ip_input.c revision 1.40
1/* $NetBSD: ip_input.c,v 1.40 1996/10/22 11:27:05 veego 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#ifdef PFIL_HOOKS 442 m0 = m; 443 for (pfh = pfil_hook_get(PFIL_BAD); pfh; pfh = pfh->pfil_link.le_next) 444 if (pfh->pfil_func) { 445 (void)pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 2, &m0); 446 ip = mtod(m = m0, struct ip *); 447 } 448#endif /* PFIL_HOOKS */ 449 m_freem(m); 450 goto next; 451} 452 453/* 454 * Take incoming datagram fragment and try to 455 * reassemble it into whole datagram. If a chain for 456 * reassembly of this datagram already exists, then it 457 * is given as fp; otherwise have to make a chain. 458 */ 459struct ip * 460ip_reass(ipqe, fp) 461 register struct ipqent *ipqe; 462 register struct ipq *fp; 463{ 464 register struct mbuf *m = dtom(ipqe->ipqe_ip); 465 register struct ipqent *nq, *p, *q; 466 struct ip *ip; 467 struct mbuf *t; 468 int hlen = ipqe->ipqe_ip->ip_hl << 2; 469 int i, next; 470 471 /* 472 * Presence of header sizes in mbufs 473 * would confuse code below. 474 */ 475 m->m_data += hlen; 476 m->m_len -= hlen; 477 478 /* 479 * If first fragment to arrive, create a reassembly queue. 480 */ 481 if (fp == 0) { 482 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 483 goto dropfrag; 484 fp = mtod(t, struct ipq *); 485 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 486 fp->ipq_ttl = IPFRAGTTL; 487 fp->ipq_p = ipqe->ipqe_ip->ip_p; 488 fp->ipq_id = ipqe->ipqe_ip->ip_id; 489 LIST_INIT(&fp->ipq_fragq); 490 fp->ipq_src = ipqe->ipqe_ip->ip_src; 491 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 492 p = NULL; 493 goto insert; 494 } 495 496 /* 497 * Find a segment which begins after this one does. 498 */ 499 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 500 p = q, q = q->ipqe_q.le_next) 501 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 502 break; 503 504 /* 505 * If there is a preceding segment, it may provide some of 506 * our data already. If so, drop the data from the incoming 507 * segment. If it provides all of our data, drop us. 508 */ 509 if (p != NULL) { 510 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 511 ipqe->ipqe_ip->ip_off; 512 if (i > 0) { 513 if (i >= ipqe->ipqe_ip->ip_len) 514 goto dropfrag; 515 m_adj(dtom(ipqe->ipqe_ip), i); 516 ipqe->ipqe_ip->ip_off += i; 517 ipqe->ipqe_ip->ip_len -= i; 518 } 519 } 520 521 /* 522 * While we overlap succeeding segments trim them or, 523 * if they are completely covered, dequeue them. 524 */ 525 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 526 q->ipqe_ip->ip_off; q = nq) { 527 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 528 q->ipqe_ip->ip_off; 529 if (i < q->ipqe_ip->ip_len) { 530 q->ipqe_ip->ip_len -= i; 531 q->ipqe_ip->ip_off += i; 532 m_adj(dtom(q->ipqe_ip), i); 533 break; 534 } 535 nq = q->ipqe_q.le_next; 536 m_freem(dtom(q->ipqe_ip)); 537 LIST_REMOVE(q, ipqe_q); 538 FREE(q, M_IPQ); 539 } 540 541insert: 542 /* 543 * Stick new segment in its place; 544 * check for complete reassembly. 545 */ 546 if (p == NULL) { 547 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 548 } else { 549 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 550 } 551 next = 0; 552 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 553 p = q, q = q->ipqe_q.le_next) { 554 if (q->ipqe_ip->ip_off != next) 555 return (0); 556 next += q->ipqe_ip->ip_len; 557 } 558 if (p->ipqe_mff) 559 return (0); 560 561 /* 562 * Reassembly is complete; concatenate fragments. 563 */ 564 q = fp->ipq_fragq.lh_first; 565 ip = q->ipqe_ip; 566 m = dtom(q->ipqe_ip); 567 t = m->m_next; 568 m->m_next = 0; 569 m_cat(m, t); 570 nq = q->ipqe_q.le_next; 571 FREE(q, M_IPQ); 572 for (q = nq; q != NULL; q = nq) { 573 t = dtom(q->ipqe_ip); 574 nq = q->ipqe_q.le_next; 575 FREE(q, M_IPQ); 576 m_cat(m, t); 577 } 578 579 /* 580 * Create header for new ip packet by 581 * modifying header of first packet; 582 * dequeue and discard fragment reassembly header. 583 * Make header visible. 584 */ 585 ip->ip_len = next; 586 ip->ip_src = fp->ipq_src; 587 ip->ip_dst = fp->ipq_dst; 588 LIST_REMOVE(fp, ipq_q); 589 (void) m_free(dtom(fp)); 590 m->m_len += (ip->ip_hl << 2); 591 m->m_data -= (ip->ip_hl << 2); 592 /* some debugging cruft by sklower, below, will go away soon */ 593 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 594 register int plen = 0; 595 for (t = m; m; m = m->m_next) 596 plen += m->m_len; 597 t->m_pkthdr.len = plen; 598 } 599 return (ip); 600 601dropfrag: 602 ipstat.ips_fragdropped++; 603 m_freem(m); 604 FREE(ipqe, M_IPQ); 605 return (0); 606} 607 608/* 609 * Free a fragment reassembly header and all 610 * associated datagrams. 611 */ 612void 613ip_freef(fp) 614 struct ipq *fp; 615{ 616 register struct ipqent *q, *p; 617 618 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 619 p = q->ipqe_q.le_next; 620 m_freem(dtom(q->ipqe_ip)); 621 LIST_REMOVE(q, ipqe_q); 622 FREE(q, M_IPQ); 623 } 624 LIST_REMOVE(fp, ipq_q); 625 (void) m_free(dtom(fp)); 626} 627 628/* 629 * IP timer processing; 630 * if a timer expires on a reassembly 631 * queue, discard it. 632 */ 633void 634ip_slowtimo() 635{ 636 register struct ipq *fp, *nfp; 637 int s = splsoftnet(); 638 639 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 640 nfp = fp->ipq_q.le_next; 641 if (--fp->ipq_ttl == 0) { 642 ipstat.ips_fragtimeout++; 643 ip_freef(fp); 644 } 645 } 646 splx(s); 647} 648 649/* 650 * Drain off all datagram fragments. 651 */ 652void 653ip_drain() 654{ 655 656 while (ipq.lh_first != NULL) { 657 ipstat.ips_fragdropped++; 658 ip_freef(ipq.lh_first); 659 } 660} 661 662/* 663 * Do option processing on a datagram, 664 * possibly discarding it if bad options are encountered, 665 * or forwarding it if source-routed. 666 * Returns 1 if packet has been forwarded/freed, 667 * 0 if the packet should be processed further. 668 */ 669int 670ip_dooptions(m) 671 struct mbuf *m; 672{ 673 register struct ip *ip = mtod(m, struct ip *); 674 register u_char *cp; 675 register struct ip_timestamp *ipt; 676 register struct in_ifaddr *ia; 677 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 678 struct in_addr *sin, dst; 679 n_time ntime; 680 681 dst = ip->ip_dst; 682 cp = (u_char *)(ip + 1); 683 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 684 for (; cnt > 0; cnt -= optlen, cp += optlen) { 685 opt = cp[IPOPT_OPTVAL]; 686 if (opt == IPOPT_EOL) 687 break; 688 if (opt == IPOPT_NOP) 689 optlen = 1; 690 else { 691 optlen = cp[IPOPT_OLEN]; 692 if (optlen <= 0 || optlen > cnt) { 693 code = &cp[IPOPT_OLEN] - (u_char *)ip; 694 goto bad; 695 } 696 } 697 switch (opt) { 698 699 default: 700 break; 701 702 /* 703 * Source routing with record. 704 * Find interface with current destination address. 705 * If none on this machine then drop if strictly routed, 706 * or do nothing if loosely routed. 707 * Record interface address and bring up next address 708 * component. If strictly routed make sure next 709 * address is on directly accessible net. 710 */ 711 case IPOPT_LSRR: 712 case IPOPT_SSRR: 713 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 714 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 715 goto bad; 716 } 717 ipaddr.sin_addr = ip->ip_dst; 718 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 719 if (ia == 0) { 720 if (opt == IPOPT_SSRR) { 721 type = ICMP_UNREACH; 722 code = ICMP_UNREACH_SRCFAIL; 723 goto bad; 724 } 725 /* 726 * Loose routing, and not at next destination 727 * yet; nothing to do except forward. 728 */ 729 break; 730 } 731 off--; /* 0 origin */ 732 if (off > optlen - sizeof(struct in_addr)) { 733 /* 734 * End of source route. Should be for us. 735 */ 736 save_rte(cp, ip->ip_src); 737 break; 738 } 739 /* 740 * locate outgoing interface 741 */ 742 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 743 sizeof(ipaddr.sin_addr)); 744 if (opt == IPOPT_SSRR) { 745#define INA struct in_ifaddr * 746#define SA struct sockaddr * 747 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 748 } else 749 ia = ip_rtaddr(ipaddr.sin_addr); 750 if (ia == 0) { 751 type = ICMP_UNREACH; 752 code = ICMP_UNREACH_SRCFAIL; 753 goto bad; 754 } 755 ip->ip_dst = ipaddr.sin_addr; 756 bcopy((caddr_t)&ia->ia_addr.sin_addr, 757 (caddr_t)(cp + off), sizeof(struct in_addr)); 758 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 759 /* 760 * Let ip_intr's mcast routing check handle mcast pkts 761 */ 762 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 763 break; 764 765 case IPOPT_RR: 766 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 767 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 768 goto bad; 769 } 770 /* 771 * If no space remains, ignore. 772 */ 773 off--; /* 0 origin */ 774 if (off > optlen - sizeof(struct in_addr)) 775 break; 776 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 777 sizeof(ipaddr.sin_addr)); 778 /* 779 * locate outgoing interface; if we're the destination, 780 * use the incoming interface (should be same). 781 */ 782 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 783 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 784 type = ICMP_UNREACH; 785 code = ICMP_UNREACH_HOST; 786 goto bad; 787 } 788 bcopy((caddr_t)&ia->ia_addr.sin_addr, 789 (caddr_t)(cp + off), sizeof(struct in_addr)); 790 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 791 break; 792 793 case IPOPT_TS: 794 code = cp - (u_char *)ip; 795 ipt = (struct ip_timestamp *)cp; 796 if (ipt->ipt_len < 5) 797 goto bad; 798 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 799 if (++ipt->ipt_oflw == 0) 800 goto bad; 801 break; 802 } 803 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 804 switch (ipt->ipt_flg) { 805 806 case IPOPT_TS_TSONLY: 807 break; 808 809 case IPOPT_TS_TSANDADDR: 810 if (ipt->ipt_ptr + sizeof(n_time) + 811 sizeof(struct in_addr) > ipt->ipt_len) 812 goto bad; 813 ipaddr.sin_addr = dst; 814 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 815 m->m_pkthdr.rcvif); 816 if (ia == 0) 817 continue; 818 bcopy((caddr_t)&ia->ia_addr.sin_addr, 819 (caddr_t)sin, sizeof(struct in_addr)); 820 ipt->ipt_ptr += sizeof(struct in_addr); 821 break; 822 823 case IPOPT_TS_PRESPEC: 824 if (ipt->ipt_ptr + sizeof(n_time) + 825 sizeof(struct in_addr) > ipt->ipt_len) 826 goto bad; 827 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 828 sizeof(struct in_addr)); 829 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 830 continue; 831 ipt->ipt_ptr += sizeof(struct in_addr); 832 break; 833 834 default: 835 goto bad; 836 } 837 ntime = iptime(); 838 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 839 sizeof(n_time)); 840 ipt->ipt_ptr += sizeof(n_time); 841 } 842 } 843 if (forward) { 844 if (ip_forwsrcrt == 0) { 845 type = ICMP_UNREACH; 846 code = ICMP_UNREACH_SRCFAIL; 847 goto bad; 848 } 849 ip_forward(m, 1); 850 return (1); 851 } 852 return (0); 853bad: 854 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 855 icmp_error(m, type, code, 0, 0); 856 ipstat.ips_badoptions++; 857 return (1); 858} 859 860/* 861 * Given address of next destination (final or next hop), 862 * return internet address info of interface to be used to get there. 863 */ 864struct in_ifaddr * 865ip_rtaddr(dst) 866 struct in_addr dst; 867{ 868 register struct sockaddr_in *sin; 869 870 sin = satosin(&ipforward_rt.ro_dst); 871 872 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 873 if (ipforward_rt.ro_rt) { 874 RTFREE(ipforward_rt.ro_rt); 875 ipforward_rt.ro_rt = 0; 876 } 877 sin->sin_family = AF_INET; 878 sin->sin_len = sizeof(*sin); 879 sin->sin_addr = dst; 880 881 rtalloc(&ipforward_rt); 882 } 883 if (ipforward_rt.ro_rt == 0) 884 return ((struct in_ifaddr *)0); 885 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 886} 887 888/* 889 * Save incoming source route for use in replies, 890 * to be picked up later by ip_srcroute if the receiver is interested. 891 */ 892void 893save_rte(option, dst) 894 u_char *option; 895 struct in_addr dst; 896{ 897 unsigned olen; 898 899 olen = option[IPOPT_OLEN]; 900#ifdef DIAGNOSTIC 901 if (ipprintfs) 902 printf("save_rte: olen %d\n", olen); 903#endif 904 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 905 return; 906 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 907 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 908 ip_srcrt.dst = dst; 909} 910 911/* 912 * Retrieve incoming source route for use in replies, 913 * in the same form used by setsockopt. 914 * The first hop is placed before the options, will be removed later. 915 */ 916struct mbuf * 917ip_srcroute() 918{ 919 register struct in_addr *p, *q; 920 register struct mbuf *m; 921 922 if (ip_nhops == 0) 923 return ((struct mbuf *)0); 924 m = m_get(M_DONTWAIT, MT_SOOPTS); 925 if (m == 0) 926 return ((struct mbuf *)0); 927 928#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 929 930 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 931 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 932 OPTSIZ; 933#ifdef DIAGNOSTIC 934 if (ipprintfs) 935 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 936#endif 937 938 /* 939 * First save first hop for return route 940 */ 941 p = &ip_srcrt.route[ip_nhops - 1]; 942 *(mtod(m, struct in_addr *)) = *p--; 943#ifdef DIAGNOSTIC 944 if (ipprintfs) 945 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 946#endif 947 948 /* 949 * Copy option fields and padding (nop) to mbuf. 950 */ 951 ip_srcrt.nop = IPOPT_NOP; 952 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 953 bcopy((caddr_t)&ip_srcrt.nop, 954 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 955 q = (struct in_addr *)(mtod(m, caddr_t) + 956 sizeof(struct in_addr) + OPTSIZ); 957#undef OPTSIZ 958 /* 959 * Record return path as an IP source route, 960 * reversing the path (pointers are now aligned). 961 */ 962 while (p >= ip_srcrt.route) { 963#ifdef DIAGNOSTIC 964 if (ipprintfs) 965 printf(" %x", ntohl(q->s_addr)); 966#endif 967 *q++ = *p--; 968 } 969 /* 970 * Last hop goes to final destination. 971 */ 972 *q = ip_srcrt.dst; 973#ifdef DIAGNOSTIC 974 if (ipprintfs) 975 printf(" %x\n", ntohl(q->s_addr)); 976#endif 977 return (m); 978} 979 980/* 981 * Strip out IP options, at higher 982 * level protocol in the kernel. 983 * Second argument is buffer to which options 984 * will be moved, and return value is their length. 985 * XXX should be deleted; last arg currently ignored. 986 */ 987void 988ip_stripoptions(m, mopt) 989 register struct mbuf *m; 990 struct mbuf *mopt; 991{ 992 register int i; 993 struct ip *ip = mtod(m, struct ip *); 994 register caddr_t opts; 995 int olen; 996 997 olen = (ip->ip_hl<<2) - sizeof (struct ip); 998 opts = (caddr_t)(ip + 1); 999 i = m->m_len - (sizeof (struct ip) + olen); 1000 bcopy(opts + olen, opts, (unsigned)i); 1001 m->m_len -= olen; 1002 if (m->m_flags & M_PKTHDR) 1003 m->m_pkthdr.len -= olen; 1004 ip->ip_hl = sizeof(struct ip) >> 2; 1005} 1006 1007int inetctlerrmap[PRC_NCMDS] = { 1008 0, 0, 0, 0, 1009 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1010 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1011 EMSGSIZE, EHOSTUNREACH, 0, 0, 1012 0, 0, 0, 0, 1013 ENOPROTOOPT 1014}; 1015 1016/* 1017 * Forward a packet. If some error occurs return the sender 1018 * an icmp packet. Note we can't always generate a meaningful 1019 * icmp message because icmp doesn't have a large enough repertoire 1020 * of codes and types. 1021 * 1022 * If not forwarding, just drop the packet. This could be confusing 1023 * if ipforwarding was zero but some routing protocol was advancing 1024 * us as a gateway to somewhere. However, we must let the routing 1025 * protocol deal with that. 1026 * 1027 * The srcrt parameter indicates whether the packet is being forwarded 1028 * via a source route. 1029 */ 1030void 1031ip_forward(m, srcrt) 1032 struct mbuf *m; 1033 int srcrt; 1034{ 1035 register struct ip *ip = mtod(m, struct ip *); 1036 register struct sockaddr_in *sin; 1037 register struct rtentry *rt; 1038 int error, type = 0, code = 0; 1039 struct mbuf *mcopy; 1040 n_long dest; 1041 struct ifnet *destifp; 1042 1043 dest = 0; 1044#ifdef DIAGNOSTIC 1045 if (ipprintfs) 1046 printf("forward: src %x dst %x ttl %x\n", 1047 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1048#endif 1049 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1050 ipstat.ips_cantforward++; 1051 m_freem(m); 1052 return; 1053 } 1054 HTONS(ip->ip_id); 1055 if (ip->ip_ttl <= IPTTLDEC) { 1056 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1057 return; 1058 } 1059 ip->ip_ttl -= IPTTLDEC; 1060 1061 sin = satosin(&ipforward_rt.ro_dst); 1062 if ((rt = ipforward_rt.ro_rt) == 0 || 1063 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1064 if (ipforward_rt.ro_rt) { 1065 RTFREE(ipforward_rt.ro_rt); 1066 ipforward_rt.ro_rt = 0; 1067 } 1068 sin->sin_family = AF_INET; 1069 sin->sin_len = sizeof(struct sockaddr_in); 1070 sin->sin_addr = ip->ip_dst; 1071 1072 rtalloc(&ipforward_rt); 1073 if (ipforward_rt.ro_rt == 0) { 1074 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1075 return; 1076 } 1077 rt = ipforward_rt.ro_rt; 1078 } 1079 1080 /* 1081 * Save at most 68 bytes of the packet in case 1082 * we need to generate an ICMP message to the src. 1083 */ 1084 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1085 1086 /* 1087 * If forwarding packet using same interface that it came in on, 1088 * perhaps should send a redirect to sender to shortcut a hop. 1089 * Only send redirect if source is sending directly to us, 1090 * and if packet was not source routed (or has any options). 1091 * Also, don't send redirect if forwarding using a default route 1092 * or a route modified by a redirect. 1093 */ 1094 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1095 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1096 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1097 ipsendredirects && !srcrt) { 1098 if (rt->rt_ifa && 1099 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1100 ifatoia(rt->rt_ifa)->ia_subnet) { 1101 if (rt->rt_flags & RTF_GATEWAY) 1102 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1103 else 1104 dest = ip->ip_dst.s_addr; 1105 /* Router requirements says to only send host redirects */ 1106 type = ICMP_REDIRECT; 1107 code = ICMP_REDIRECT_HOST; 1108#ifdef DIAGNOSTIC 1109 if (ipprintfs) 1110 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1111#endif 1112 } 1113 } 1114 1115 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1116 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1117 if (error) 1118 ipstat.ips_cantforward++; 1119 else { 1120 ipstat.ips_forward++; 1121 if (type) 1122 ipstat.ips_redirectsent++; 1123 else { 1124 if (mcopy) 1125 m_freem(mcopy); 1126 return; 1127 } 1128 } 1129 if (mcopy == NULL) 1130 return; 1131 destifp = NULL; 1132 1133 switch (error) { 1134 1135 case 0: /* forwarded, but need redirect */ 1136 /* type, code set above */ 1137 break; 1138 1139 case ENETUNREACH: /* shouldn't happen, checked above */ 1140 case EHOSTUNREACH: 1141 case ENETDOWN: 1142 case EHOSTDOWN: 1143 default: 1144 type = ICMP_UNREACH; 1145 code = ICMP_UNREACH_HOST; 1146 break; 1147 1148 case EMSGSIZE: 1149 type = ICMP_UNREACH; 1150 code = ICMP_UNREACH_NEEDFRAG; 1151 if (ipforward_rt.ro_rt) 1152 destifp = ipforward_rt.ro_rt->rt_ifp; 1153 ipstat.ips_cantfrag++; 1154 break; 1155 1156 case ENOBUFS: 1157 type = ICMP_SOURCEQUENCH; 1158 code = 0; 1159 break; 1160 } 1161 icmp_error(mcopy, type, code, dest, destifp); 1162} 1163 1164int 1165ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1166 int *name; 1167 u_int namelen; 1168 void *oldp; 1169 size_t *oldlenp; 1170 void *newp; 1171 size_t newlen; 1172{ 1173 /* All sysctl names at this level are terminal. */ 1174 if (namelen != 1) 1175 return (ENOTDIR); 1176 1177 switch (name[0]) { 1178 case IPCTL_FORWARDING: 1179 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1180 case IPCTL_SENDREDIRECTS: 1181 return (sysctl_int(oldp, oldlenp, newp, newlen, 1182 &ipsendredirects)); 1183 case IPCTL_DEFTTL: 1184 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1185#ifdef notyet 1186 case IPCTL_DEFMTU: 1187 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1188#endif 1189 case IPCTL_FORWSRCRT: 1190 /* 1191 * Don't allow this to change in a secure environment. 1192 */ 1193 if (securelevel > 0) 1194 return (EPERM); 1195 return (sysctl_int(oldp, oldlenp, newp, newlen, 1196 &ip_forwsrcrt)); 1197 case IPCTL_DIRECTEDBCAST: 1198 return (sysctl_int(oldp, oldlenp, newp, newlen, 1199 &ip_directedbcast)); 1200 default: 1201 return (EOPNOTSUPP); 1202 } 1203 /* NOTREACHED */ 1204} 1205