ip_input.c revision 1.33
1/* $NetBSD: ip_input.c,v 1.33 1996/09/06 05:07:44 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#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 * IP initialization: fill in IP protocol switch table. 122 * All protocols not implemented in kernel go to raw IP protocol handler. 123 */ 124void 125ip_init() 126{ 127 register struct protosw *pr; 128 register int i; 129 130 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 131 if (pr == 0) 132 panic("ip_init"); 133 for (i = 0; i < IPPROTO_MAX; i++) 134 ip_protox[i] = pr - inetsw; 135 for (pr = inetdomain.dom_protosw; 136 pr < inetdomain.dom_protoswNPROTOSW; pr++) 137 if (pr->pr_domain->dom_family == PF_INET && 138 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 139 ip_protox[pr->pr_protocol] = pr - inetsw; 140 LIST_INIT(&ipq); 141 ip_id = time.tv_sec & 0xffff; 142 ipintrq.ifq_maxlen = ipqmaxlen; 143 TAILQ_INIT(&in_ifaddr); 144} 145 146struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 147struct route ipforward_rt; 148 149/* 150 * Ip input routine. Checksum and byte swap header. If fragmented 151 * try to reassemble. Process options. Pass to next level. 152 */ 153void 154ipintr() 155{ 156 register struct ip *ip = NULL; 157 register struct mbuf *m; 158 register struct ipq *fp; 159 register struct in_ifaddr *ia; 160 struct ipqent *ipqe; 161 int hlen = 0, mff, s; 162#ifdef PACKET_FILTER 163 struct packet_filter_hook *pfh; 164 struct mbuf *m0; 165#endif /* PACKET_FILTER */ 166 167next: 168 /* 169 * Get next datagram off input queue and get IP header 170 * in first mbuf. 171 */ 172 s = splimp(); 173 IF_DEQUEUE(&ipintrq, m); 174 splx(s); 175 if (m == 0) 176 return; 177#ifdef DIAGNOSTIC 178 if ((m->m_flags & M_PKTHDR) == 0) 179 panic("ipintr no HDR"); 180#endif 181 /* 182 * If no IP addresses have been set yet but the interfaces 183 * are receiving, can't do anything with incoming packets yet. 184 */ 185 if (in_ifaddr.tqh_first == 0) 186 goto bad; 187 ipstat.ips_total++; 188 if (m->m_len < sizeof (struct ip) && 189 (m = m_pullup(m, sizeof (struct ip))) == 0) { 190 ipstat.ips_toosmall++; 191 goto next; 192 } 193 ip = mtod(m, struct ip *); 194 if (ip->ip_v != IPVERSION) { 195 ipstat.ips_badvers++; 196 goto bad; 197 } 198 hlen = ip->ip_hl << 2; 199 if (hlen < sizeof(struct ip)) { /* minimum header length */ 200 ipstat.ips_badhlen++; 201 goto bad; 202 } 203 if (hlen > m->m_len) { 204 if ((m = m_pullup(m, hlen)) == 0) { 205 ipstat.ips_badhlen++; 206 goto next; 207 } 208 ip = mtod(m, struct ip *); 209 } 210 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 211 ipstat.ips_badsum++; 212 goto bad; 213 } 214 215 /* 216 * Convert fields to host representation. 217 */ 218 NTOHS(ip->ip_len); 219 if (ip->ip_len < hlen) { 220 ipstat.ips_badlen++; 221 goto bad; 222 } 223 NTOHS(ip->ip_id); 224 NTOHS(ip->ip_off); 225 226 /* 227 * Check that the amount of data in the buffers 228 * is as at least much as the IP header would have us expect. 229 * Trim mbufs if longer than we expect. 230 * Drop packet if shorter than we expect. 231 */ 232 if (m->m_pkthdr.len < ip->ip_len) { 233 ipstat.ips_tooshort++; 234 goto bad; 235 } 236 if (m->m_pkthdr.len > ip->ip_len) { 237 if (m->m_len == m->m_pkthdr.len) { 238 m->m_len = ip->ip_len; 239 m->m_pkthdr.len = ip->ip_len; 240 } else 241 m_adj(m, ip->ip_len - m->m_pkthdr.len); 242 } 243 244#ifdef PACKET_FILTER 245 /* 246 * Run through list of hooks for input packets. 247 */ 248 m0 = m; 249 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 250 if (pfh->pfil_func) { 251 if (pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0)) 252 goto bad; 253 ip = mtod(m = m0, struct ip *); 254 } 255#endif /* PACKET_FILTER */ 256 257 /* 258 * Process options and, if not destined for us, 259 * ship it on. ip_dooptions returns 1 when an 260 * error was detected (causing an icmp message 261 * to be sent and the original packet to be freed). 262 */ 263 ip_nhops = 0; /* for source routed packets */ 264 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 265 goto next; 266 267 /* 268 * Check our list of addresses, to see if the packet is for us. 269 */ 270 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 271 if (ip->ip_dst.s_addr == ia->ia_addr.sin_addr.s_addr) 272 goto ours; 273 if (((ip_directedbcast == 0) || (ip_directedbcast && 274 ia->ia_ifp == m->m_pkthdr.rcvif)) && 275 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 276 if (ip->ip_dst.s_addr == ia->ia_broadaddr.sin_addr.s_addr || 277 ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr || 278 /* 279 * Look for all-0's host part (old broadcast addr), 280 * either for subnet or net. 281 */ 282 ip->ip_dst.s_addr == ia->ia_subnet || 283 ip->ip_dst.s_addr == ia->ia_net) 284 goto ours; 285 } 286 } 287 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 288 struct in_multi *inm; 289#ifdef MROUTING 290 extern struct socket *ip_mrouter; 291 292 if (m->m_flags & M_EXT) { 293 if ((m = m_pullup(m, hlen)) == 0) { 294 ipstat.ips_toosmall++; 295 goto next; 296 } 297 ip = mtod(m, struct ip *); 298 } 299 300 if (ip_mrouter) { 301 /* 302 * If we are acting as a multicast router, all 303 * incoming multicast packets are passed to the 304 * kernel-level multicast forwarding function. 305 * The packet is returned (relatively) intact; if 306 * ip_mforward() returns a non-zero value, the packet 307 * must be discarded, else it may be accepted below. 308 * 309 * (The IP ident field is put in the same byte order 310 * as expected when ip_mforward() is called from 311 * ip_output().) 312 */ 313 ip->ip_id = htons(ip->ip_id); 314 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 315 ipstat.ips_cantforward++; 316 m_freem(m); 317 goto next; 318 } 319 ip->ip_id = ntohs(ip->ip_id); 320 321 /* 322 * The process-level routing demon needs to receive 323 * all multicast IGMP packets, whether or not this 324 * host belongs to their destination groups. 325 */ 326 if (ip->ip_p == IPPROTO_IGMP) 327 goto ours; 328 ipstat.ips_forward++; 329 } 330#endif 331 /* 332 * See if we belong to the destination multicast group on the 333 * arrival interface. 334 */ 335 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 336 if (inm == NULL) { 337 ipstat.ips_cantforward++; 338 m_freem(m); 339 goto next; 340 } 341 goto ours; 342 } 343 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 344 ip->ip_dst.s_addr == INADDR_ANY) 345 goto ours; 346 347 /* 348 * Not for us; forward if possible and desirable. 349 */ 350 if (ipforwarding == 0) { 351 ipstat.ips_cantforward++; 352 m_freem(m); 353 } else 354 ip_forward(m, 0); 355 goto next; 356 357ours: 358 /* 359 * If offset or IP_MF are set, must reassemble. 360 * Otherwise, nothing need be done. 361 * (We could look in the reassembly queue to see 362 * if the packet was previously fragmented, 363 * but it's not worth the time; just let them time out.) 364 */ 365 if (ip->ip_off &~ IP_DF) { 366 if (m->m_flags & M_EXT) { /* XXX */ 367 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 368 ipstat.ips_toosmall++; 369 goto next; 370 } 371 ip = mtod(m, struct ip *); 372 } 373 /* 374 * Look for queue of fragments 375 * of this datagram. 376 */ 377 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 378 if (ip->ip_id == fp->ipq_id && 379 ip->ip_src.s_addr == fp->ipq_src.s_addr && 380 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 381 ip->ip_p == fp->ipq_p) 382 goto found; 383 fp = 0; 384found: 385 386 /* 387 * Adjust ip_len to not reflect header, 388 * set ipqe_mff if more fragments are expected, 389 * convert offset of this to bytes. 390 */ 391 ip->ip_len -= hlen; 392 mff = (ip->ip_off & IP_MF) != 0; 393 if (mff) { 394 /* 395 * Make sure that fragments have a data length 396 * that's a non-zero multiple of 8 bytes. 397 */ 398 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 399 ipstat.ips_badfrags++; 400 goto bad; 401 } 402 } 403 ip->ip_off <<= 3; 404 405 /* 406 * If datagram marked as having more fragments 407 * or if this is not the first fragment, 408 * attempt reassembly; if it succeeds, proceed. 409 */ 410 if (mff || ip->ip_off) { 411 ipstat.ips_fragments++; 412 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 413 M_IPQ, M_NOWAIT); 414 if (ipqe == NULL) { 415 ipstat.ips_rcvmemdrop++; 416 goto bad; 417 } 418 ipqe->ipqe_mff = mff; 419 ipqe->ipqe_ip = ip; 420 ip = ip_reass(ipqe, fp); 421 if (ip == 0) 422 goto next; 423 ipstat.ips_reassembled++; 424 m = dtom(ip); 425 } else 426 if (fp) 427 ip_freef(fp); 428 } else 429 ip->ip_len -= hlen; 430 431 /* 432 * Switch out to protocol's input routine. 433 */ 434 ipstat.ips_delivered++; 435 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 436 goto next; 437bad: 438#ifdef PACKET_FILTER 439 m0 = m; 440 for (pfh = pfil_hook_get(PFIL_BAD); pfh; pfh = pfh->pfil_link.le_next) 441 if (pfh->pfil_func) { 442 (void)pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 2, &m0); 443 ip = mtod(m = m0, struct ip *); 444 } 445#endif /* PACKET_FILTER */ 446 m_freem(m); 447 goto next; 448} 449 450/* 451 * Take incoming datagram fragment and try to 452 * reassemble it into whole datagram. If a chain for 453 * reassembly of this datagram already exists, then it 454 * is given as fp; otherwise have to make a chain. 455 */ 456struct ip * 457ip_reass(ipqe, fp) 458 register struct ipqent *ipqe; 459 register struct ipq *fp; 460{ 461 register struct mbuf *m = dtom(ipqe->ipqe_ip); 462 register struct ipqent *nq, *p, *q; 463 struct ip *ip; 464 struct mbuf *t; 465 int hlen = ipqe->ipqe_ip->ip_hl << 2; 466 int i, next; 467 468 /* 469 * Presence of header sizes in mbufs 470 * would confuse code below. 471 */ 472 m->m_data += hlen; 473 m->m_len -= hlen; 474 475 /* 476 * If first fragment to arrive, create a reassembly queue. 477 */ 478 if (fp == 0) { 479 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 480 goto dropfrag; 481 fp = mtod(t, struct ipq *); 482 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 483 fp->ipq_ttl = IPFRAGTTL; 484 fp->ipq_p = ipqe->ipqe_ip->ip_p; 485 fp->ipq_id = ipqe->ipqe_ip->ip_id; 486 LIST_INIT(&fp->ipq_fragq); 487 fp->ipq_src = ipqe->ipqe_ip->ip_src; 488 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 489 p = NULL; 490 goto insert; 491 } 492 493 /* 494 * Find a segment which begins after this one does. 495 */ 496 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 497 p = q, q = q->ipqe_q.le_next) 498 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 499 break; 500 501 /* 502 * If there is a preceding segment, it may provide some of 503 * our data already. If so, drop the data from the incoming 504 * segment. If it provides all of our data, drop us. 505 */ 506 if (p != NULL) { 507 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 508 ipqe->ipqe_ip->ip_off; 509 if (i > 0) { 510 if (i >= ipqe->ipqe_ip->ip_len) 511 goto dropfrag; 512 m_adj(dtom(ipqe->ipqe_ip), i); 513 ipqe->ipqe_ip->ip_off += i; 514 ipqe->ipqe_ip->ip_len -= i; 515 } 516 } 517 518 /* 519 * While we overlap succeeding segments trim them or, 520 * if they are completely covered, dequeue them. 521 */ 522 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 523 q->ipqe_ip->ip_off; q = nq) { 524 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 525 q->ipqe_ip->ip_off; 526 if (i < q->ipqe_ip->ip_len) { 527 q->ipqe_ip->ip_len -= i; 528 q->ipqe_ip->ip_off += i; 529 m_adj(dtom(q->ipqe_ip), i); 530 break; 531 } 532 nq = q->ipqe_q.le_next; 533 m_freem(dtom(q->ipqe_ip)); 534 LIST_REMOVE(q, ipqe_q); 535 FREE(q, M_IPQ); 536 } 537 538insert: 539 /* 540 * Stick new segment in its place; 541 * check for complete reassembly. 542 */ 543 if (p == NULL) { 544 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 545 } else { 546 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 547 } 548 next = 0; 549 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 550 p = q, q = q->ipqe_q.le_next) { 551 if (q->ipqe_ip->ip_off != next) 552 return (0); 553 next += q->ipqe_ip->ip_len; 554 } 555 if (p->ipqe_mff) 556 return (0); 557 558 /* 559 * Reassembly is complete; concatenate fragments. 560 */ 561 q = fp->ipq_fragq.lh_first; 562 ip = q->ipqe_ip; 563 m = dtom(q->ipqe_ip); 564 t = m->m_next; 565 m->m_next = 0; 566 m_cat(m, t); 567 nq = q->ipqe_q.le_next; 568 FREE(q, M_IPQ); 569 for (q = nq; q != NULL; q = nq) { 570 t = dtom(q->ipqe_ip); 571 nq = q->ipqe_q.le_next; 572 FREE(q, M_IPQ); 573 m_cat(m, t); 574 } 575 576 /* 577 * Create header for new ip packet by 578 * modifying header of first packet; 579 * dequeue and discard fragment reassembly header. 580 * Make header visible. 581 */ 582 ip->ip_len = next; 583 ip->ip_src = fp->ipq_src; 584 ip->ip_dst = fp->ipq_dst; 585 LIST_REMOVE(fp, ipq_q); 586 (void) m_free(dtom(fp)); 587 m->m_len += (ip->ip_hl << 2); 588 m->m_data -= (ip->ip_hl << 2); 589 /* some debugging cruft by sklower, below, will go away soon */ 590 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 591 register int plen = 0; 592 for (t = m; m; m = m->m_next) 593 plen += m->m_len; 594 t->m_pkthdr.len = plen; 595 } 596 return (ip); 597 598dropfrag: 599 ipstat.ips_fragdropped++; 600 m_freem(m); 601 FREE(ipqe, M_IPQ); 602 return (0); 603} 604 605/* 606 * Free a fragment reassembly header and all 607 * associated datagrams. 608 */ 609void 610ip_freef(fp) 611 struct ipq *fp; 612{ 613 register struct ipqent *q, *p; 614 615 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 616 p = q->ipqe_q.le_next; 617 m_freem(dtom(q->ipqe_ip)); 618 LIST_REMOVE(q, ipqe_q); 619 FREE(q, M_IPQ); 620 } 621 LIST_REMOVE(fp, ipq_q); 622 (void) m_free(dtom(fp)); 623} 624 625/* 626 * IP timer processing; 627 * if a timer expires on a reassembly 628 * queue, discard it. 629 */ 630void 631ip_slowtimo() 632{ 633 register struct ipq *fp, *nfp; 634 int s = splsoftnet(); 635 636 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 637 nfp = fp->ipq_q.le_next; 638 if (--fp->ipq_ttl == 0) { 639 ipstat.ips_fragtimeout++; 640 ip_freef(fp); 641 } 642 } 643 splx(s); 644} 645 646/* 647 * Drain off all datagram fragments. 648 */ 649void 650ip_drain() 651{ 652 653 while (ipq.lh_first != NULL) { 654 ipstat.ips_fragdropped++; 655 ip_freef(ipq.lh_first); 656 } 657} 658 659/* 660 * Do option processing on a datagram, 661 * possibly discarding it if bad options are encountered, 662 * or forwarding it if source-routed. 663 * Returns 1 if packet has been forwarded/freed, 664 * 0 if the packet should be processed further. 665 */ 666int 667ip_dooptions(m) 668 struct mbuf *m; 669{ 670 register struct ip *ip = mtod(m, struct ip *); 671 register u_char *cp; 672 register struct ip_timestamp *ipt; 673 register struct in_ifaddr *ia; 674 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 675 struct in_addr *sin, dst; 676 n_time ntime; 677 678 dst = ip->ip_dst; 679 cp = (u_char *)(ip + 1); 680 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 681 for (; cnt > 0; cnt -= optlen, cp += optlen) { 682 opt = cp[IPOPT_OPTVAL]; 683 if (opt == IPOPT_EOL) 684 break; 685 if (opt == IPOPT_NOP) 686 optlen = 1; 687 else { 688 optlen = cp[IPOPT_OLEN]; 689 if (optlen <= 0 || optlen > cnt) { 690 code = &cp[IPOPT_OLEN] - (u_char *)ip; 691 goto bad; 692 } 693 } 694 switch (opt) { 695 696 default: 697 break; 698 699 /* 700 * Source routing with record. 701 * Find interface with current destination address. 702 * If none on this machine then drop if strictly routed, 703 * or do nothing if loosely routed. 704 * Record interface address and bring up next address 705 * component. If strictly routed make sure next 706 * address is on directly accessible net. 707 */ 708 case IPOPT_LSRR: 709 case IPOPT_SSRR: 710 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 711 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 712 goto bad; 713 } 714 ipaddr.sin_addr = ip->ip_dst; 715 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 716 if (ia == 0) { 717 if (opt == IPOPT_SSRR) { 718 type = ICMP_UNREACH; 719 code = ICMP_UNREACH_SRCFAIL; 720 goto bad; 721 } 722 /* 723 * Loose routing, and not at next destination 724 * yet; nothing to do except forward. 725 */ 726 break; 727 } 728 off--; /* 0 origin */ 729 if (off > optlen - sizeof(struct in_addr)) { 730 /* 731 * End of source route. Should be for us. 732 */ 733 save_rte(cp, ip->ip_src); 734 break; 735 } 736 /* 737 * locate outgoing interface 738 */ 739 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 740 sizeof(ipaddr.sin_addr)); 741 if (opt == IPOPT_SSRR) { 742#define INA struct in_ifaddr * 743#define SA struct sockaddr * 744 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 745 } else 746 ia = ip_rtaddr(ipaddr.sin_addr); 747 if (ia == 0) { 748 type = ICMP_UNREACH; 749 code = ICMP_UNREACH_SRCFAIL; 750 goto bad; 751 } 752 ip->ip_dst = ipaddr.sin_addr; 753 bcopy((caddr_t)&ia->ia_addr.sin_addr, 754 (caddr_t)(cp + off), sizeof(struct in_addr)); 755 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 756 /* 757 * Let ip_intr's mcast routing check handle mcast pkts 758 */ 759 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 760 break; 761 762 case IPOPT_RR: 763 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 764 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 765 goto bad; 766 } 767 /* 768 * If no space remains, ignore. 769 */ 770 off--; /* 0 origin */ 771 if (off > optlen - sizeof(struct in_addr)) 772 break; 773 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 774 sizeof(ipaddr.sin_addr)); 775 /* 776 * locate outgoing interface; if we're the destination, 777 * use the incoming interface (should be same). 778 */ 779 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 780 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 781 type = ICMP_UNREACH; 782 code = ICMP_UNREACH_HOST; 783 goto bad; 784 } 785 bcopy((caddr_t)&ia->ia_addr.sin_addr, 786 (caddr_t)(cp + off), sizeof(struct in_addr)); 787 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 788 break; 789 790 case IPOPT_TS: 791 code = cp - (u_char *)ip; 792 ipt = (struct ip_timestamp *)cp; 793 if (ipt->ipt_len < 5) 794 goto bad; 795 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 796 if (++ipt->ipt_oflw == 0) 797 goto bad; 798 break; 799 } 800 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 801 switch (ipt->ipt_flg) { 802 803 case IPOPT_TS_TSONLY: 804 break; 805 806 case IPOPT_TS_TSANDADDR: 807 if (ipt->ipt_ptr + sizeof(n_time) + 808 sizeof(struct in_addr) > ipt->ipt_len) 809 goto bad; 810 ipaddr.sin_addr = dst; 811 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 812 m->m_pkthdr.rcvif); 813 if (ia == 0) 814 continue; 815 bcopy((caddr_t)&ia->ia_addr.sin_addr, 816 (caddr_t)sin, sizeof(struct in_addr)); 817 ipt->ipt_ptr += sizeof(struct in_addr); 818 break; 819 820 case IPOPT_TS_PRESPEC: 821 if (ipt->ipt_ptr + sizeof(n_time) + 822 sizeof(struct in_addr) > ipt->ipt_len) 823 goto bad; 824 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 825 sizeof(struct in_addr)); 826 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 827 continue; 828 ipt->ipt_ptr += sizeof(struct in_addr); 829 break; 830 831 default: 832 goto bad; 833 } 834 ntime = iptime(); 835 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 836 sizeof(n_time)); 837 ipt->ipt_ptr += sizeof(n_time); 838 } 839 } 840 if (forward) { 841 if (ip_forwsrcrt == 0) { 842 type = ICMP_UNREACH; 843 code = ICMP_UNREACH_SRCFAIL; 844 goto bad; 845 } 846 ip_forward(m, 1); 847 return (1); 848 } 849 return (0); 850bad: 851 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 852 icmp_error(m, type, code, 0, 0); 853 ipstat.ips_badoptions++; 854 return (1); 855} 856 857/* 858 * Given address of next destination (final or next hop), 859 * return internet address info of interface to be used to get there. 860 */ 861struct in_ifaddr * 862ip_rtaddr(dst) 863 struct in_addr dst; 864{ 865 register struct sockaddr_in *sin; 866 867 sin = satosin(&ipforward_rt.ro_dst); 868 869 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 870 if (ipforward_rt.ro_rt) { 871 RTFREE(ipforward_rt.ro_rt); 872 ipforward_rt.ro_rt = 0; 873 } 874 sin->sin_family = AF_INET; 875 sin->sin_len = sizeof(*sin); 876 sin->sin_addr = dst; 877 878 rtalloc(&ipforward_rt); 879 } 880 if (ipforward_rt.ro_rt == 0) 881 return ((struct in_ifaddr *)0); 882 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 883} 884 885/* 886 * Save incoming source route for use in replies, 887 * to be picked up later by ip_srcroute if the receiver is interested. 888 */ 889void 890save_rte(option, dst) 891 u_char *option; 892 struct in_addr dst; 893{ 894 unsigned olen; 895 896 olen = option[IPOPT_OLEN]; 897#ifdef DIAGNOSTIC 898 if (ipprintfs) 899 printf("save_rte: olen %d\n", olen); 900#endif 901 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 902 return; 903 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 904 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 905 ip_srcrt.dst = dst; 906} 907 908/* 909 * Retrieve incoming source route for use in replies, 910 * in the same form used by setsockopt. 911 * The first hop is placed before the options, will be removed later. 912 */ 913struct mbuf * 914ip_srcroute() 915{ 916 register struct in_addr *p, *q; 917 register struct mbuf *m; 918 919 if (ip_nhops == 0) 920 return ((struct mbuf *)0); 921 m = m_get(M_DONTWAIT, MT_SOOPTS); 922 if (m == 0) 923 return ((struct mbuf *)0); 924 925#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 926 927 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 928 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 929 OPTSIZ; 930#ifdef DIAGNOSTIC 931 if (ipprintfs) 932 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 933#endif 934 935 /* 936 * First save first hop for return route 937 */ 938 p = &ip_srcrt.route[ip_nhops - 1]; 939 *(mtod(m, struct in_addr *)) = *p--; 940#ifdef DIAGNOSTIC 941 if (ipprintfs) 942 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 943#endif 944 945 /* 946 * Copy option fields and padding (nop) to mbuf. 947 */ 948 ip_srcrt.nop = IPOPT_NOP; 949 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 950 bcopy((caddr_t)&ip_srcrt.nop, 951 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 952 q = (struct in_addr *)(mtod(m, caddr_t) + 953 sizeof(struct in_addr) + OPTSIZ); 954#undef OPTSIZ 955 /* 956 * Record return path as an IP source route, 957 * reversing the path (pointers are now aligned). 958 */ 959 while (p >= ip_srcrt.route) { 960#ifdef DIAGNOSTIC 961 if (ipprintfs) 962 printf(" %x", ntohl(q->s_addr)); 963#endif 964 *q++ = *p--; 965 } 966 /* 967 * Last hop goes to final destination. 968 */ 969 *q = ip_srcrt.dst; 970#ifdef DIAGNOSTIC 971 if (ipprintfs) 972 printf(" %x\n", ntohl(q->s_addr)); 973#endif 974 return (m); 975} 976 977/* 978 * Strip out IP options, at higher 979 * level protocol in the kernel. 980 * Second argument is buffer to which options 981 * will be moved, and return value is their length. 982 * XXX should be deleted; last arg currently ignored. 983 */ 984void 985ip_stripoptions(m, mopt) 986 register struct mbuf *m; 987 struct mbuf *mopt; 988{ 989 register int i; 990 struct ip *ip = mtod(m, struct ip *); 991 register caddr_t opts; 992 int olen; 993 994 olen = (ip->ip_hl<<2) - sizeof (struct ip); 995 opts = (caddr_t)(ip + 1); 996 i = m->m_len - (sizeof (struct ip) + olen); 997 bcopy(opts + olen, opts, (unsigned)i); 998 m->m_len -= olen; 999 if (m->m_flags & M_PKTHDR) 1000 m->m_pkthdr.len -= olen; 1001 ip->ip_hl = sizeof(struct ip) >> 2; 1002} 1003 1004int inetctlerrmap[PRC_NCMDS] = { 1005 0, 0, 0, 0, 1006 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1007 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1008 EMSGSIZE, EHOSTUNREACH, 0, 0, 1009 0, 0, 0, 0, 1010 ENOPROTOOPT 1011}; 1012 1013/* 1014 * Forward a packet. If some error occurs return the sender 1015 * an icmp packet. Note we can't always generate a meaningful 1016 * icmp message because icmp doesn't have a large enough repertoire 1017 * of codes and types. 1018 * 1019 * If not forwarding, just drop the packet. This could be confusing 1020 * if ipforwarding was zero but some routing protocol was advancing 1021 * us as a gateway to somewhere. However, we must let the routing 1022 * protocol deal with that. 1023 * 1024 * The srcrt parameter indicates whether the packet is being forwarded 1025 * via a source route. 1026 */ 1027void 1028ip_forward(m, srcrt) 1029 struct mbuf *m; 1030 int srcrt; 1031{ 1032 register struct ip *ip = mtod(m, struct ip *); 1033 register struct sockaddr_in *sin; 1034 register struct rtentry *rt; 1035 int error, type = 0, code = 0; 1036 struct mbuf *mcopy; 1037 n_long dest; 1038 struct ifnet *destifp; 1039 1040 dest = 0; 1041#ifdef DIAGNOSTIC 1042 if (ipprintfs) 1043 printf("forward: src %x dst %x ttl %x\n", 1044 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1045#endif 1046 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1047 ipstat.ips_cantforward++; 1048 m_freem(m); 1049 return; 1050 } 1051 HTONS(ip->ip_id); 1052 if (ip->ip_ttl <= IPTTLDEC) { 1053 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1054 return; 1055 } 1056 ip->ip_ttl -= IPTTLDEC; 1057 1058 sin = satosin(&ipforward_rt.ro_dst); 1059 if ((rt = ipforward_rt.ro_rt) == 0 || 1060 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1061 if (ipforward_rt.ro_rt) { 1062 RTFREE(ipforward_rt.ro_rt); 1063 ipforward_rt.ro_rt = 0; 1064 } 1065 sin->sin_family = AF_INET; 1066 sin->sin_len = sizeof(*sin); 1067 sin->sin_addr = ip->ip_dst; 1068 1069 rtalloc(&ipforward_rt); 1070 if (ipforward_rt.ro_rt == 0) { 1071 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1072 return; 1073 } 1074 rt = ipforward_rt.ro_rt; 1075 } 1076 1077 /* 1078 * Save at most 64 bytes of the packet in case 1079 * we need to generate an ICMP message to the src. 1080 */ 1081 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1082 1083 /* 1084 * If forwarding packet using same interface that it came in on, 1085 * perhaps should send a redirect to sender to shortcut a hop. 1086 * Only send redirect if source is sending directly to us, 1087 * and if packet was not source routed (or has any options). 1088 * Also, don't send redirect if forwarding using a default route 1089 * or a route modified by a redirect. 1090 */ 1091 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1092 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1093 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1094 ipsendredirects && !srcrt) { 1095 if (rt->rt_ifa && 1096 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1097 ifatoia(rt->rt_ifa)->ia_subnet) { 1098 if (rt->rt_flags & RTF_GATEWAY) 1099 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1100 else 1101 dest = ip->ip_dst.s_addr; 1102 /* Router requirements says to only send host redirects */ 1103 type = ICMP_REDIRECT; 1104 code = ICMP_REDIRECT_HOST; 1105#ifdef DIAGNOSTIC 1106 if (ipprintfs) 1107 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1108#endif 1109 } 1110 } 1111 1112 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1113 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1114 if (error) 1115 ipstat.ips_cantforward++; 1116 else { 1117 ipstat.ips_forward++; 1118 if (type) 1119 ipstat.ips_redirectsent++; 1120 else { 1121 if (mcopy) 1122 m_freem(mcopy); 1123 return; 1124 } 1125 } 1126 if (mcopy == NULL) 1127 return; 1128 destifp = NULL; 1129 1130 switch (error) { 1131 1132 case 0: /* forwarded, but need redirect */ 1133 /* type, code set above */ 1134 break; 1135 1136 case ENETUNREACH: /* shouldn't happen, checked above */ 1137 case EHOSTUNREACH: 1138 case ENETDOWN: 1139 case EHOSTDOWN: 1140 default: 1141 type = ICMP_UNREACH; 1142 code = ICMP_UNREACH_HOST; 1143 break; 1144 1145 case EMSGSIZE: 1146 type = ICMP_UNREACH; 1147 code = ICMP_UNREACH_NEEDFRAG; 1148 if (ipforward_rt.ro_rt) 1149 destifp = ipforward_rt.ro_rt->rt_ifp; 1150 ipstat.ips_cantfrag++; 1151 break; 1152 1153 case ENOBUFS: 1154 type = ICMP_SOURCEQUENCH; 1155 code = 0; 1156 break; 1157 } 1158 icmp_error(mcopy, type, code, dest, destifp); 1159} 1160 1161int 1162ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1163 int *name; 1164 u_int namelen; 1165 void *oldp; 1166 size_t *oldlenp; 1167 void *newp; 1168 size_t newlen; 1169{ 1170 /* All sysctl names at this level are terminal. */ 1171 if (namelen != 1) 1172 return (ENOTDIR); 1173 1174 switch (name[0]) { 1175 case IPCTL_FORWARDING: 1176 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1177 case IPCTL_SENDREDIRECTS: 1178 return (sysctl_int(oldp, oldlenp, newp, newlen, 1179 &ipsendredirects)); 1180 case IPCTL_DEFTTL: 1181 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1182#ifdef notyet 1183 case IPCTL_DEFMTU: 1184 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1185#endif 1186 case IPCTL_FORWSRCRT: 1187 /* 1188 * Don't allow this to change in a secure environment. 1189 */ 1190 if (securelevel > 0) 1191 return (EPERM); 1192 return (sysctl_int(oldp, oldlenp, newp, newlen, 1193 &ip_forwsrcrt)); 1194 case IPCTL_DIRECTEDBCAST: 1195 return (sysctl_int(oldp, oldlenp, newp, newlen, 1196 &ip_directedbcast)); 1197 default: 1198 return (EOPNOTSUPP); 1199 } 1200 /* NOTREACHED */ 1201} 1202