len == m->m_pkthdr.len) { 236 m->m_len = len; 237 m->m_pkthdr.len = len; 238 } else 239 m_adj(m, len - m->m_pkthdr.len); 240 } 241 242#ifdef PACKET_FILTER 243 /* 244 * Run through list of hooks for input packets. 245 */ 246 m0 = m; 247 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 248 if (pfh->pfil_func) { 249 if (pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0)) 250 goto bad; 251 ip = mtod(m = m0, struct ip *); 252 } 253#endif /* PACKET_FILTER */ 254 255 /* 256 * Process options and, if not destined for us, 257 * ship it on. ip_dooptions returns 1 when an 258 * error was detected (causing an icmp message 259 * to be sent and the original packet to be freed). 260 */ 261 ip_nhops = 0; /* for source routed packets */ 262 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 263 goto next; 264 265 /* 266 * Check our list of addresses, to see if the packet is for us. 267 */ 268 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 269 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 270 goto ours; 271 if (((ip_directedbcast == 0) || (ip_directedbcast && 272 ia->ia_ifp == m->m_pkthdr.rcvif)) && 273 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 274 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 275 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 276 /* 277 * Look for all-0's host part (old broadcast addr), 278 * either for subnet or net. 279 */ 280 ip->ip_dst.s_addr == ia->ia_subnet || 281 ip->ip_dst.s_addr == ia->ia_net) 282 goto ours; 283 } 284 } 285 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 286 struct in_multi *inm; 287#ifdef MROUTING 288 extern struct socket *ip_mrouter; 289 290 if (m->m_flags & M_EXT) { 291 if ((m = m_pullup(m, hlen)) == 0) { 292 ipstat.ips_toosmall++; 293 goto next; 294 } 295 ip = mtod(m, struct ip *); 296 } 297 298 if (ip_mrouter) { 299 /* 300 * If we are acting as a multicast router, all 301 * incoming multicast packets are passed to the 302 * kernel-level multicast forwarding function. 303 * The packet is returned (relatively) intact; if 304 * ip_mforward() returns a non-zero value, the packet 305 * must be discarded, else it may be accepted below. 306 * 307 * (The IP ident field is put in the same byte order 308 * as expected when ip_mforward() is called from 309 * ip_output().) 310 */ 311 ip->ip_id = htons(ip->ip_id); 312 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 313 ipstat.ips_cantforward++; 314 m_freem(m); 315 goto next; 316 } 317 ip->ip_id = ntohs(ip->ip_id); 318 319 /* 320 * The process-level routing demon needs to receive 321 * all multicast IGMP packets, whether or not this 322 * host belongs to their destination groups. 323 */ 324 if (ip->ip_p == IPPROTO_IGMP) 325 goto ours; 326 ipstat.ips_forward++; 327 } 328#endif 329 /* 330 * See if we belong to the destination multicast group on the 331 * arrival interface. 332 */ 333 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 334 if (inm == NULL) { 335 ipstat.ips_cantforward++; 336 m_freem(m); 337 goto next; 338 } 339 goto ours; 340 } 341 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 342 in_nullhost(ip->ip_dst)) 343 goto ours; 344 345 /* 346 * Not for us; forward if possible and desirable. 347 */ 348 if (ipforwarding == 0) { 349 ipstat.ips_cantforward++; 350 m_freem(m); 351 } else 352 ip_forward(m, 0); 353 goto next; 354 355ours: 356 /* 357 * If offset or IP_MF are set, must reassemble. 358 * Otherwise, nothing need be done. 359 * (We could look in the reassembly queue to see 360 * if the packet was previously fragmented, 361 * but it's not worth the time; just let them time out.) 362 */ 363 if (ip->ip_off &~ IP_DF) { 364 if (m->m_flags & M_EXT) { /* XXX */ 365 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 366 ipstat.ips_toosmall++; 367 goto next; 368 } 369 ip = mtod(m, struct ip *); 370 } 371 /* 372 * Look for queue of fragments 373 * of this datagram. 374 */ 375 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 376 if (ip->ip_id == fp->ipq_id && 377 in_hosteq(ip->ip_src, fp->ipq_src) && 378 in_hosteq(ip->ip_dst, fp->ipq_dst) && 379 ip->ip_p == fp->ipq_p) 380 goto found; 381 fp = 0; 382found: 383 384 /* 385 * Adjust ip_len to not reflect header, 386 * set ipqe_mff if more fragments are expected, 387 * convert offset of this to bytes. 388 */ 389 ip->ip_len -= hlen; 390 mff = (ip->ip_off & IP_MF) != 0; 391 if (mff) { 392 /* 393 * Make sure that fragments have a data length 394 * that's a non-zero multiple of 8 bytes. 395 */ 396 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 397 ipstat.ips_badfrags++; 398 goto bad; 399 } 400 } 401 ip->ip_off <<= 3; 402 403 /* 404 * If datagram marked as having more fragments 405 * or if this is not the first fragment, 406 * attempt reassembly; if it succeeds, proceed. 407 */ 408 if (mff || ip->ip_off) { 409 ipstat.ips_fragments++; 410 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 411 M_IPQ, M_NOWAIT); 412 if (ipqe == NULL) { 413 ipstat.ips_rcvmemdrop++; 414 goto bad; 415 } 416 ipqe->ipqe_mff = mff; 417 ipqe->ipqe_ip = ip; 418 ip = ip_reass(ipqe, fp); 419 if (ip == 0) 420 goto next; 421 ipstat.ips_reassembled++; 422 m = dtom(ip); 423 } else 424 if (fp) 425 ip_freef(fp); 426 } else 427 ip->ip_len -= hlen; 428 429 /* 430 * Switch out to protocol's input routine. 431 */ 432 ipstat.ips_delivered++; 433 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 434 goto next; 435bad: 436#ifdef PACKET_FILTER 437 m0 = m; 438 for (pfh = pfil_hook_get(PFIL_BAD); pfh; pfh = pfh->pfil_link.le_next) 439 if (pfh->pfil_func) { 440 (void)pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 2, &m0); 441 ip = mtod(m = m0, struct ip *); 442 } 443#endif /* PACKET_FILTER */ 444 m_freem(m); 445 goto next; 446} 447 448/* 449 * Take incoming datagram fragment and try to 450 * reassemble it into whole datagram. If a chain for 451 * reassembly of this datagram already exists, then it 452 * is given as fp; otherwise have to make a chain. 453 */ 454struct ip * 455ip_reass(ipqe, fp) 456 register struct ipqent *ipqe; 457 register struct ipq *fp; 458{ 459 register struct mbuf *m = dtom(ipqe->ipqe_ip); 460 register struct ipqent *nq, *p, *q; 461 struct ip *ip; 462 struct mbuf *t; 463 int hlen = ipqe->ipqe_ip->ip_hl << 2; 464 int i, next; 465 466 /* 467 * Presence of header sizes in mbufs 468 * would confuse code below. 469 */ 470 m->m_data += hlen; 471 m->m_len -= hlen; 472 473 /* 474 * If first fragment to arrive, create a reassembly queue. 475 */ 476 if (fp == 0) { 477 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 478 goto dropfrag; 479 fp = mtod(t, struct ipq *); 480 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 481 fp->ipq_ttl = IPFRAGTTL; 482 fp->ipq_p = ipqe->ipqe_ip->ip_p; 483 fp->ipq_id = ipqe->ipqe_ip->ip_id; 484 LIST_INIT(&fp->ipq_fragq); 485 fp->ipq_src = ipqe->ipqe_ip->ip_src; 486 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 487 p = NULL; 488 goto insert; 489 } 490 491 /* 492 * Find a segment which begins after this one does. 493 */ 494 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 495 p = q, q = q->ipqe_q.le_next) 496 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 497 break; 498 499 /* 500 * If there is a preceding segment, it may provide some of 501 * our data already. If so, drop the data from the incoming 502 * segment. If it provides all of our data, drop us. 503 */ 504 if (p != NULL) { 505 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 506 ipqe->ipqe_ip->ip_off; 507 if (i > 0) { 508 if (i >= ipqe->ipqe_ip->ip_len) 509 goto dropfrag; 510 m_adj(dtom(ipqe->ipqe_ip), i); 511 ipqe->ipqe_ip->ip_off += i; 512 ipqe->ipqe_ip->ip_len -= i; 513 } 514 } 515 516 /* 517 * While we overlap succeeding segments trim them or, 518 * if they are completely covered, dequeue them. 519 */ 520 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 521 q->ipqe_ip->ip_off; q = nq) { 522 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 523 q->ipqe_ip->ip_off; 524 if (i < q->ipqe_ip->ip_len) { 525 q->ipqe_ip->ip_len -= i; 526 q->ipqe_ip->ip_off += i; 527 m_adj(dtom(q->ipqe_ip), i); 528 break; 529 } 530 nq = q->ipqe_q.le_next; 531 m_freem(dtom(q->ipqe_ip)); 532 LIST_REMOVE(q, ipqe_q); 533 FREE(q, M_IPQ); 534 } 535 536insert: 537 /* 538 * Stick new segment in its place; 539 * check for complete reassembly. 540 */ 541 if (p == NULL) { 542 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 543 } else { 544 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 545 } 546 next = 0; 547 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 548 p = q, q = q->ipqe_q.le_next) { 549 if (q->ipqe_ip->ip_off != next) 550 return (0); 551 next += q->ipqe_ip->ip_len; 552 } 553 if (p->ipqe_mff) 554 return (0); 555 556 /* 557 * Reassembly is complete; concatenate fragments. 558 */ 559 q = fp->ipq_fragq.lh_first; 560 ip = q->ipqe_ip; 561 m = dtom(q->ipqe_ip); 562 t = m->m_next; 563 m->m_next = 0; 564 m_cat(m, t); 565 nq = q->ipqe_q.le_next; 566 FREE(q, M_IPQ); 567 for (q = nq; q != NULL; q = nq) { 568 t = dtom(q->ipqe_ip); 569 nq = q->ipqe_q.le_next; 570 FREE(q, M_IPQ); 571 m_cat(m, t); 572 } 573 574 /* 575 * Create header for new ip packet by 576 * modifying header of first packet; 577 * dequeue and discard fragment reassembly header. 578 * Make header visible. 579 */ 580 ip->ip_len = next; 581 ip->ip_src = fp->ipq_src; 582 ip->ip_dst = fp->ipq_dst; 583 LIST_REMOVE(fp, ipq_q); 584 (void) m_free(dtom(fp)); 585 m->m_len += (ip->ip_hl << 2); 586 m->m_data -= (ip->ip_hl << 2); 587 /* some debugging cruft by sklower, below, will go away soon */ 588 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 589 register int plen = 0; 590 for (t = m; m; m = m->m_next) 591 plen += m->m_len; 592 t->m_pkthdr.len = plen; 593 } 594 return (ip); 595 596dropfrag: 597 ipstat.ips_fragdropped++; 598 m_freem(m); 599 FREE(ipqe, M_IPQ); 600 return (0); 601} 602 603/* 604 * Free a fragment reassembly header and all 605 * associated datagrams. 606 */ 607void 608ip_freef(fp) 609 struct ipq *fp; 610{ 611 register struct ipqent *q, *p; 612 613 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 614 p = q->ipqe_q.le_next; 615 m_freem(dtom(q->ipqe_ip)); 616 LIST_REMOVE(q, ipqe_q); 617 FREE(q, M_IPQ); 618 } 619 LIST_REMOVE(fp, ipq_q); 620 (void) m_free(dtom(fp)); 621} 622 623/* 624 * IP timer processing; 625 * if a timer expires on a reassembly 626 * queue, discard it. 627 */ 628void 629ip_slowtimo() 630{ 631 register struct ipq *fp, *nfp; 632 int s = splsoftnet(); 633 634 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 635 nfp = fp->ipq_q.le_next; 636 if (--fp->ipq_ttl == 0) { 637 ipstat.ips_fragtimeout++; 638 ip_freef(fp); 639 } 640 } 641 splx(s); 642} 643 644/* 645 * Drain off all datagram fragments. 646 */ 647void 648ip_drain() 649{ 650 651 while (ipq.lh_first != NULL) { 652 ipstat.ips_fragdropped++; 653 ip_freef(ipq.lh_first); 654 } 655} 656 657/* 658 * Do option processing on a datagram, 659 * possibly discarding it if bad options are encountered, 660 * or forwarding it if source-routed. 661 * Returns 1 if packet has been forwarded/freed, 662 * 0 if the packet should be processed further. 663 */ 664int 665ip_dooptions(m) 666 struct mbuf *m; 667{ 668 register struct ip *ip = mtod(m, struct ip *); 669 register u_char *cp; 670 register struct ip_timestamp *ipt; 671 register struct in_ifaddr *ia; 672 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 673 struct in_addr *sin, dst; 674 n_time ntime; 675 676 dst = ip->ip_dst; 677 cp = (u_char *)(ip + 1); 678 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 679 for (; cnt > 0; cnt -= optlen, cp += optlen) { 680 opt = cp[IPOPT_OPTVAL]; 681 if (opt == IPOPT_EOL) 682 break; 683 if (opt == IPOPT_NOP) 684 optlen = 1; 685 else { 686 optlen = cp[IPOPT_OLEN]; 687 if (optlen <= 0 || optlen > cnt) { 688 code = &cp[IPOPT_OLEN] - (u_char *)ip; 689 goto bad; 690 } 691 } 692 switch (opt) { 693 694 default: 695 break; 696 697 /* 698 * Source routing with record. 699 * Find interface with current destination address. 700 * If none on this machine then drop if strictly routed, 701 * or do nothing if loosely routed. 702 * Record interface address and bring up next address 703 * component. If strictly routed make sure next 704 * address is on directly accessible net. 705 */ 706 case IPOPT_LSRR: 707 case IPOPT_SSRR: 708 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 709 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 710 goto bad; 711 } 712 ipaddr.sin_addr = ip->ip_dst; 713 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 714 if (ia == 0) { 715 if (opt == IPOPT_SSRR) { 716 type = ICMP_UNREACH; 717 code = ICMP_UNREACH_SRCFAIL; 718 goto bad; 719 } 720 /* 721 * Loose routing, and not at next destination 722 * yet; nothing to do except forward. 723 */ 724 break; 725 } 726 off--; /* 0 origin */ 727 if (off > optlen - sizeof(struct in_addr)) { 728 /* 729 * End of source route. Should be for us. 730 */ 731 save_rte(cp, ip->ip_src); 732 break; 733 } 734 /* 735 * locate outgoing interface 736 */ 737 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 738 sizeof(ipaddr.sin_addr)); 739 if (opt == IPOPT_SSRR) { 740#define INA struct in_ifaddr * 741#define SA struct sockaddr * 742 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 743 } else 744 ia = ip_rtaddr(ipaddr.sin_addr); 745 if (ia == 0) { 746 type = ICMP_UNREACH; 747 code = ICMP_UNREACH_SRCFAIL; 748 goto bad; 749 } 750 ip->ip_dst = ipaddr.sin_addr; 751 bcopy((caddr_t)&ia->ia_addr.sin_addr, 752 (caddr_t)(cp + off), sizeof(struct in_addr)); 753 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 754 /* 755 * Let ip_intr's mcast routing check handle mcast pkts 756 */ 757 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 758 break; 759 760 case IPOPT_RR: 761 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 762 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 763 goto bad; 764 } 765 /* 766 * If no space remains, ignore. 767 */ 768 off--; /* 0 origin */ 769 if (off > optlen - sizeof(struct in_addr)) 770 break; 771 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 772 sizeof(ipaddr.sin_addr)); 773 /* 774 * locate outgoing interface; if we're the destination, 775 * use the incoming interface (should be same). 776 */ 777 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 778 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 779 type = ICMP_UNREACH; 780 code = ICMP_UNREACH_HOST; 781 goto bad; 782 } 783 bcopy((caddr_t)&ia->ia_addr.sin_addr, 784 (caddr_t)(cp + off), sizeof(struct in_addr)); 785 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 786 break; 787 788 case IPOPT_TS: 789 code = cp - (u_char *)ip; 790 ipt = (struct ip_timestamp *)cp; 791 if (ipt->ipt_len < 5) 792 goto bad; 793 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 794 if (++ipt->ipt_oflw == 0) 795 goto bad; 796 break; 797 } 798 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 799 switch (ipt->ipt_flg) { 800 801 case IPOPT_TS_TSONLY: 802 break; 803 804 case IPOPT_TS_TSANDADDR: 805 if (ipt->ipt_ptr + sizeof(n_time) + 806 sizeof(struct in_addr) > ipt->ipt_len) 807 goto bad; 808 ipaddr.sin_addr = dst; 809 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 810 m->m_pkthdr.rcvif); 811 if (ia == 0) 812 continue; 813 bcopy((caddr_t)&ia->ia_addr.sin_addr, 814 (caddr_t)sin, sizeof(struct in_addr)); 815 ipt->ipt_ptr += sizeof(struct in_addr); 816 break; 817 818 case IPOPT_TS_PRESPEC: 819 if (ipt->ipt_ptr + sizeof(n_time) + 820 sizeof(struct in_addr) > ipt->ipt_len) 821 goto bad; 822 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 823 sizeof(struct in_addr)); 824 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 825 continue; 826 ipt->ipt_ptr += sizeof(struct in_addr); 827 break; 828 829 default: 830 goto bad; 831 } 832 ntime = iptime(); 833 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 834 sizeof(n_time)); 835 ipt->ipt_ptr += sizeof(n_time); 836 } 837 } 838 if (forward) { 839 if (ip_forwsrcrt == 0) { 840 type = ICMP_UNREACH; 841 code = ICMP_UNREACH_SRCFAIL; 842 goto bad; 843 } 844 ip_forward(m, 1); 845 return (1); 846 } 847 return (0); 848bad: 849 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 850 icmp_error(m, type, code, 0, 0); 851 ipstat.ips_badoptions++; 852 return (1); 853} 854 855/* 856 * Given address of next destination (final or next hop), 857 * return internet address info of interface to be used to get there. 858 */ 859struct in_ifaddr * 860ip_rtaddr(dst) 861 struct in_addr dst; 862{ 863 register struct sockaddr_in *sin; 864 865 sin = satosin(&ipforward_rt.ro_dst); 866 867 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 868 if (ipforward_rt.ro_rt) { 869 RTFREE(ipforward_rt.ro_rt); 870 ipforward_rt.ro_rt = 0; 871 } 872 sin->sin_family = AF_INET; 873 sin->sin_len = sizeof(*sin); 874 sin->sin_addr = dst; 875 876 rtalloc(&ipforward_rt); 877 } 878 if (ipforward_rt.ro_rt == 0) 879 return ((struct in_ifaddr *)0); 880 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 881} 882 883/* 884 * Save incoming source route for use in replies, 885 * to be picked up later by ip_srcroute if the receiver is interested. 886 */ 887void 888save_rte(option, dst) 889 u_char *option; 890 struct in_addr dst; 891{ 892 unsigned olen; 893 894 olen = option[IPOPT_OLEN]; 895#ifdef DIAGNOSTIC 896 if (ipprintfs) 897 printf("save_rte: olen %d\n", olen); 898#endif 899 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 900 return; 901 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 902 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 903 ip_srcrt.dst = dst; 904} 905 906/* 907 * Retrieve incoming source route for use in replies, 908 * in the same form used by setsockopt. 909 * The first hop is placed before the options, will be removed later. 910 */ 911struct mbuf * 912ip_srcroute() 913{ 914 register struct in_addr *p, *q; 915 register struct mbuf *m; 916 917 if (ip_nhops == 0) 918 return ((struct mbuf *)0); 919 m = m_get(M_DONTWAIT, MT_SOOPTS); 920 if (m == 0) 921 return ((struct mbuf *)0); 922 923#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 924 925 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 926 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 927 OPTSIZ; 928#ifdef DIAGNOSTIC 929 if (ipprintfs) 930 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 931#endif 932 933 /* 934 * First save first hop for return route 935 */ 936 p = &ip_srcrt.route[ip_nhops - 1]; 937 *(mtod(m, struct in_addr *)) = *p--; 938#ifdef DIAGNOSTIC 939 if (ipprintfs) 940 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 941#endif 942 943 /* 944 * Copy option fields and padding (nop) to mbuf. 945 */ 946 ip_srcrt.nop = IPOPT_NOP; 947 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 948 bcopy((caddr_t)&ip_srcrt.nop, 949 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 950 q = (struct in_addr *)(mtod(m, caddr_t) + 951 sizeof(struct in_addr) + OPTSIZ); 952#undef OPTSIZ 953 /* 954 * Record return path as an IP source route, 955 * reversing the path (pointers are now aligned). 956 */ 957 while (p >= ip_srcrt.route) { 958#ifdef DIAGNOSTIC 959 if (ipprintfs) 960 printf(" %x", ntohl(q->s_addr)); 961#endif 962 *q++ = *p--; 963 } 964 /* 965 * Last hop goes to final destination. 966 */ 967 *q = ip_srcrt.dst; 968#ifdef DIAGNOSTIC 969 if (ipprintfs) 970 printf(" %x\n", ntohl(q->s_addr)); 971#endif 972 return (m); 973} 974 975/* 976 * Strip out IP options, at higher 977 * level protocol in the kernel. 978 * Second argument is buffer to which options 979 * will be moved, and return value is their length. 980 * XXX should be deleted; last arg currently ignored. 981 */ 982void 983ip_stripoptions(m, mopt) 984 register struct mbuf *m; 985 struct mbuf *mopt; 986{ 987 register int i; 988 struct ip *ip = mtod(m, struct ip *); 989 register caddr_t opts; 990 int olen; 991 992 olen = (ip->ip_hl<<2) - sizeof (struct ip); 993 opts = (caddr_t)(ip + 1); 994 i = m->m_len - (sizeof (struct ip) + olen); 995 bcopy(opts + olen, opts, (unsigned)i); 996 m->m_len -= olen; 997 if (m->m_flags & M_PKTHDR) 998 m->m_pkthdr.len -= olen; 999 ip->ip_hl = sizeof(struct ip) >> 2; 1000} 1001 1002int inetctlerrmap[PRC_NCMDS] = { 1003 0, 0, 0, 0, 1004 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1005 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1006 EMSGSIZE, EHOSTUNREACH, 0, 0, 1007 0, 0, 0, 0, 1008 ENOPROTOOPT 1009}; 1010 1011/* 1012 * Forward a packet. If some error occurs return the sender 1013 * an icmp packet. Note we can't always generate a meaningful 1014 * icmp message because icmp doesn't have a large enough repertoire 1015 * of codes and types. 1016 * 1017 * If not forwarding, just drop the packet. This could be confusing 1018 * if ipforwarding was zero but some routing protocol was advancing 1019 * us as a gateway to somewhere. However, we must let the routing 1020 * protocol deal with that. 1021 * 1022 * The srcrt parameter indicates whether the packet is being forwarded 1023 * via a source route. 1024 */ 1025void 1026ip_forward(m, srcrt) 1027 struct mbuf *m; 1028 int srcrt; 1029{ 1030 register struct ip *ip = mtod(m, struct ip *); 1031 register struct sockaddr_in *sin; 1032 register struct rtentry *rt; 1033 int error, type = 0, code = 0; 1034 struct mbuf *mcopy; 1035 n_long dest; 1036 struct ifnet *destifp; 1037 1038 dest = 0; 1039#ifdef DIAGNOSTIC 1040 if (ipprintfs) 1041 printf("forward: src %x dst %x ttl %x\n", 1042 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1043#endif 1044 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1045 ipstat.ips_cantforward++; 1046 m_freem(m); 1047 return; 1048 } 1049 HTONS(ip->ip_id); 1050 if (ip->ip_ttl <= IPTTLDEC) { 1051 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1052 return; 1053 } 1054 ip->ip_ttl -= IPTTLDEC; 1055 1056 sin = satosin(&ipforward_rt.ro_dst); 1057 if ((rt = ipforward_rt.ro_rt) == 0 || 1058 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1059 if (ipforward_rt.ro_rt) { 1060 RTFREE(ipforward_rt.ro_rt); 1061 ipforward_rt.ro_rt = 0; 1062 } 1063 sin->sin_family = AF_INET; 1064 sin->sin_len = sizeof(struct sockaddr_in); 1065 sin->sin_addr = ip->ip_dst; 1066 1067 rtalloc(&ipforward_rt); 1068 if (ipforward_rt.ro_rt == 0) { 1069 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1070 return; 1071 } 1072 rt = ipforward_rt.ro_rt; 1073 } 1074 1075 /* 1076 * Save at most 68 bytes of the packet in case 1077 * we need to generate an ICMP message to the src. 1078 */ 1079 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1080 1081 /* 1082 * If forwarding packet using same interface that it came in on, 1083 * perhaps should send a redirect to sender to shortcut a hop. 1084 * Only send redirect if source is sending directly to us, 1085 * and if packet was not source routed (or has any options). 1086 * Also, don't send redirect if forwarding using a default route 1087 * or a route modified by a redirect. 1088 */ 1089 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1090 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1091 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1092 ipsendredirects && !srcrt) { 1093 if (rt->rt_ifa && 1094 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1095 ifatoia(rt->rt_ifa)->ia_subnet) { 1096 if (rt->rt_flags & RTF_GATEWAY) 1097 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1098 else 1099 dest = ip->ip_dst.s_addr; 1100 /* Router requirements says to only send host redirects */ 1101 type = ICMP_REDIRECT; 1102 code = ICMP_REDIRECT_HOST; 1103#ifdef DIAGNOSTIC 1104 if (ipprintfs) 1105 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1106#endif 1107 } 1108 } 1109 1110 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1111 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1112 if (error) 1113 ipstat.ips_cantforward++; 1114 else { 1115 ipstat.ips_forward++; 1116 if (type) 1117 ipstat.ips_redirectsent++; 1118 else { 1119 if (mcopy) 1120 m_freem(mcopy); 1121 return; 1122 } 1123 } 1124 if (mcopy == NULL) 1125 return; 1126 destifp = NULL; 1127 1128 switch (error) { 1129 1130 case 0: /* forwarded, but need redirect */ 1131 /* type, code set above */ 1132 break; 1133 1134 case ENETUNREACH: /* shouldn't happen, checked above */ 1135 case EHOSTUNREACH: 1136 case ENETDOWN: 1137 case EHOSTDOWN: 1138 default: 1139 type = ICMP_UNREACH; 1140 code = ICMP_UNREACH_HOST; 1141 break; 1142 1143 case EMSGSIZE: 1144 type = ICMP_UNREACH; 1145 code = ICMP_UNREACH_NEEDFRAG; 1146 if (ipforward_rt.ro_rt) 1147 destifp = ipforward_rt.ro_rt->rt_ifp; 1148 ipstat.ips_cantfrag++; 1149 break; 1150 1151 case ENOBUFS: 1152 type = ICMP_SOURCEQUENCH; 1153 code = 0; 1154 break; 1155 } 1156 icmp_error(mcopy, type, code, dest, destifp); 1157} 1158 1159int 1160ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1161 int *name; 1162 u_int namelen; 1163 void *oldp; 1164 size_t *oldlenp; 1165 void *newp; 1166 size_t newlen; 1167{ 1168 /* All sysctl names at this level are terminal. */ 1169 if (namelen != 1) 1170 return (ENOTDIR); 1171 1172 switch (name[0]) { 1173 case IPCTL_FORWARDING: 1174 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1175 case IPCTL_SENDREDIRECTS: 1176 return (sysctl_int(oldp, oldlenp, newp, newlen, 1177 &ipsendredirects)); 1178 case IPCTL_DEFTTL: 1179 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1180#ifdef notyet 1181 case IPCTL_DEFMTU: 1182 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1183#endif 1184 case IPCTL_FORWSRCRT: 1185 /* 1186 * Don't allow this to change in a secure environment. 1187 */ 1188 if (securelevel > 0) 1189 return (EPERM); 1190 return (sysctl_int(oldp, oldlenp, newp, newlen, 1191 &ip_forwsrcrt)); 1192 case IPCTL_DIRECTEDBCAST: 1193 return (sysctl_int(oldp, oldlenp, newp, newlen, 1194 &ip_directedbcast)); 1195 default: 1196 return (EOPNOTSUPP); 1197 } 1198 /* NOTREACHED */ 1199} 1200