ip_input.c revision 1.36
1/* $NetBSD: ip_input.c,v 1.36 1996/09/14 14:40:26 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 bad; 255 ip = mtod(m = m0, struct ip *); 256 } 257#endif /* PFIL_HOOKS */ 258 259 /* 260 * Process options and, if not destined for us, 261 * ship it on. ip_dooptions returns 1 when an 262 * error was detected (causing an icmp message 263 * to be sent and the original packet to be freed). 264 */ 265 ip_nhops = 0; /* for source routed packets */ 266 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 267 goto next; 268 269 /* 270 * Check our list of addresses, to see if the packet is for us. 271 */ 272 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 273 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 274 goto ours; 275 if (((ip_directedbcast == 0) || (ip_directedbcast && 276 ia->ia_ifp == m->m_pkthdr.rcvif)) && 277 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 278 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 279 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 280 /* 281 * Look for all-0's host part (old broadcast addr), 282 * either for subnet or net. 283 */ 284 ip->ip_dst.s_addr == ia->ia_subnet || 285 ip->ip_dst.s_addr == ia->ia_net) 286 goto ours; 287 } 288 } 289 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 290 struct in_multi *inm; 291#ifdef MROUTING 292 extern struct socket *ip_mrouter; 293 294 if (m->m_flags & M_EXT) { 295 if ((m = m_pullup(m, hlen)) == 0) { 296 ipstat.ips_toosmall++; 297 goto next; 298 } 299 ip = mtod(m, struct ip *); 300 } 301 302 if (ip_mrouter) { 303 /* 304 * If we are acting as a multicast router, all 305 * incoming multicast packets are passed to the 306 * kernel-level multicast forwarding function. 307 * The packet is returned (relatively) intact; if 308 * ip_mforward() returns a non-zero value, the packet 309 * must be discarded, else it may be accepted below. 310 * 311 * (The IP ident field is put in the same byte order 312 * as expected when ip_mforward() is called from 313 * ip_output().) 314 */ 315 ip->ip_id = htons(ip->ip_id); 316 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 317 ipstat.ips_cantforward++; 318 m_freem(m); 319 goto next; 320 } 321 ip->ip_id = ntohs(ip->ip_id); 322 323 /* 324 * The process-level routing demon needs to receive 325 * all multicast IGMP packets, whether or not this 326 * host belongs to their destination groups. 327 */ 328 if (ip->ip_p == IPPROTO_IGMP) 329 goto ours; 330 ipstat.ips_forward++; 331 } 332#endif 333 /* 334 * See if we belong to the destination multicast group on the 335 * arrival interface. 336 */ 337 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 338 if (inm == NULL) { 339 ipstat.ips_cantforward++; 340 m_freem(m); 341 goto next; 342 } 343 goto ours; 344 } 345 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 346 in_nullhost(ip->ip_dst)) 347 goto ours; 348 349 /* 350 * Not for us; forward if possible and desirable. 351 */ 352 if (ipforwarding == 0) { 353 ipstat.ips_cantforward++; 354 m_freem(m); 355 } else 356 ip_forward(m, 0); 357 goto next; 358 359ours: 360 /* 361 * If offset or IP_MF are set, must reassemble. 362 * Otherwise, nothing need be done. 363 * (We could look in the reassembly queue to see 364 * if the packet was previously fragmented, 365 * but it's not worth the time; just let them time out.) 366 */ 367 if (ip->ip_off &~ IP_DF) { 368 if (m->m_flags & M_EXT) { /* XXX */ 369 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 370 ipstat.ips_toosmall++; 371 goto next; 372 } 373 ip = mtod(m, struct ip *); 374 } 375 /* 376 * Look for queue of fragments 377 * of this datagram. 378 */ 379 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 380 if (ip->ip_id == fp->ipq_id && 381 in_hosteq(ip->ip_src, fp->ipq_src) && 382 in_hosteq(ip->ip_dst, fp->ipq_dst) && 383 ip->ip_p == fp->ipq_p) 384 goto found; 385 fp = 0; 386found: 387 388 /* 389 * Adjust ip_len to not reflect header, 390 * set ipqe_mff if more fragments are expected, 391 * convert offset of this to bytes. 392 */ 393 ip->ip_len -= hlen; 394 mff = (ip->ip_off & IP_MF) != 0; 395 if (mff) { 396 /* 397 * Make sure that fragments have a data length 398 * that's a non-zero multiple of 8 bytes. 399 */ 400 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 401 ipstat.ips_badfrags++; 402 goto bad; 403 } 404 } 405 ip->ip_off <<= 3; 406 407 /* 408 * If datagram marked as having more fragments 409 * or if this is not the first fragment, 410 * attempt reassembly; if it succeeds, proceed. 411 */ 412 if (mff || ip->ip_off) { 413 ipstat.ips_fragments++; 414 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 415 M_IPQ, M_NOWAIT); 416 if (ipqe == NULL) { 417 ipstat.ips_rcvmemdrop++; 418 goto bad; 419 } 420 ipqe->ipqe_mff = mff; 421 ipqe->ipqe_ip = ip; 422 ip = ip_reass(ipqe, fp); 423 if (ip == 0) 424 goto next; 425 ipstat.ips_reassembled++; 426 m = dtom(ip); 427 } else 428 if (fp) 429 ip_freef(fp); 430 } else 431 ip->ip_len -= hlen; 432 433 /* 434 * Switch out to protocol's input routine. 435 */ 436 ipstat.ips_delivered++; 437 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 438 goto next; 439bad: 440#ifdef PFIL_HOOKS 441 m0 = m; 442 for (pfh = pfil_hook_get(PFIL_BAD); pfh; pfh = pfh->pfil_link.le_next) 443 if (pfh->pfil_func) { 444 (void)pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 2, &m0); 445 ip = mtod(m = m0, struct ip *); 446 } 447#endif /* PFIL_HOOKS */ 448 m_freem(m); 449 goto next; 450} 451 452/* 453 * Take incoming datagram fragment and try to 454 * reassemble it into whole datagram. If a chain for 455 * reassembly of this datagram already exists, then it 456 * is given as fp; otherwise have to make a chain. 457 */ 458struct ip * 459ip_reass(ipqe, fp) 460 register struct ipqent *ipqe; 461 register struct ipq *fp; 462{ 463 register struct mbuf *m = dtom(ipqe->ipqe_ip); 464 register struct ipqent *nq, *p, *q; 465 struct ip *ip; 466 struct mbuf *t; 467 int hlen = ipqe->ipqe_ip->ip_hl << 2; 468 int i, next; 469 470 /* 471 * Presence of header sizes in mbufs 472 * would confuse code below. 473 */ 474 m->m_data += hlen; 475 m->m_len -= hlen; 476 477 /* 478 * If first fragment to arrive, create a reassembly queue. 479 */ 480 if (fp == 0) { 481 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 482 goto dropfrag; 483 fp = mtod(t, struct ipq *); 484 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 485 fp->ipq_ttl = IPFRAGTTL; 486 fp->ipq_p = ipqe->ipqe_ip->ip_p; 487 fp->ipq_id = ipqe->ipqe_ip->ip_id; 488 LIST_INIT(&fp->ipq_fragq); 489 fp->ipq_src = ipqe->ipqe_ip->ip_src; 490 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 491 p = NULL; 492 goto insert; 493 } 494 495 /* 496 * Find a segment which begins after this one does. 497 */ 498 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 499 p = q, q = q->ipqe_q.le_next) 500 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 501 break; 502 503 /* 504 * If there is a preceding segment, it may provide some of 505 * our data already. If so, drop the data from the incoming 506 * segment. If it provides all of our data, drop us. 507 */ 508 if (p != NULL) { 509 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 510 ipqe->ipqe_ip->ip_off; 511 if (i > 0) { 512 if (i >= ipqe->ipqe_ip->ip_len) 513 goto dropfrag; 514 m_adj(dtom(ipqe->ipqe_ip), i); 515 ipqe->ipqe_ip->ip_off += i; 516 ipqe->ipqe_ip->ip_len -= i; 517 } 518 } 519 520 /* 521 * While we overlap succeeding segments trim them or, 522 * if they are completely covered, dequeue them. 523 */ 524 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 525 q->ipqe_ip->ip_off; q = nq) { 526 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 527 q->ipqe_ip->ip_off; 528 if (i < q->ipqe_ip->ip_len) { 529 q->ipqe_ip->ip_len -= i; 530 q->ipqe_ip->ip_off += i; 531 m_adj(dtom(q->ipqe_ip), i); 532 break; 533 } 534 nq = q->ipqe_q.le_next; 535 m_freem(dtom(q->ipqe_ip)); 536 LIST_REMOVE(q, ipqe_q); 537 FREE(q, M_IPQ); 538 } 539 540insert: 541 /* 542 * Stick new segment in its place; 543 * check for complete reassembly. 544 */ 545 if (p == NULL) { 546 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 547 } else { 548 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 549 } 550 next = 0; 551 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 552 p = q, q = q->ipqe_q.le_next) { 553 if (q->ipqe_ip->ip_off != next) 554 return (0); 555 next += q->ipqe_ip->ip_len; 556 } 557 if (p->ipqe_mff) 558 return (0); 559 560 /* 561 * Reassembly is complete; concatenate fragments. 562 */ 563 q = fp->ipq_fragq.lh_first; 564 ip = q->ipqe_ip; 565 m = dtom(q->ipqe_ip); 566 t = m->m_next; 567 m->m_next = 0; 568 m_cat(m, t); 569 nq = q->ipqe_q.le_next; 570 FREE(q, M_IPQ); 571 for (q = nq; q != NULL; q = nq) { 572 t = dtom(q->ipqe_ip); 573 nq = q->ipqe_q.le_next; 574 FREE(q, M_IPQ); 575 m_cat(m, t); 576 } 577 578 /* 579 * Create header for new ip packet by 580 * modifying header of first packet; 581 * dequeue and discard fragment reassembly header. 582 * Make header visible. 583 */ 584 ip->ip_len = next; 585 ip->ip_src = fp->ipq_src; 586 ip->ip_dst = fp->ipq_dst; 587 LIST_REMOVE(fp, ipq_q); 588 (void) m_free(dtom(fp)); 589 m->m_len += (ip->ip_hl << 2); 590 m->m_data -= (ip->ip_hl << 2); 591 /* some debugging cruft by sklower, below, will go away soon */ 592 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 593 register int plen = 0; 594 for (t = m; m; m = m->m_next) 595 plen += m->m_len; 596 t->m_pkthdr.len = plen; 597 } 598 return (ip); 599 600dropfrag: 601 ipstat.ips_fragdropped++; 602 m_freem(m); 603 FREE(ipqe, M_IPQ); 604 return (0); 605} 606 607/* 608 * Free a fragment reassembly header and all 609 * associated datagrams. 610 */ 611void 612ip_freef(fp) 613 struct ipq *fp; 614{ 615 register struct ipqent *q, *p; 616 617 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 618 p = q->ipqe_q.le_next; 619 m_freem(dtom(q->ipqe_ip)); 620 LIST_REMOVE(q, ipqe_q); 621 FREE(q, M_IPQ); 622 } 623 LIST_REMOVE(fp, ipq_q); 624 (void) m_free(dtom(fp)); 625} 626 627/* 628 * IP timer processing; 629 * if a timer expires on a reassembly 630 * queue, discard it. 631 */ 632void 633ip_slowtimo() 634{ 635 register struct ipq *fp, *nfp; 636 int s = splsoftnet(); 637 638 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 639 nfp = fp->ipq_q.le_next; 640 if (--fp->ipq_ttl == 0) { 641 ipstat.ips_fragtimeout++; 642 ip_freef(fp); 643 } 644 } 645 splx(s); 646} 647 648/* 649 * Drain off all datagram fragments. 650 */ 651void 652ip_drain() 653{ 654 655 while (ipq.lh_first != NULL) { 656 ipstat.ips_fragdropped++; 657 ip_freef(ipq.lh_first); 658 } 659} 660 661/* 662 * Do option processing on a datagram, 663 * possibly discarding it if bad options are encountered, 664 * or forwarding it if source-routed. 665 * Returns 1 if packet has been forwarded/freed, 666 * 0 if the packet should be processed further. 667 */ 668int 669ip_dooptions(m) 670 struct mbuf *m; 671{ 672 register struct ip *ip = mtod(m, struct ip *); 673 register u_char *cp; 674 register struct ip_timestamp *ipt; 675 register struct in_ifaddr *ia; 676 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 677 struct in_addr *sin, dst; 678 n_time ntime; 679 680 dst = ip->ip_dst; 681 cp = (u_char *)(ip + 1); 682 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 683 for (; cnt > 0; cnt -= optlen, cp += optlen) { 684 opt = cp[IPOPT_OPTVAL]; 685 if (opt == IPOPT_EOL) 686 break; 687 if (opt == IPOPT_NOP) 688 optlen = 1; 689 else { 690 optlen = cp[IPOPT_OLEN]; 691 if (optlen <= 0 || optlen > cnt) { 692 code = &cp[IPOPT_OLEN] - (u_char *)ip; 693 goto bad; 694 } 695 } 696 switch (opt) { 697 698 default: 699 break; 700 701 /* 702 * Source routing with record. 703 * Find interface with current destination address. 704 * If none on this machine then drop if strictly routed, 705 * or do nothing if loosely routed. 706 * Record interface address and bring up next address 707 * component. If strictly routed make sure next 708 * address is on directly accessible net. 709 */ 710 case IPOPT_LSRR: 711 case IPOPT_SSRR: 712 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 713 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 714 goto bad; 715 } 716 ipaddr.sin_addr = ip->ip_dst; 717 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 718 if (ia == 0) { 719 if (opt == IPOPT_SSRR) { 720 type = ICMP_UNREACH; 721 code = ICMP_UNREACH_SRCFAIL; 722 goto bad; 723 } 724 /* 725 * Loose routing, and not at next destination 726 * yet; nothing to do except forward. 727 */ 728 break; 729 } 730 off--; /* 0 origin */ 731 if (off > optlen - sizeof(struct in_addr)) { 732 /* 733 * End of source route. Should be for us. 734 */ 735 save_rte(cp, ip->ip_src); 736 break; 737 } 738 /* 739 * locate outgoing interface 740 */ 741 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 742 sizeof(ipaddr.sin_addr)); 743 if (opt == IPOPT_SSRR) { 744#define INA struct in_ifaddr * 745#define SA struct sockaddr * 746 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 747 } else 748 ia = ip_rtaddr(ipaddr.sin_addr); 749 if (ia == 0) { 750 type = ICMP_UNREACH; 751 code = ICMP_UNREACH_SRCFAIL; 752 goto bad; 753 } 754 ip->ip_dst = ipaddr.sin_addr; 755 bcopy((caddr_t)&ia->ia_addr.sin_addr, 756 (caddr_t)(cp + off), sizeof(struct in_addr)); 757 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 758 /* 759 * Let ip_intr's mcast routing check handle mcast pkts 760 */ 761 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 762 break; 763 764 case IPOPT_RR: 765 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 766 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 767 goto bad; 768 } 769 /* 770 * If no space remains, ignore. 771 */ 772 off--; /* 0 origin */ 773 if (off > optlen - sizeof(struct in_addr)) 774 break; 775 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 776 sizeof(ipaddr.sin_addr)); 777 /* 778 * locate outgoing interface; if we're the destination, 779 * use the incoming interface (should be same). 780 */ 781 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 782 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 783 type = ICMP_UNREACH; 784 code = ICMP_UNREACH_HOST; 785 goto bad; 786 } 787 bcopy((caddr_t)&ia->ia_addr.sin_addr, 788 (caddr_t)(cp + off), sizeof(struct in_addr)); 789 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 790 break; 791 792 case IPOPT_TS: 793 code = cp - (u_char *)ip; 794 ipt = (struct ip_timestamp *)cp; 795 if (ipt->ipt_len < 5) 796 goto bad; 797 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 798 if (++ipt->ipt_oflw == 0) 799 goto bad; 800 break; 801 } 802 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 803 switch (ipt->ipt_flg) { 804 805 case IPOPT_TS_TSONLY: 806 break; 807 808 case IPOPT_TS_TSANDADDR: 809 if (ipt->ipt_ptr + sizeof(n_time) + 810 sizeof(struct in_addr) > ipt->ipt_len) 811 goto bad; 812 ipaddr.sin_addr = dst; 813 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 814 m->m_pkthdr.rcvif); 815 if (ia == 0) 816 continue; 817 bcopy((caddr_t)&ia->ia_addr.sin_addr, 818 (caddr_t)sin, sizeof(struct in_addr)); 819 ipt->ipt_ptr += sizeof(struct in_addr); 820 break; 821 822 case IPOPT_TS_PRESPEC: 823 if (ipt->ipt_ptr + sizeof(n_time) + 824 sizeof(struct in_addr) > ipt->ipt_len) 825 goto bad; 826 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 827 sizeof(struct in_addr)); 828 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 829 continue; 830 ipt->ipt_ptr += sizeof(struct in_addr); 831 break; 832 833 default: 834 goto bad; 835 } 836 ntime = iptime(); 837 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 838 sizeof(n_time)); 839 ipt->ipt_ptr += sizeof(n_time); 840 } 841 } 842 if (forward) { 843 if (ip_forwsrcrt == 0) { 844 type = ICMP_UNREACH; 845 code = ICMP_UNREACH_SRCFAIL; 846 goto bad; 847 } 848 ip_forward(m, 1); 849 return (1); 850 } 851 return (0); 852bad: 853 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 854 icmp_error(m, type, code, 0, 0); 855 ipstat.ips_badoptions++; 856 return (1); 857} 858 859/* 860 * Given address of next destination (final or next hop), 861 * return internet address info of interface to be used to get there. 862 */ 863struct in_ifaddr * 864ip_rtaddr(dst) 865 struct in_addr dst; 866{ 867 register struct sockaddr_in *sin; 868 869 sin = satosin(&ipforward_rt.ro_dst); 870 871 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 872 if (ipforward_rt.ro_rt) { 873 RTFREE(ipforward_rt.ro_rt); 874 ipforward_rt.ro_rt = 0; 875 } 876 sin->sin_family = AF_INET; 877 sin->sin_len = sizeof(*sin); 878 sin->sin_addr = dst; 879 880 rtalloc(&ipforward_rt); 881 } 882 if (ipforward_rt.ro_rt == 0) 883 return ((struct in_ifaddr *)0); 884 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 885} 886 887/* 888 * Save incoming source route for use in replies, 889 * to be picked up later by ip_srcroute if the receiver is interested. 890 */ 891void 892save_rte(option, dst) 893 u_char *option; 894 struct in_addr dst; 895{ 896 unsigned olen; 897 898 olen = option[IPOPT_OLEN]; 899#ifdef DIAGNOSTIC 900 if (ipprintfs) 901 printf("save_rte: olen %d\n", olen); 902#endif 903 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 904 return; 905 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 906 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 907 ip_srcrt.dst = dst; 908} 909 910/* 911 * Retrieve incoming source route for use in replies, 912 * in the same form used by setsockopt. 913 * The first hop is placed before the options, will be removed later. 914 */ 915struct mbuf * 916ip_srcroute() 917{ 918 register struct in_addr *p, *q; 919 register struct mbuf *m; 920 921 if (ip_nhops == 0) 922 return ((struct mbuf *)0); 923 m = m_get(M_DONTWAIT, MT_SOOPTS); 924 if (m == 0) 925 return ((struct mbuf *)0); 926 927#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 928 929 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 930 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 931 OPTSIZ; 932#ifdef DIAGNOSTIC 933 if (ipprintfs) 934 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 935#endif 936 937 /* 938 * First save first hop for return route 939 */ 940 p = &ip_srcrt.route[ip_nhops - 1]; 941 *(mtod(m, struct in_addr *)) = *p--; 942#ifdef DIAGNOSTIC 943 if (ipprintfs) 944 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 945#endif 946 947 /* 948 * Copy option fields and padding (nop) to mbuf. 949 */ 950 ip_srcrt.nop = IPOPT_NOP; 951 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 952 bcopy((caddr_t)&ip_srcrt.nop, 953 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 954 q = (struct in_addr *)(mtod(m, caddr_t) + 955 sizeof(struct in_addr) + OPTSIZ); 956#undef OPTSIZ 957 /* 958 * Record return path as an IP source route, 959 * reversing the path (pointers are now aligned). 960 */ 961 while (p >= ip_srcrt.route) { 962#ifdef DIAGNOSTIC 963 if (ipprintfs) 964 printf(" %x", ntohl(q->s_addr)); 965#endif 966 *q++ = *p--; 967 } 968 /* 969 * Last hop goes to final destination. 970 */ 971 *q = ip_srcrt.dst; 972#ifdef DIAGNOSTIC 973 if (ipprintfs) 974 printf(" %x\n", ntohl(q->s_addr)); 975#endif 976 return (m); 977} 978 979/* 980 * Strip out IP options, at higher 981 * level protocol in the kernel. 982 * Second argument is buffer to which options 983 * will be moved, and return value is their length. 984 * XXX should be deleted; last arg currently ignored. 985 */ 986void 987ip_stripoptions(m, mopt) 988 register struct mbuf *m; 989 struct mbuf *mopt; 990{ 991 register int i; 992 struct ip *ip = mtod(m, struct ip *); 993 register caddr_t opts; 994 int olen; 995 996 olen = (ip->ip_hl<<2) - sizeof (struct ip); 997 opts = (caddr_t)(ip + 1); 998 i = m->m_len - (sizeof (struct ip) + olen); 999 bcopy(opts + olen, opts, (unsigned)i); 1000 m->m_len -= olen; 1001 if (m->m_flags & M_PKTHDR) 1002 m->m_pkthdr.len -= olen; 1003 ip->ip_hl = sizeof(struct ip) >> 2; 1004} 1005 1006int inetctlerrmap[PRC_NCMDS] = { 1007 0, 0, 0, 0, 1008 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1009 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1010 EMSGSIZE, EHOSTUNREACH, 0, 0, 1011 0, 0, 0, 0, 1012 ENOPROTOOPT 1013}; 1014 1015/* 1016 * Forward a packet. If some error occurs return the sender 1017 * an icmp packet. Note we can't always generate a meaningful 1018 * icmp message because icmp doesn't have a large enough repertoire 1019 * of codes and types. 1020 * 1021 * If not forwarding, just drop the packet. This could be confusing 1022 * if ipforwarding was zero but some routing protocol was advancing 1023 * us as a gateway to somewhere. However, we must let the routing 1024 * protocol deal with that. 1025 * 1026 * The srcrt parameter indicates whether the packet is being forwarded 1027 * via a source route. 1028 */ 1029void 1030ip_forward(m, srcrt) 1031 struct mbuf *m; 1032 int srcrt; 1033{ 1034 register struct ip *ip = mtod(m, struct ip *); 1035 register struct sockaddr_in *sin; 1036 register struct rtentry *rt; 1037 int error, type = 0, code = 0; 1038 struct mbuf *mcopy; 1039 n_long dest; 1040 struct ifnet *destifp; 1041 1042 dest = 0; 1043#ifdef DIAGNOSTIC 1044 if (ipprintfs) 1045 printf("forward: src %x dst %x ttl %x\n", 1046 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1047#endif 1048 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1049 ipstat.ips_cantforward++; 1050 m_freem(m); 1051 return; 1052 } 1053 HTONS(ip->ip_id); 1054 if (ip->ip_ttl <= IPTTLDEC) { 1055 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1056 return; 1057 } 1058 ip->ip_ttl -= IPTTLDEC; 1059 1060 sin = satosin(&ipforward_rt.ro_dst); 1061 if ((rt = ipforward_rt.ro_rt) == 0 || 1062 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1063 if (ipforward_rt.ro_rt) { 1064 RTFREE(ipforward_rt.ro_rt); 1065 ipforward_rt.ro_rt = 0; 1066 } 1067 sin->sin_family = AF_INET; 1068 sin->sin_len = sizeof(struct sockaddr_in); 1069 sin->sin_addr = ip->ip_dst; 1070 1071 rtalloc(&ipforward_rt); 1072 if (ipforward_rt.ro_rt == 0) { 1073 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1074 return; 1075 } 1076 rt = ipforward_rt.ro_rt; 1077 } 1078 1079 /* 1080 * Save at most 68 bytes of the packet in case 1081 * we need to generate an ICMP message to the src. 1082 */ 1083 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1084 1085 /* 1086 * If forwarding packet using same interface that it came in on, 1087 * perhaps should send a redirect to sender to shortcut a hop. 1088 * Only send redirect if source is sending directly to us, 1089 * and if packet was not source routed (or has any options). 1090 * Also, don't send redirect if forwarding using a default route 1091 * or a route modified by a redirect. 1092 */ 1093 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1094 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1095 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1096 ipsendredirects && !srcrt) { 1097 if (rt->rt_ifa && 1098 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1099 ifatoia(rt->rt_ifa)->ia_subnet) { 1100 if (rt->rt_flags & RTF_GATEWAY) 1101 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1102 else 1103 dest = ip->ip_dst.s_addr; 1104 /* Router requirements says to only send host redirects */ 1105 type = ICMP_REDIRECT; 1106 code = ICMP_REDIRECT_HOST; 1107#ifdef DIAGNOSTIC 1108 if (ipprintfs) 1109 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1110#endif 1111 } 1112 } 1113 1114 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1115 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1116 if (error) 1117 ipstat.ips_cantforward++; 1118 else { 1119 ipstat.ips_forward++; 1120 if (type) 1121 ipstat.ips_redirectsent++; 1122 else { 1123 if (mcopy) 1124 m_freem(mcopy); 1125 return; 1126 } 1127 } 1128 if (mcopy == NULL) 1129 return; 1130 destifp = NULL; 1131 1132 switch (error) { 1133 1134 case 0: /* forwarded, but need redirect */ 1135 /* type, code set above */ 1136 break; 1137 1138 case ENETUNREACH: /* shouldn't happen, checked above */ 1139 case EHOSTUNREACH: 1140 case ENETDOWN: 1141 case EHOSTDOWN: 1142 default: 1143 type = ICMP_UNREACH; 1144 code = ICMP_UNREACH_HOST; 1145 break; 1146 1147 case EMSGSIZE: 1148 type = ICMP_UNREACH; 1149 code = ICMP_UNREACH_NEEDFRAG; 1150 if (ipforward_rt.ro_rt) 1151 destifp = ipforward_rt.ro_rt->rt_ifp; 1152 ipstat.ips_cantfrag++; 1153 break; 1154 1155 case ENOBUFS: 1156 type = ICMP_SOURCEQUENCH; 1157 code = 0; 1158 break; 1159 } 1160 icmp_error(mcopy, type, code, dest, destifp); 1161} 1162 1163int 1164ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1165 int *name; 1166 u_int namelen; 1167 void *oldp; 1168 size_t *oldlenp; 1169 void *newp; 1170 size_t newlen; 1171{ 1172 /* All sysctl names at this level are terminal. */ 1173 if (namelen != 1) 1174 return (ENOTDIR); 1175 1176 switch (name[0]) { 1177 case IPCTL_FORWARDING: 1178 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1179 case IPCTL_SENDREDIRECTS: 1180 return (sysctl_int(oldp, oldlenp, newp, newlen, 1181 &ipsendredirects)); 1182 case IPCTL_DEFTTL: 1183 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1184#ifdef notyet 1185 case IPCTL_DEFMTU: 1186 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1187#endif 1188 case IPCTL_FORWSRCRT: 1189 /* 1190 * Don't allow this to change in a secure environment. 1191 */ 1192 if (securelevel > 0) 1193 return (EPERM); 1194 return (sysctl_int(oldp, oldlenp, newp, newlen, 1195 &ip_forwsrcrt)); 1196 case IPCTL_DIRECTEDBCAST: 1197 return (sysctl_int(oldp, oldlenp, newp, newlen, 1198 &ip_directedbcast)); 1199 default: 1200 return (EOPNOTSUPP); 1201 } 1202 /* NOTREACHED */ 1203} 1204