ip_input.c revision 1.42
1/* $NetBSD: ip_input.c,v 1.42 1996/12/20 08:39:27 mrg 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#ifdef PFIL_HOOKS 65#include <net/pfil.h> 66#endif /* PFIL_HOOKS */ 67 68#ifndef IPFORWARDING 69#ifdef GATEWAY 70#define IPFORWARDING 1 /* forward IP packets not for us */ 71#else /* GATEWAY */ 72#define IPFORWARDING 0 /* don't forward IP packets not for us */ 73#endif /* GATEWAY */ 74#endif /* IPFORWARDING */ 75#ifndef IPSENDREDIRECTS 76#define IPSENDREDIRECTS 1 77#endif 78#ifndef IPFORWSRCRT 79#define IPFORWSRCRT 1 /* allow source-routed packets */ 80#endif 81/* 82 * Note: DIRECTED_BROADCAST is handled this way so that previous 83 * configuration using this option will Just Work. 84 */ 85#ifndef IPDIRECTEDBCAST 86#ifdef DIRECTED_BROADCAST 87#define IPDIRECTEDBCAST 1 88#else 89#define IPDIRECTEDBCAST 0 90#endif /* DIRECTED_BROADCAST */ 91#endif /* IPDIRECTEDBCAST */ 92int ipforwarding = IPFORWARDING; 93int ipsendredirects = IPSENDREDIRECTS; 94int ip_defttl = IPDEFTTL; 95int ip_forwsrcrt = IPFORWSRCRT; 96int ip_directedbcast = IPDIRECTEDBCAST; 97#ifdef DIAGNOSTIC 98int ipprintfs = 0; 99#endif 100 101extern struct domain inetdomain; 102extern struct protosw inetsw[]; 103u_char ip_protox[IPPROTO_MAX]; 104int ipqmaxlen = IFQ_MAXLEN; 105struct in_ifaddrhead in_ifaddr; 106struct ifqueue ipintrq; 107 108/* 109 * We need to save the IP options in case a protocol wants to respond 110 * to an incoming packet over the same route if the packet got here 111 * using IP source routing. This allows connection establishment and 112 * maintenance when the remote end is on a network that is not known 113 * to us. 114 */ 115int ip_nhops = 0; 116static struct ip_srcrt { 117 struct in_addr dst; /* final destination */ 118 char nop; /* one NOP to align */ 119 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 120 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 121} ip_srcrt; 122 123static void save_rte __P((u_char *, struct in_addr)); 124 125/* 126 * IP initialization: fill in IP protocol switch table. 127 * All protocols not implemented in kernel go to raw IP protocol handler. 128 */ 129void 130ip_init() 131{ 132 register struct protosw *pr; 133 register int i; 134 135 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 136 if (pr == 0) 137 panic("ip_init"); 138 for (i = 0; i < IPPROTO_MAX; i++) 139 ip_protox[i] = pr - inetsw; 140 for (pr = inetdomain.dom_protosw; 141 pr < inetdomain.dom_protoswNPROTOSW; pr++) 142 if (pr->pr_domain->dom_family == PF_INET && 143 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 144 ip_protox[pr->pr_protocol] = pr - inetsw; 145 LIST_INIT(&ipq); 146 ip_id = time.tv_sec & 0xffff; 147 ipintrq.ifq_maxlen = ipqmaxlen; 148 TAILQ_INIT(&in_ifaddr); 149} 150 151struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 152struct route ipforward_rt; 153 154/* 155 * Ip input routine. Checksum and byte swap header. If fragmented 156 * try to reassemble. Process options. Pass to next level. 157 */ 158void 159ipintr() 160{ 161 register struct ip *ip = NULL; 162 register struct mbuf *m; 163 register struct ipq *fp; 164 register struct in_ifaddr *ia; 165 struct ipqent *ipqe; 166 int hlen = 0, mff, len, s; 167#ifdef PFIL_HOOKS 168 struct packet_filter_hook *pfh; 169 struct mbuf *m0; 170#endif /* PFIL_HOOKS */ 171 172next: 173 /* 174 * Get next datagram off input queue and get IP header 175 * in first mbuf. 176 */ 177 s = splimp(); 178 IF_DEQUEUE(&ipintrq, m); 179 splx(s); 180 if (m == 0) 181 return; 182#ifdef DIAGNOSTIC 183 if ((m->m_flags & M_PKTHDR) == 0) 184 panic("ipintr no HDR"); 185#endif 186 /* 187 * If no IP addresses have been set yet but the interfaces 188 * are receiving, can't do anything with incoming packets yet. 189 */ 190 if (in_ifaddr.tqh_first == 0) 191 goto bad; 192 ipstat.ips_total++; 193 if (m->m_len < sizeof (struct ip) && 194 (m = m_pullup(m, sizeof (struct ip))) == 0) { 195 ipstat.ips_toosmall++; 196 goto next; 197 } 198 ip = mtod(m, struct ip *); 199 if (ip->ip_v != IPVERSION) { 200 ipstat.ips_badvers++; 201 goto bad; 202 } 203 hlen = ip->ip_hl << 2; 204 if (hlen < sizeof(struct ip)) { /* minimum header length */ 205 ipstat.ips_badhlen++; 206 goto bad; 207 } 208 if (hlen > m->m_len) { 209 if ((m = m_pullup(m, hlen)) == 0) { 210 ipstat.ips_badhlen++; 211 goto next; 212 } 213 ip = mtod(m, struct ip *); 214 } 215 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 216 ipstat.ips_badsum++; 217 goto bad; 218 } 219 220 /* 221 * Convert fields to host representation. 222 */ 223 NTOHS(ip->ip_len); 224 NTOHS(ip->ip_id); 225 NTOHS(ip->ip_off); 226 len = ip->ip_len; 227 228 /* 229 * Check that the amount of data in the buffers 230 * is as at least much as the IP header would have us expect. 231 * Trim mbufs if longer than we expect. 232 * Drop packet if shorter than we expect. 233 */ 234 if (m->m_pkthdr.len < len) { 235 ipstat.ips_tooshort++; 236 goto bad; 237 } 238 if (m->m_pkthdr.len > len) { 239 if (m->m_len == m->m_pkthdr.len) { 240 m->m_len = len; 241 m->m_pkthdr.len = len; 242 } else 243 m_adj(m, len - m->m_pkthdr.len); 244 } 245 246#ifdef PFIL_HOOKS 247 /* 248 * Run through list of hooks for input packets. 249 */ 250 m0 = m; 251 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 252 if (pfh->pfil_func) { 253 if (pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0)) 254 goto next; 255 else 256 ip = mtod(m = m0, struct ip *); 257 } 258#endif /* PFIL_HOOKS */ 259 260 /* 261 * Process options and, if not destined for us, 262 * ship it on. ip_dooptions returns 1 when an 263 * error was detected (causing an icmp message 264 * to be sent and the original packet to be freed). 265 */ 266 ip_nhops = 0; /* for source routed packets */ 267 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 268 goto next; 269 270 /* 271 * Check our list of addresses, to see if the packet is for us. 272 */ 273 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 274 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 275 goto ours; 276 if (((ip_directedbcast == 0) || (ip_directedbcast && 277 ia->ia_ifp == m->m_pkthdr.rcvif)) && 278 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 279 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 280 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 281 /* 282 * Look for all-0's host part (old broadcast addr), 283 * either for subnet or net. 284 */ 285 ip->ip_dst.s_addr == ia->ia_subnet || 286 ip->ip_dst.s_addr == ia->ia_net) 287 goto ours; 288 } 289 } 290 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 291 struct in_multi *inm; 292#ifdef MROUTING 293 extern struct socket *ip_mrouter; 294 295 if (m->m_flags & M_EXT) { 296 if ((m = m_pullup(m, hlen)) == 0) { 297 ipstat.ips_toosmall++; 298 goto next; 299 } 300 ip = mtod(m, struct ip *); 301 } 302 303 if (ip_mrouter) { 304 /* 305 * If we are acting as a multicast router, all 306 * incoming multicast packets are passed to the 307 * kernel-level multicast forwarding function. 308 * The packet is returned (relatively) intact; if 309 * ip_mforward() returns a non-zero value, the packet 310 * must be discarded, else it may be accepted below. 311 * 312 * (The IP ident field is put in the same byte order 313 * as expected when ip_mforward() is called from 314 * ip_output().) 315 */ 316 ip->ip_id = htons(ip->ip_id); 317 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 318 ipstat.ips_cantforward++; 319 m_freem(m); 320 goto next; 321 } 322 ip->ip_id = ntohs(ip->ip_id); 323 324 /* 325 * The process-level routing demon needs to receive 326 * all multicast IGMP packets, whether or not this 327 * host belongs to their destination groups. 328 */ 329 if (ip->ip_p == IPPROTO_IGMP) 330 goto ours; 331 ipstat.ips_forward++; 332 } 333#endif 334 /* 335 * See if we belong to the destination multicast group on the 336 * arrival interface. 337 */ 338 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 339 if (inm == NULL) { 340 ipstat.ips_cantforward++; 341 m_freem(m); 342 goto next; 343 } 344 goto ours; 345 } 346 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 347 in_nullhost(ip->ip_dst)) 348 goto ours; 349 350 /* 351 * Not for us; forward if possible and desirable. 352 */ 353 if (ipforwarding == 0) { 354 ipstat.ips_cantforward++; 355 m_freem(m); 356 } else 357 ip_forward(m, 0); 358 goto next; 359 360ours: 361 /* 362 * If offset or IP_MF are set, must reassemble. 363 * Otherwise, nothing need be done. 364 * (We could look in the reassembly queue to see 365 * if the packet was previously fragmented, 366 * but it's not worth the time; just let them time out.) 367 */ 368 if (ip->ip_off & ~(IP_DF|IP_RF)) { 369 if (m->m_flags & M_EXT) { /* XXX */ 370 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 371 ipstat.ips_toosmall++; 372 goto next; 373 } 374 ip = mtod(m, struct ip *); 375 } 376 /* 377 * Look for queue of fragments 378 * of this datagram. 379 */ 380 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 381 if (ip->ip_id == fp->ipq_id && 382 in_hosteq(ip->ip_src, fp->ipq_src) && 383 in_hosteq(ip->ip_dst, fp->ipq_dst) && 384 ip->ip_p == fp->ipq_p) 385 goto found; 386 fp = 0; 387found: 388 389 /* 390 * Adjust ip_len to not reflect header, 391 * set ipqe_mff if more fragments are expected, 392 * convert offset of this to bytes. 393 */ 394 ip->ip_len -= hlen; 395 mff = (ip->ip_off & IP_MF) != 0; 396 if (mff) { 397 /* 398 * Make sure that fragments have a data length 399 * that's a non-zero multiple of 8 bytes. 400 */ 401 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 402 ipstat.ips_badfrags++; 403 goto bad; 404 } 405 } 406 ip->ip_off <<= 3; 407 408 /* 409 * If datagram marked as having more fragments 410 * or if this is not the first fragment, 411 * attempt reassembly; if it succeeds, proceed. 412 */ 413 if (mff || ip->ip_off) { 414 ipstat.ips_fragments++; 415 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 416 M_IPQ, M_NOWAIT); 417 if (ipqe == NULL) { 418 ipstat.ips_rcvmemdrop++; 419 goto bad; 420 } 421 ipqe->ipqe_mff = mff; 422 ipqe->ipqe_ip = ip; 423 ip = ip_reass(ipqe, fp); 424 if (ip == 0) 425 goto next; 426 ipstat.ips_reassembled++; 427 m = dtom(ip); 428 } else 429 if (fp) 430 ip_freef(fp); 431 } else 432 ip->ip_len -= hlen; 433 434 /* 435 * Switch out to protocol's input routine. 436 */ 437 ipstat.ips_delivered++; 438 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 439 goto next; 440bad: 441 m_freem(m); 442 goto next; 443} 444 445/* 446 * Take incoming datagram fragment and try to 447 * reassemble it into whole datagram. If a chain for 448 * reassembly of this datagram already exists, then it 449 * is given as fp; otherwise have to make a chain. 450 */ 451struct ip * 452ip_reass(ipqe, fp) 453 register struct ipqent *ipqe; 454 register struct ipq *fp; 455{ 456 register struct mbuf *m = dtom(ipqe->ipqe_ip); 457 register struct ipqent *nq, *p, *q; 458 struct ip *ip; 459 struct mbuf *t; 460 int hlen = ipqe->ipqe_ip->ip_hl << 2; 461 int i, next; 462 463 /* 464 * Presence of header sizes in mbufs 465 * would confuse code below. 466 */ 467 m->m_data += hlen; 468 m->m_len -= hlen; 469 470 /* 471 * If first fragment to arrive, create a reassembly queue. 472 */ 473 if (fp == 0) { 474 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 475 goto dropfrag; 476 fp = mtod(t, struct ipq *); 477 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 478 fp->ipq_ttl = IPFRAGTTL; 479 fp->ipq_p = ipqe->ipqe_ip->ip_p; 480 fp->ipq_id = ipqe->ipqe_ip->ip_id; 481 LIST_INIT(&fp->ipq_fragq); 482 fp->ipq_src = ipqe->ipqe_ip->ip_src; 483 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 484 p = NULL; 485 goto insert; 486 } 487 488 /* 489 * Find a segment which begins after this one does. 490 */ 491 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 492 p = q, q = q->ipqe_q.le_next) 493 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 494 break; 495 496 /* 497 * If there is a preceding segment, it may provide some of 498 * our data already. If so, drop the data from the incoming 499 * segment. If it provides all of our data, drop us. 500 */ 501 if (p != NULL) { 502 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 503 ipqe->ipqe_ip->ip_off; 504 if (i > 0) { 505 if (i >= ipqe->ipqe_ip->ip_len) 506 goto dropfrag; 507 m_adj(dtom(ipqe->ipqe_ip), i); 508 ipqe->ipqe_ip->ip_off += i; 509 ipqe->ipqe_ip->ip_len -= i; 510 } 511 } 512 513 /* 514 * While we overlap succeeding segments trim them or, 515 * if they are completely covered, dequeue them. 516 */ 517 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 518 q->ipqe_ip->ip_off; q = nq) { 519 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 520 q->ipqe_ip->ip_off; 521 if (i < q->ipqe_ip->ip_len) { 522 q->ipqe_ip->ip_len -= i; 523 q->ipqe_ip->ip_off += i; 524 m_adj(dtom(q->ipqe_ip), i); 525 break; 526 } 527 nq = q->ipqe_q.le_next; 528 m_freem(dtom(q->ipqe_ip)); 529 LIST_REMOVE(q, ipqe_q); 530 FREE(q, M_IPQ); 531 } 532 533insert: 534 /* 535 * Stick new segment in its place; 536 * check for complete reassembly. 537 */ 538 if (p == NULL) { 539 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 540 } else { 541 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 542 } 543 next = 0; 544 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 545 p = q, q = q->ipqe_q.le_next) { 546 if (q->ipqe_ip->ip_off != next) 547 return (0); 548 next += q->ipqe_ip->ip_len; 549 } 550 if (p->ipqe_mff) 551 return (0); 552 553 /* 554 * Reassembly is complete. Check for a bogus message size and 555 * concatenate fragments. 556 */ 557 q = fp->ipq_fragq.lh_first; 558 ip = q->ipqe_ip; 559 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 560 ipstat.ips_toolong++; 561 ip_freef(fp); 562 return (0); 563 } 564 m = dtom(q->ipqe_ip); 565 t = m->m_next; 566 m->m_next = 0; 567 m_cat(m, t); 568 nq = q->ipqe_q.le_next; 569 FREE(q, M_IPQ); 570 for (q = nq; q != NULL; q = nq) { 571 t = dtom(q->ipqe_ip); 572 nq = q->ipqe_q.le_next; 573 FREE(q, M_IPQ); 574 m_cat(m, t); 575 } 576 577 /* 578 * Create header for new ip packet by 579 * modifying header of first packet; 580 * dequeue and discard fragment reassembly header. 581 * Make header visible. 582 */ 583 ip->ip_len = next; 584 ip->ip_src = fp->ipq_src; 585 ip->ip_dst = fp->ipq_dst; 586 LIST_REMOVE(fp, ipq_q); 587 (void) m_free(dtom(fp)); 588 m->m_len += (ip->ip_hl << 2); 589 m->m_data -= (ip->ip_hl << 2); 590 /* some debugging cruft by sklower, below, will go away soon */ 591 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 592 register int plen = 0; 593 for (t = m; m; m = m->m_next) 594 plen += m->m_len; 595 t->m_pkthdr.len = plen; 596 } 597 return (ip); 598 599dropfrag: 600 ipstat.ips_fragdropped++; 601 m_freem(m); 602 FREE(ipqe, M_IPQ); 603 return (0); 604} 605 606/* 607 * Free a fragment reassembly header and all 608 * associated datagrams. 609 */ 610void 611ip_freef(fp) 612 struct ipq *fp; 613{ 614 register struct ipqent *q, *p; 615 616 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 617 p = q->ipqe_q.le_next; 618 m_freem(dtom(q->ipqe_ip)); 619 LIST_REMOVE(q, ipqe_q); 620 FREE(q, M_IPQ); 621 } 622 LIST_REMOVE(fp, ipq_q); 623 (void) m_free(dtom(fp)); 624} 625 626/* 627 * IP timer processing; 628 * if a timer expires on a reassembly 629 * queue, discard it. 630 */ 631void 632ip_slowtimo() 633{ 634 register struct ipq *fp, *nfp; 635 int s = splsoftnet(); 636 637 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 638 nfp = fp->ipq_q.le_next; 639 if (--fp->ipq_ttl == 0) { 640 ipstat.ips_fragtimeout++; 641 ip_freef(fp); 642 } 643 } 644 splx(s); 645} 646 647/* 648 * Drain off all datagram fragments. 649 */ 650void 651ip_drain() 652{ 653 654 while (ipq.lh_first != NULL) { 655 ipstat.ips_fragdropped++; 656 ip_freef(ipq.lh_first); 657 } 658} 659 660/* 661 * Do option processing on a datagram, 662 * possibly discarding it if bad options are encountered, 663 * or forwarding it if source-routed. 664 * Returns 1 if packet has been forwarded/freed, 665 * 0 if the packet should be processed further. 666 */ 667int 668ip_dooptions(m) 669 struct mbuf *m; 670{ 671 register struct ip *ip = mtod(m, struct ip *); 672 register u_char *cp; 673 register struct ip_timestamp *ipt; 674 register struct in_ifaddr *ia; 675 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 676 struct in_addr *sin, dst; 677 n_time ntime; 678 679 dst = ip->ip_dst; 680 cp = (u_char *)(ip + 1); 681 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 682 for (; cnt > 0; cnt -= optlen, cp += optlen) { 683 opt = cp[IPOPT_OPTVAL]; 684 if (opt == IPOPT_EOL) 685 break; 686 if (opt == IPOPT_NOP) 687 optlen = 1; 688 else { 689 optlen = cp[IPOPT_OLEN]; 690 if (optlen <= 0 || optlen > cnt) { 691 code = &cp[IPOPT_OLEN] - (u_char *)ip; 692 goto bad; 693 } 694 } 695 switch (opt) { 696 697 default: 698 break; 699 700 /* 701 * Source routing with record. 702 * Find interface with current destination address. 703 * If none on this machine then drop if strictly routed, 704 * or do nothing if loosely routed. 705 * Record interface address and bring up next address 706 * component. If strictly routed make sure next 707 * address is on directly accessible net. 708 */ 709 case IPOPT_LSRR: 710 case IPOPT_SSRR: 711 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 712 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 713 goto bad; 714 } 715 ipaddr.sin_addr = ip->ip_dst; 716 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 717 if (ia == 0) { 718 if (opt == IPOPT_SSRR) { 719 type = ICMP_UNREACH; 720 code = ICMP_UNREACH_SRCFAIL; 721 goto bad; 722 } 723 /* 724 * Loose routing, and not at next destination 725 * yet; nothing to do except forward. 726 */ 727 break; 728 } 729 off--; /* 0 origin */ 730 if (off > optlen - sizeof(struct in_addr)) { 731 /* 732 * End of source route. Should be for us. 733 */ 734 save_rte(cp, ip->ip_src); 735 break; 736 } 737 /* 738 * locate outgoing interface 739 */ 740 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 741 sizeof(ipaddr.sin_addr)); 742 if (opt == IPOPT_SSRR) { 743#define INA struct in_ifaddr * 744#define SA struct sockaddr * 745 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 746 } else 747 ia = ip_rtaddr(ipaddr.sin_addr); 748 if (ia == 0) { 749 type = ICMP_UNREACH; 750 code = ICMP_UNREACH_SRCFAIL; 751 goto bad; 752 } 753 ip->ip_dst = ipaddr.sin_addr; 754 bcopy((caddr_t)&ia->ia_addr.sin_addr, 755 (caddr_t)(cp + off), sizeof(struct in_addr)); 756 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 757 /* 758 * Let ip_intr's mcast routing check handle mcast pkts 759 */ 760 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 761 break; 762 763 case IPOPT_RR: 764 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 765 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 766 goto bad; 767 } 768 /* 769 * If no space remains, ignore. 770 */ 771 off--; /* 0 origin */ 772 if (off > optlen - sizeof(struct in_addr)) 773 break; 774 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 775 sizeof(ipaddr.sin_addr)); 776 /* 777 * locate outgoing interface; if we're the destination, 778 * use the incoming interface (should be same). 779 */ 780 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 781 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 782 type = ICMP_UNREACH; 783 code = ICMP_UNREACH_HOST; 784 goto bad; 785 } 786 bcopy((caddr_t)&ia->ia_addr.sin_addr, 787 (caddr_t)(cp + off), sizeof(struct in_addr)); 788 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 789 break; 790 791 case IPOPT_TS: 792 code = cp - (u_char *)ip; 793 ipt = (struct ip_timestamp *)cp; 794 if (ipt->ipt_len < 5) 795 goto bad; 796 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 797 if (++ipt->ipt_oflw == 0) 798 goto bad; 799 break; 800 } 801 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 802 switch (ipt->ipt_flg) { 803 804 case IPOPT_TS_TSONLY: 805 break; 806 807 case IPOPT_TS_TSANDADDR: 808 if (ipt->ipt_ptr + sizeof(n_time) + 809 sizeof(struct in_addr) > ipt->ipt_len) 810 goto bad; 811 ipaddr.sin_addr = dst; 812 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 813 m->m_pkthdr.rcvif); 814 if (ia == 0) 815 continue; 816 bcopy((caddr_t)&ia->ia_addr.sin_addr, 817 (caddr_t)sin, sizeof(struct in_addr)); 818 ipt->ipt_ptr += sizeof(struct in_addr); 819 break; 820 821 case IPOPT_TS_PRESPEC: 822 if (ipt->ipt_ptr + sizeof(n_time) + 823 sizeof(struct in_addr) > ipt->ipt_len) 824 goto bad; 825 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 826 sizeof(struct in_addr)); 827 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 828 continue; 829 ipt->ipt_ptr += sizeof(struct in_addr); 830 break; 831 832 default: 833 goto bad; 834 } 835 ntime = iptime(); 836 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 837 sizeof(n_time)); 838 ipt->ipt_ptr += sizeof(n_time); 839 } 840 } 841 if (forward) { 842 if (ip_forwsrcrt == 0) { 843 type = ICMP_UNREACH; 844 code = ICMP_UNREACH_SRCFAIL; 845 goto bad; 846 } 847 ip_forward(m, 1); 848 return (1); 849 } 850 return (0); 851bad: 852 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 853 icmp_error(m, type, code, 0, 0); 854 ipstat.ips_badoptions++; 855 return (1); 856} 857 858/* 859 * Given address of next destination (final or next hop), 860 * return internet address info of interface to be used to get there. 861 */ 862struct in_ifaddr * 863ip_rtaddr(dst) 864 struct in_addr dst; 865{ 866 register struct sockaddr_in *sin; 867 868 sin = satosin(&ipforward_rt.ro_dst); 869 870 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 871 if (ipforward_rt.ro_rt) { 872 RTFREE(ipforward_rt.ro_rt); 873 ipforward_rt.ro_rt = 0; 874 } 875 sin->sin_family = AF_INET; 876 sin->sin_len = sizeof(*sin); 877 sin->sin_addr = dst; 878 879 rtalloc(&ipforward_rt); 880 } 881 if (ipforward_rt.ro_rt == 0) 882 return ((struct in_ifaddr *)0); 883 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 884} 885 886/* 887 * Save incoming source route for use in replies, 888 * to be picked up later by ip_srcroute if the receiver is interested. 889 */ 890void 891save_rte(option, dst) 892 u_char *option; 893 struct in_addr dst; 894{ 895 unsigned olen; 896 897 olen = option[IPOPT_OLEN]; 898#ifdef DIAGNOSTIC 899 if (ipprintfs) 900 printf("save_rte: olen %d\n", olen); 901#endif 902 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 903 return; 904 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 905 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 906 ip_srcrt.dst = dst; 907} 908 909/* 910 * Retrieve incoming source route for use in replies, 911 * in the same form used by setsockopt. 912 * The first hop is placed before the options, will be removed later. 913 */ 914struct mbuf * 915ip_srcroute() 916{ 917 register struct in_addr *p, *q; 918 register struct mbuf *m; 919 920 if (ip_nhops == 0) 921 return ((struct mbuf *)0); 922 m = m_get(M_DONTWAIT, MT_SOOPTS); 923 if (m == 0) 924 return ((struct mbuf *)0); 925 926#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 927 928 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 929 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 930 OPTSIZ; 931#ifdef DIAGNOSTIC 932 if (ipprintfs) 933 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 934#endif 935 936 /* 937 * First save first hop for return route 938 */ 939 p = &ip_srcrt.route[ip_nhops - 1]; 940 *(mtod(m, struct in_addr *)) = *p--; 941#ifdef DIAGNOSTIC 942 if (ipprintfs) 943 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 944#endif 945 946 /* 947 * Copy option fields and padding (nop) to mbuf. 948 */ 949 ip_srcrt.nop = IPOPT_NOP; 950 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 951 bcopy((caddr_t)&ip_srcrt.nop, 952 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 953 q = (struct in_addr *)(mtod(m, caddr_t) + 954 sizeof(struct in_addr) + OPTSIZ); 955#undef OPTSIZ 956 /* 957 * Record return path as an IP source route, 958 * reversing the path (pointers are now aligned). 959 */ 960 while (p >= ip_srcrt.route) { 961#ifdef DIAGNOSTIC 962 if (ipprintfs) 963 printf(" %x", ntohl(q->s_addr)); 964#endif 965 *q++ = *p--; 966 } 967 /* 968 * Last hop goes to final destination. 969 */ 970 *q = ip_srcrt.dst; 971#ifdef DIAGNOSTIC 972 if (ipprintfs) 973 printf(" %x\n", ntohl(q->s_addr)); 974#endif 975 return (m); 976} 977 978/* 979 * Strip out IP options, at higher 980 * level protocol in the kernel. 981 * Second argument is buffer to which options 982 * will be moved, and return value is their length. 983 * XXX should be deleted; last arg currently ignored. 984 */ 985void 986ip_stripoptions(m, mopt) 987 register struct mbuf *m; 988 struct mbuf *mopt; 989{ 990 register int i; 991 struct ip *ip = mtod(m, struct ip *); 992 register caddr_t opts; 993 int olen; 994 995 olen = (ip->ip_hl<<2) - sizeof (struct ip); 996 opts = (caddr_t)(ip + 1); 997 i = m->m_len - (sizeof (struct ip) + olen); 998 bcopy(opts + olen, opts, (unsigned)i); 999 m->m_len -= olen; 1000 if (m->m_flags & M_PKTHDR) 1001 m->m_pkthdr.len -= olen; 1002 ip->ip_hl = sizeof(struct ip) >> 2; 1003} 1004 1005int inetctlerrmap[PRC_NCMDS] = { 1006 0, 0, 0, 0, 1007 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1008 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1009 EMSGSIZE, EHOSTUNREACH, 0, 0, 1010 0, 0, 0, 0, 1011 ENOPROTOOPT 1012}; 1013 1014/* 1015 * Forward a packet. If some error occurs return the sender 1016 * an icmp packet. Note we can't always generate a meaningful 1017 * icmp message because icmp doesn't have a large enough repertoire 1018 * of codes and types. 1019 * 1020 * If not forwarding, just drop the packet. This could be confusing 1021 * if ipforwarding was zero but some routing protocol was advancing 1022 * us as a gateway to somewhere. However, we must let the routing 1023 * protocol deal with that. 1024 * 1025 * The srcrt parameter indicates whether the packet is being forwarded 1026 * via a source route. 1027 */ 1028void 1029ip_forward(m, srcrt) 1030 struct mbuf *m; 1031 int srcrt; 1032{ 1033 register struct ip *ip = mtod(m, struct ip *); 1034 register struct sockaddr_in *sin; 1035 register struct rtentry *rt; 1036 int error, type = 0, code = 0; 1037 struct mbuf *mcopy; 1038 n_long dest; 1039 struct ifnet *destifp; 1040 1041 dest = 0; 1042#ifdef DIAGNOSTIC 1043 if (ipprintfs) 1044 printf("forward: src %x dst %x ttl %x\n", 1045 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1046#endif 1047 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1048 ipstat.ips_cantforward++; 1049 m_freem(m); 1050 return; 1051 } 1052 HTONS(ip->ip_id); 1053 if (ip->ip_ttl <= IPTTLDEC) { 1054 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1055 return; 1056 } 1057 ip->ip_ttl -= IPTTLDEC; 1058 1059 sin = satosin(&ipforward_rt.ro_dst); 1060 if ((rt = ipforward_rt.ro_rt) == 0 || 1061 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1062 if (ipforward_rt.ro_rt) { 1063 RTFREE(ipforward_rt.ro_rt); 1064 ipforward_rt.ro_rt = 0; 1065 } 1066 sin->sin_family = AF_INET; 1067 sin->sin_len = sizeof(struct sockaddr_in); 1068 sin->sin_addr = ip->ip_dst; 1069 1070 rtalloc(&ipforward_rt); 1071 if (ipforward_rt.ro_rt == 0) { 1072 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1073 return; 1074 } 1075 rt = ipforward_rt.ro_rt; 1076 } 1077 1078 /* 1079 * Save at most 68 bytes of the packet in case 1080 * we need to generate an ICMP message to the src. 1081 */ 1082 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1083 1084 /* 1085 * If forwarding packet using same interface that it came in on, 1086 * perhaps should send a redirect to sender to shortcut a hop. 1087 * Only send redirect if source is sending directly to us, 1088 * and if packet was not source routed (or has any options). 1089 * Also, don't send redirect if forwarding using a default route 1090 * or a route modified by a redirect. 1091 */ 1092 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1093 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1094 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1095 ipsendredirects && !srcrt) { 1096 if (rt->rt_ifa && 1097 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1098 ifatoia(rt->rt_ifa)->ia_subnet) { 1099 if (rt->rt_flags & RTF_GATEWAY) 1100 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1101 else 1102 dest = ip->ip_dst.s_addr; 1103 /* Router requirements says to only send host redirects */ 1104 type = ICMP_REDIRECT; 1105 code = ICMP_REDIRECT_HOST; 1106#ifdef DIAGNOSTIC 1107 if (ipprintfs) 1108 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1109#endif 1110 } 1111 } 1112 1113 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1114 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1115 if (error) 1116 ipstat.ips_cantforward++; 1117 else { 1118 ipstat.ips_forward++; 1119 if (type) 1120 ipstat.ips_redirectsent++; 1121 else { 1122 if (mcopy) 1123 m_freem(mcopy); 1124 return; 1125 } 1126 } 1127 if (mcopy == NULL) 1128 return; 1129 destifp = NULL; 1130 1131 switch (error) { 1132 1133 case 0: /* forwarded, but need redirect */ 1134 /* type, code set above */ 1135 break; 1136 1137 case ENETUNREACH: /* shouldn't happen, checked above */ 1138 case EHOSTUNREACH: 1139 case ENETDOWN: 1140 case EHOSTDOWN: 1141 default: 1142 type = ICMP_UNREACH; 1143 code = ICMP_UNREACH_HOST; 1144 break; 1145 1146 case EMSGSIZE: 1147 type = ICMP_UNREACH; 1148 code = ICMP_UNREACH_NEEDFRAG; 1149 if (ipforward_rt.ro_rt) 1150 destifp = ipforward_rt.ro_rt->rt_ifp; 1151 ipstat.ips_cantfrag++; 1152 break; 1153 1154 case ENOBUFS: 1155 type = ICMP_SOURCEQUENCH; 1156 code = 0; 1157 break; 1158 } 1159 icmp_error(mcopy, type, code, dest, destifp); 1160} 1161 1162int 1163ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1164 int *name; 1165 u_int namelen; 1166 void *oldp; 1167 size_t *oldlenp; 1168 void *newp; 1169 size_t newlen; 1170{ 1171 /* All sysctl names at this level are terminal. */ 1172 if (namelen != 1) 1173 return (ENOTDIR); 1174 1175 switch (name[0]) { 1176 case IPCTL_FORWARDING: 1177 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1178 case IPCTL_SENDREDIRECTS: 1179 return (sysctl_int(oldp, oldlenp, newp, newlen, 1180 &ipsendredirects)); 1181 case IPCTL_DEFTTL: 1182 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1183#ifdef notyet 1184 case IPCTL_DEFMTU: 1185 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1186#endif 1187 case IPCTL_FORWSRCRT: 1188 /* 1189 * Don't allow this to change in a secure environment. 1190 */ 1191 if (securelevel > 0) 1192 return (EPERM); 1193 return (sysctl_int(oldp, oldlenp, newp, newlen, 1194 &ip_forwsrcrt)); 1195 case IPCTL_DIRECTEDBCAST: 1196 return (sysctl_int(oldp, oldlenp, newp, newlen, 1197 &ip_directedbcast)); 1198 default: 1199 return (EOPNOTSUPP); 1200 } 1201 /* NOTREACHED */ 1202} 1203