ip_input.c revision 1.27
1/* $NetBSD: ip_input.c,v 1.27 1996/01/16 04:17:37 thorpej 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 49#include <net/if.h> 50#include <net/route.h> 51 52#include <netinet/in.h> 53#include <netinet/in_systm.h> 54#include <netinet/ip.h> 55#include <netinet/in_pcb.h> 56#include <netinet/in_var.h> 57#include <netinet/ip_var.h> 58#include <netinet/ip_icmp.h> 59 60#ifndef IPFORWARDING 61#ifdef GATEWAY 62#define IPFORWARDING 1 /* forward IP packets not for us */ 63#else /* GATEWAY */ 64#define IPFORWARDING 0 /* don't forward IP packets not for us */ 65#endif /* GATEWAY */ 66#endif /* IPFORWARDING */ 67#ifndef IPSENDREDIRECTS 68#define IPSENDREDIRECTS 1 69#endif 70#ifndef IPFORWSRCRT 71#define IPFORWSRCRT 1 /* allow source-routed packets */ 72#endif 73/* 74 * Note: DIRECTED_BROADCAST is handled this way so that previous 75 * configuration using this option will Just Work. 76 */ 77#ifndef IPDIRECTEDBCAST 78#ifdef DIRECTED_BROADCAST 79#define IPDIRECTEDBCAST 1 80#else 81#define IPDIRECTEDBCAST 0 82#endif /* DIRECTED_BROADCAST */ 83#endif /* IPDIRECTEDBCAST */ 84int ipforwarding = IPFORWARDING; 85int ipsendredirects = IPSENDREDIRECTS; 86int ip_defttl = IPDEFTTL; 87int ip_forwsrcrt = IPFORWSRCRT; 88int ip_directedbcast = IPDIRECTEDBCAST; 89#ifdef DIAGNOSTIC 90int ipprintfs = 0; 91#endif 92 93extern struct domain inetdomain; 94extern struct protosw inetsw[]; 95u_char ip_protox[IPPROTO_MAX]; 96int ipqmaxlen = IFQ_MAXLEN; 97struct in_ifaddrhead in_ifaddr; 98struct ifqueue ipintrq; 99 100/* 101 * We need to save the IP options in case a protocol wants to respond 102 * to an incoming packet over the same route if the packet got here 103 * using IP source routing. This allows connection establishment and 104 * maintenance when the remote end is on a network that is not known 105 * to us. 106 */ 107int ip_nhops = 0; 108static struct ip_srcrt { 109 struct in_addr dst; /* final destination */ 110 char nop; /* one NOP to align */ 111 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 112 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 113} ip_srcrt; 114 115static void save_rte __P((u_char *, struct in_addr)); 116/* 117 * IP initialization: fill in IP protocol switch table. 118 * All protocols not implemented in kernel go to raw IP protocol handler. 119 */ 120void 121ip_init() 122{ 123 register struct protosw *pr; 124 register int i; 125 126 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 127 if (pr == 0) 128 panic("ip_init"); 129 for (i = 0; i < IPPROTO_MAX; i++) 130 ip_protox[i] = pr - inetsw; 131 for (pr = inetdomain.dom_protosw; 132 pr < inetdomain.dom_protoswNPROTOSW; pr++) 133 if (pr->pr_domain->dom_family == PF_INET && 134 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 135 ip_protox[pr->pr_protocol] = pr - inetsw; 136 LIST_INIT(&ipq); 137 ip_id = time.tv_sec & 0xffff; 138 ipintrq.ifq_maxlen = ipqmaxlen; 139 TAILQ_INIT(&in_ifaddr); 140} 141 142struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 143struct route ipforward_rt; 144 145/* 146 * Ip input routine. Checksum and byte swap header. If fragmented 147 * try to reassemble. Process options. Pass to next level. 148 */ 149void 150ipintr() 151{ 152 register struct ip *ip; 153 register struct mbuf *m; 154 register struct ipq *fp; 155 register struct in_ifaddr *ia; 156 struct ipqent *ipqe; 157 int hlen, mff, s; 158 159next: 160 /* 161 * Get next datagram off input queue and get IP header 162 * in first mbuf. 163 */ 164 s = splimp(); 165 IF_DEQUEUE(&ipintrq, m); 166 splx(s); 167 if (m == 0) 168 return; 169#ifdef DIAGNOSTIC 170 if ((m->m_flags & M_PKTHDR) == 0) 171 panic("ipintr no HDR"); 172#endif 173 /* 174 * If no IP addresses have been set yet but the interfaces 175 * are receiving, can't do anything with incoming packets yet. 176 */ 177 if (in_ifaddr.tqh_first == 0) 178 goto bad; 179 ipstat.ips_total++; 180 if (m->m_len < sizeof (struct ip) && 181 (m = m_pullup(m, sizeof (struct ip))) == 0) { 182 ipstat.ips_toosmall++; 183 goto next; 184 } 185 ip = mtod(m, struct ip *); 186 if (ip->ip_v != IPVERSION) { 187 ipstat.ips_badvers++; 188 goto bad; 189 } 190 hlen = ip->ip_hl << 2; 191 if (hlen < sizeof(struct ip)) { /* minimum header length */ 192 ipstat.ips_badhlen++; 193 goto bad; 194 } 195 if (hlen > m->m_len) { 196 if ((m = m_pullup(m, hlen)) == 0) { 197 ipstat.ips_badhlen++; 198 goto next; 199 } 200 ip = mtod(m, struct ip *); 201 } 202 if (ip->ip_sum = in_cksum(m, hlen)) { 203 ipstat.ips_badsum++; 204 goto bad; 205 } 206 207 /* 208 * Convert fields to host representation. 209 */ 210 NTOHS(ip->ip_len); 211 if (ip->ip_len < hlen) { 212 ipstat.ips_badlen++; 213 goto bad; 214 } 215 NTOHS(ip->ip_id); 216 NTOHS(ip->ip_off); 217 218 /* 219 * Check that the amount of data in the buffers 220 * is as at least much as the IP header would have us expect. 221 * Trim mbufs if longer than we expect. 222 * Drop packet if shorter than we expect. 223 */ 224 if (m->m_pkthdr.len < ip->ip_len) { 225 ipstat.ips_tooshort++; 226 goto bad; 227 } 228 if (m->m_pkthdr.len > ip->ip_len) { 229 if (m->m_len == m->m_pkthdr.len) { 230 m->m_len = ip->ip_len; 231 m->m_pkthdr.len = ip->ip_len; 232 } else 233 m_adj(m, ip->ip_len - m->m_pkthdr.len); 234 } 235 236 /* 237 * Process options and, if not destined for us, 238 * ship it on. ip_dooptions returns 1 when an 239 * error was detected (causing an icmp message 240 * to be sent and the original packet to be freed). 241 */ 242 ip_nhops = 0; /* for source routed packets */ 243 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 244 goto next; 245 246 /* 247 * Check our list of addresses, to see if the packet is for us. 248 */ 249 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 250 if (ip->ip_dst.s_addr == ia->ia_addr.sin_addr.s_addr) 251 goto ours; 252 if (((ip_directedbcast == 0) || (ip_directedbcast && 253 ia->ia_ifp == m->m_pkthdr.rcvif)) && 254 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 255 if (ip->ip_dst.s_addr == ia->ia_broadaddr.sin_addr.s_addr || 256 ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr || 257 /* 258 * Look for all-0's host part (old broadcast addr), 259 * either for subnet or net. 260 */ 261 ip->ip_dst.s_addr == ia->ia_subnet || 262 ip->ip_dst.s_addr == ia->ia_net) 263 goto ours; 264 } 265 } 266 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 267 struct in_multi *inm; 268#ifdef MROUTING 269 extern struct socket *ip_mrouter; 270 271 if (m->m_flags & M_EXT) { 272 if ((m = m_pullup(m, hlen)) == 0) { 273 ipstat.ips_toosmall++; 274 goto next; 275 } 276 ip = mtod(m, struct ip *); 277 } 278 279 if (ip_mrouter) { 280 /* 281 * If we are acting as a multicast router, all 282 * incoming multicast packets are passed to the 283 * kernel-level multicast forwarding function. 284 * The packet is returned (relatively) intact; if 285 * ip_mforward() returns a non-zero value, the packet 286 * must be discarded, else it may be accepted below. 287 * 288 * (The IP ident field is put in the same byte order 289 * as expected when ip_mforward() is called from 290 * ip_output().) 291 */ 292 ip->ip_id = htons(ip->ip_id); 293 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 294 ipstat.ips_cantforward++; 295 m_freem(m); 296 goto next; 297 } 298 ip->ip_id = ntohs(ip->ip_id); 299 300 /* 301 * The process-level routing demon needs to receive 302 * all multicast IGMP packets, whether or not this 303 * host belongs to their destination groups. 304 */ 305 if (ip->ip_p == IPPROTO_IGMP) 306 goto ours; 307 ipstat.ips_forward++; 308 } 309#endif 310 /* 311 * See if we belong to the destination multicast group on the 312 * arrival interface. 313 */ 314 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 315 if (inm == NULL) { 316 ipstat.ips_cantforward++; 317 m_freem(m); 318 goto next; 319 } 320 goto ours; 321 } 322 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 323 ip->ip_dst.s_addr == INADDR_ANY) 324 goto ours; 325 326 /* 327 * Not for us; forward if possible and desirable. 328 */ 329 if (ipforwarding == 0) { 330 ipstat.ips_cantforward++; 331 m_freem(m); 332 } else 333 ip_forward(m, 0); 334 goto next; 335 336ours: 337 /* 338 * If offset or IP_MF are set, must reassemble. 339 * Otherwise, nothing need be done. 340 * (We could look in the reassembly queue to see 341 * if the packet was previously fragmented, 342 * but it's not worth the time; just let them time out.) 343 */ 344 if (ip->ip_off &~ IP_DF) { 345 if (m->m_flags & M_EXT) { /* XXX */ 346 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 347 ipstat.ips_toosmall++; 348 goto next; 349 } 350 ip = mtod(m, struct ip *); 351 } 352 /* 353 * Look for queue of fragments 354 * of this datagram. 355 */ 356 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 357 if (ip->ip_id == fp->ipq_id && 358 ip->ip_src.s_addr == fp->ipq_src.s_addr && 359 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 360 ip->ip_p == fp->ipq_p) 361 goto found; 362 fp = 0; 363found: 364 365 /* 366 * Adjust ip_len to not reflect header, 367 * set ipqe_mff if more fragments are expected, 368 * convert offset of this to bytes. 369 */ 370 ip->ip_len -= hlen; 371 mff = (ip->ip_off & IP_MF) != 0; 372 if (mff) { 373 /* 374 * Make sure that fragments have a data length 375 * that's a non-zero multiple of 8 bytes. 376 */ 377 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 378 ipstat.ips_badfrags++; 379 goto bad; 380 } 381 } 382 ip->ip_off <<= 3; 383 384 /* 385 * If datagram marked as having more fragments 386 * or if this is not the first fragment, 387 * attempt reassembly; if it succeeds, proceed. 388 */ 389 if (mff || ip->ip_off) { 390 ipstat.ips_fragments++; 391 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 392 M_IPQ, M_NOWAIT); 393 if (ipqe == NULL) { 394 ipstat.ips_rcvmemdrop++; 395 goto bad; 396 } 397 ipqe->ipqe_mff = mff; 398 ipqe->ipqe_ip = ip; 399 ip = ip_reass(ipqe, fp); 400 if (ip == 0) 401 goto next; 402 ipstat.ips_reassembled++; 403 m = dtom(ip); 404 } else 405 if (fp) 406 ip_freef(fp); 407 } else 408 ip->ip_len -= hlen; 409 410 /* 411 * Switch out to protocol's input routine. 412 */ 413 ipstat.ips_delivered++; 414 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 415 goto next; 416bad: 417 m_freem(m); 418 goto next; 419} 420 421/* 422 * Take incoming datagram fragment and try to 423 * reassemble it into whole datagram. If a chain for 424 * reassembly of this datagram already exists, then it 425 * is given as fp; otherwise have to make a chain. 426 */ 427struct ip * 428ip_reass(ipqe, fp) 429 register struct ipqent *ipqe; 430 register struct ipq *fp; 431{ 432 register struct mbuf *m = dtom(ipqe->ipqe_ip); 433 register struct ipqent *nq, *p, *q; 434 struct ip *ip; 435 struct mbuf *t; 436 int hlen = ipqe->ipqe_ip->ip_hl << 2; 437 int i, next; 438 439 /* 440 * Presence of header sizes in mbufs 441 * would confuse code below. 442 */ 443 m->m_data += hlen; 444 m->m_len -= hlen; 445 446 /* 447 * If first fragment to arrive, create a reassembly queue. 448 */ 449 if (fp == 0) { 450 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 451 goto dropfrag; 452 fp = mtod(t, struct ipq *); 453 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 454 fp->ipq_ttl = IPFRAGTTL; 455 fp->ipq_p = ipqe->ipqe_ip->ip_p; 456 fp->ipq_id = ipqe->ipqe_ip->ip_id; 457 LIST_INIT(&fp->ipq_fragq); 458 fp->ipq_src = ipqe->ipqe_ip->ip_src; 459 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 460 p = NULL; 461 goto insert; 462 } 463 464 /* 465 * Find a segment which begins after this one does. 466 */ 467 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 468 p = q, q = q->ipqe_q.le_next) 469 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 470 break; 471 472 /* 473 * If there is a preceding segment, it may provide some of 474 * our data already. If so, drop the data from the incoming 475 * segment. If it provides all of our data, drop us. 476 */ 477 if (p != NULL) { 478 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 479 ipqe->ipqe_ip->ip_off; 480 if (i > 0) { 481 if (i >= ipqe->ipqe_ip->ip_len) 482 goto dropfrag; 483 m_adj(dtom(ipqe->ipqe_ip), i); 484 ipqe->ipqe_ip->ip_off += i; 485 ipqe->ipqe_ip->ip_len -= i; 486 } 487 } 488 489 /* 490 * While we overlap succeeding segments trim them or, 491 * if they are completely covered, dequeue them. 492 */ 493 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 494 q->ipqe_ip->ip_off; q = nq) { 495 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 496 q->ipqe_ip->ip_off; 497 if (i < q->ipqe_ip->ip_len) { 498 q->ipqe_ip->ip_len -= i; 499 q->ipqe_ip->ip_off += i; 500 m_adj(dtom(q->ipqe_ip), i); 501 break; 502 } 503 nq = q->ipqe_q.le_next; 504 m_freem(dtom(q->ipqe_ip)); 505 LIST_REMOVE(q, ipqe_q); 506 FREE(q, M_IPQ); 507 } 508 509insert: 510 /* 511 * Stick new segment in its place; 512 * check for complete reassembly. 513 */ 514 if (p == NULL) { 515 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 516 } else { 517 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 518 } 519 next = 0; 520 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 521 p = q, q = q->ipqe_q.le_next) { 522 if (q->ipqe_ip->ip_off != next) 523 return (0); 524 next += q->ipqe_ip->ip_len; 525 } 526 if (p->ipqe_mff) 527 return (0); 528 529 /* 530 * Reassembly is complete; concatenate fragments. 531 */ 532 q = fp->ipq_fragq.lh_first; 533 ip = q->ipqe_ip; 534 m = dtom(q->ipqe_ip); 535 t = m->m_next; 536 m->m_next = 0; 537 m_cat(m, t); 538 nq = q->ipqe_q.le_next; 539 FREE(q, M_IPQ); 540 for (q = nq; q != NULL; q = nq) { 541 t = dtom(q->ipqe_ip); 542 nq = q->ipqe_q.le_next; 543 FREE(q, M_IPQ); 544 m_cat(m, t); 545 } 546 547 /* 548 * Create header for new ip packet by 549 * modifying header of first packet; 550 * dequeue and discard fragment reassembly header. 551 * Make header visible. 552 */ 553 ip->ip_len = next; 554 ip->ip_src = fp->ipq_src; 555 ip->ip_dst = fp->ipq_dst; 556 LIST_REMOVE(fp, ipq_q); 557 (void) m_free(dtom(fp)); 558 m->m_len += (ip->ip_hl << 2); 559 m->m_data -= (ip->ip_hl << 2); 560 /* some debugging cruft by sklower, below, will go away soon */ 561 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 562 register int plen = 0; 563 for (t = m; m; m = m->m_next) 564 plen += m->m_len; 565 t->m_pkthdr.len = plen; 566 } 567 return (ip); 568 569dropfrag: 570 ipstat.ips_fragdropped++; 571 m_freem(m); 572 FREE(ipqe, M_IPQ); 573 return (0); 574} 575 576/* 577 * Free a fragment reassembly header and all 578 * associated datagrams. 579 */ 580void 581ip_freef(fp) 582 struct ipq *fp; 583{ 584 register struct ipqent *q, *p; 585 586 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 587 p = q->ipqe_q.le_next; 588 m_freem(dtom(q->ipqe_ip)); 589 LIST_REMOVE(q, ipqe_q); 590 FREE(q, M_IPQ); 591 } 592 LIST_REMOVE(fp, ipq_q); 593 (void) m_free(dtom(fp)); 594} 595 596/* 597 * IP timer processing; 598 * if a timer expires on a reassembly 599 * queue, discard it. 600 */ 601void 602ip_slowtimo() 603{ 604 register struct ipq *fp, *nfp; 605 int s = splsoftnet(); 606 607 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 608 nfp = fp->ipq_q.le_next; 609 if (--fp->ipq_ttl == 0) { 610 ipstat.ips_fragtimeout++; 611 ip_freef(fp); 612 } 613 } 614 splx(s); 615} 616 617/* 618 * Drain off all datagram fragments. 619 */ 620void 621ip_drain() 622{ 623 624 while (ipq.lh_first != NULL) { 625 ipstat.ips_fragdropped++; 626 ip_freef(ipq.lh_first); 627 } 628} 629 630/* 631 * Do option processing on a datagram, 632 * possibly discarding it if bad options are encountered, 633 * or forwarding it if source-routed. 634 * Returns 1 if packet has been forwarded/freed, 635 * 0 if the packet should be processed further. 636 */ 637int 638ip_dooptions(m) 639 struct mbuf *m; 640{ 641 register struct ip *ip = mtod(m, struct ip *); 642 register u_char *cp; 643 register struct ip_timestamp *ipt; 644 register struct in_ifaddr *ia; 645 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 646 struct in_addr *sin, dst; 647 n_time ntime; 648 649 dst = ip->ip_dst; 650 cp = (u_char *)(ip + 1); 651 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 652 for (; cnt > 0; cnt -= optlen, cp += optlen) { 653 opt = cp[IPOPT_OPTVAL]; 654 if (opt == IPOPT_EOL) 655 break; 656 if (opt == IPOPT_NOP) 657 optlen = 1; 658 else { 659 optlen = cp[IPOPT_OLEN]; 660 if (optlen <= 0 || optlen > cnt) { 661 code = &cp[IPOPT_OLEN] - (u_char *)ip; 662 goto bad; 663 } 664 } 665 switch (opt) { 666 667 default: 668 break; 669 670 /* 671 * Source routing with record. 672 * Find interface with current destination address. 673 * If none on this machine then drop if strictly routed, 674 * or do nothing if loosely routed. 675 * Record interface address and bring up next address 676 * component. If strictly routed make sure next 677 * address is on directly accessible net. 678 */ 679 case IPOPT_LSRR: 680 case IPOPT_SSRR: 681 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 682 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 683 goto bad; 684 } 685 ipaddr.sin_addr = ip->ip_dst; 686 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 687 if (ia == 0) { 688 if (opt == IPOPT_SSRR) { 689 type = ICMP_UNREACH; 690 code = ICMP_UNREACH_SRCFAIL; 691 goto bad; 692 } 693 /* 694 * Loose routing, and not at next destination 695 * yet; nothing to do except forward. 696 */ 697 break; 698 } 699 off--; /* 0 origin */ 700 if (off > optlen - sizeof(struct in_addr)) { 701 /* 702 * End of source route. Should be for us. 703 */ 704 save_rte(cp, ip->ip_src); 705 break; 706 } 707 /* 708 * locate outgoing interface 709 */ 710 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 711 sizeof(ipaddr.sin_addr)); 712 if (opt == IPOPT_SSRR) { 713#define INA struct in_ifaddr * 714#define SA struct sockaddr * 715 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 716 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 717 } else 718 ia = ip_rtaddr(ipaddr.sin_addr); 719 if (ia == 0) { 720 type = ICMP_UNREACH; 721 code = ICMP_UNREACH_SRCFAIL; 722 goto bad; 723 } 724 ip->ip_dst = ipaddr.sin_addr; 725 bcopy((caddr_t)&ia->ia_addr.sin_addr, 726 (caddr_t)(cp + off), sizeof(struct in_addr)); 727 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 728 /* 729 * Let ip_intr's mcast routing check handle mcast pkts 730 */ 731 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 732 break; 733 734 case IPOPT_RR: 735 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 736 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 737 goto bad; 738 } 739 /* 740 * If no space remains, ignore. 741 */ 742 off--; /* 0 origin */ 743 if (off > optlen - sizeof(struct in_addr)) 744 break; 745 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 746 sizeof(ipaddr.sin_addr)); 747 /* 748 * locate outgoing interface; if we're the destination, 749 * use the incoming interface (should be same). 750 */ 751 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 752 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 753 type = ICMP_UNREACH; 754 code = ICMP_UNREACH_HOST; 755 goto bad; 756 } 757 bcopy((caddr_t)&ia->ia_addr.sin_addr, 758 (caddr_t)(cp + off), sizeof(struct in_addr)); 759 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 760 break; 761 762 case IPOPT_TS: 763 code = cp - (u_char *)ip; 764 ipt = (struct ip_timestamp *)cp; 765 if (ipt->ipt_len < 5) 766 goto bad; 767 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 768 if (++ipt->ipt_oflw == 0) 769 goto bad; 770 break; 771 } 772 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 773 switch (ipt->ipt_flg) { 774 775 case IPOPT_TS_TSONLY: 776 break; 777 778 case IPOPT_TS_TSANDADDR: 779 if (ipt->ipt_ptr + sizeof(n_time) + 780 sizeof(struct in_addr) > ipt->ipt_len) 781 goto bad; 782 ipaddr.sin_addr = dst; 783 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 784 m->m_pkthdr.rcvif); 785 if (ia == 0) 786 continue; 787 bcopy((caddr_t)&ia->ia_addr.sin_addr, 788 (caddr_t)sin, sizeof(struct in_addr)); 789 ipt->ipt_ptr += sizeof(struct in_addr); 790 break; 791 792 case IPOPT_TS_PRESPEC: 793 if (ipt->ipt_ptr + sizeof(n_time) + 794 sizeof(struct in_addr) > ipt->ipt_len) 795 goto bad; 796 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 797 sizeof(struct in_addr)); 798 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 799 continue; 800 ipt->ipt_ptr += sizeof(struct in_addr); 801 break; 802 803 default: 804 goto bad; 805 } 806 ntime = iptime(); 807 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 808 sizeof(n_time)); 809 ipt->ipt_ptr += sizeof(n_time); 810 } 811 } 812 if (forward) { 813 if (ip_forwsrcrt == 0) { 814 type = ICMP_UNREACH; 815 code = ICMP_UNREACH_SRCFAIL; 816 goto bad; 817 } 818 ip_forward(m, 1); 819 return (1); 820 } 821 return (0); 822bad: 823 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 824 icmp_error(m, type, code, 0, 0); 825 ipstat.ips_badoptions++; 826 return (1); 827} 828 829/* 830 * Given address of next destination (final or next hop), 831 * return internet address info of interface to be used to get there. 832 */ 833struct in_ifaddr * 834ip_rtaddr(dst) 835 struct in_addr dst; 836{ 837 register struct sockaddr_in *sin; 838 839 sin = satosin(&ipforward_rt.ro_dst); 840 841 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 842 if (ipforward_rt.ro_rt) { 843 RTFREE(ipforward_rt.ro_rt); 844 ipforward_rt.ro_rt = 0; 845 } 846 sin->sin_family = AF_INET; 847 sin->sin_len = sizeof(*sin); 848 sin->sin_addr = dst; 849 850 rtalloc(&ipforward_rt); 851 } 852 if (ipforward_rt.ro_rt == 0) 853 return ((struct in_ifaddr *)0); 854 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 855} 856 857/* 858 * Save incoming source route for use in replies, 859 * to be picked up later by ip_srcroute if the receiver is interested. 860 */ 861void 862save_rte(option, dst) 863 u_char *option; 864 struct in_addr dst; 865{ 866 unsigned olen; 867 868 olen = option[IPOPT_OLEN]; 869#ifdef DIAGNOSTIC 870 if (ipprintfs) 871 printf("save_rte: olen %d\n", olen); 872#endif 873 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 874 return; 875 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 876 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 877 ip_srcrt.dst = dst; 878} 879 880/* 881 * Retrieve incoming source route for use in replies, 882 * in the same form used by setsockopt. 883 * The first hop is placed before the options, will be removed later. 884 */ 885struct mbuf * 886ip_srcroute() 887{ 888 register struct in_addr *p, *q; 889 register struct mbuf *m; 890 891 if (ip_nhops == 0) 892 return ((struct mbuf *)0); 893 m = m_get(M_DONTWAIT, MT_SOOPTS); 894 if (m == 0) 895 return ((struct mbuf *)0); 896 897#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 898 899 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 900 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 901 OPTSIZ; 902#ifdef DIAGNOSTIC 903 if (ipprintfs) 904 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 905#endif 906 907 /* 908 * First save first hop for return route 909 */ 910 p = &ip_srcrt.route[ip_nhops - 1]; 911 *(mtod(m, struct in_addr *)) = *p--; 912#ifdef DIAGNOSTIC 913 if (ipprintfs) 914 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 915#endif 916 917 /* 918 * Copy option fields and padding (nop) to mbuf. 919 */ 920 ip_srcrt.nop = IPOPT_NOP; 921 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 922 bcopy((caddr_t)&ip_srcrt.nop, 923 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 924 q = (struct in_addr *)(mtod(m, caddr_t) + 925 sizeof(struct in_addr) + OPTSIZ); 926#undef OPTSIZ 927 /* 928 * Record return path as an IP source route, 929 * reversing the path (pointers are now aligned). 930 */ 931 while (p >= ip_srcrt.route) { 932#ifdef DIAGNOSTIC 933 if (ipprintfs) 934 printf(" %lx", ntohl(q->s_addr)); 935#endif 936 *q++ = *p--; 937 } 938 /* 939 * Last hop goes to final destination. 940 */ 941 *q = ip_srcrt.dst; 942#ifdef DIAGNOSTIC 943 if (ipprintfs) 944 printf(" %lx\n", ntohl(q->s_addr)); 945#endif 946 return (m); 947} 948 949/* 950 * Strip out IP options, at higher 951 * level protocol in the kernel. 952 * Second argument is buffer to which options 953 * will be moved, and return value is their length. 954 * XXX should be deleted; last arg currently ignored. 955 */ 956void 957ip_stripoptions(m, mopt) 958 register struct mbuf *m; 959 struct mbuf *mopt; 960{ 961 register int i; 962 struct ip *ip = mtod(m, struct ip *); 963 register caddr_t opts; 964 int olen; 965 966 olen = (ip->ip_hl<<2) - sizeof (struct ip); 967 opts = (caddr_t)(ip + 1); 968 i = m->m_len - (sizeof (struct ip) + olen); 969 bcopy(opts + olen, opts, (unsigned)i); 970 m->m_len -= olen; 971 if (m->m_flags & M_PKTHDR) 972 m->m_pkthdr.len -= olen; 973 ip->ip_hl = sizeof(struct ip) >> 2; 974} 975 976int inetctlerrmap[PRC_NCMDS] = { 977 0, 0, 0, 0, 978 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 979 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 980 EMSGSIZE, EHOSTUNREACH, 0, 0, 981 0, 0, 0, 0, 982 ENOPROTOOPT 983}; 984 985/* 986 * Forward a packet. If some error occurs return the sender 987 * an icmp packet. Note we can't always generate a meaningful 988 * icmp message because icmp doesn't have a large enough repertoire 989 * of codes and types. 990 * 991 * If not forwarding, just drop the packet. This could be confusing 992 * if ipforwarding was zero but some routing protocol was advancing 993 * us as a gateway to somewhere. However, we must let the routing 994 * protocol deal with that. 995 * 996 * The srcrt parameter indicates whether the packet is being forwarded 997 * via a source route. 998 */ 999void 1000ip_forward(m, srcrt) 1001 struct mbuf *m; 1002 int srcrt; 1003{ 1004 register struct ip *ip = mtod(m, struct ip *); 1005 register struct sockaddr_in *sin; 1006 register struct rtentry *rt; 1007 int error, type = 0, code; 1008 struct mbuf *mcopy; 1009 n_long dest; 1010 struct ifnet *destifp; 1011 1012 dest = 0; 1013#ifdef DIAGNOSTIC 1014 if (ipprintfs) 1015 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1016 ip->ip_dst, ip->ip_ttl); 1017#endif 1018 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1019 ipstat.ips_cantforward++; 1020 m_freem(m); 1021 return; 1022 } 1023 HTONS(ip->ip_id); 1024 if (ip->ip_ttl <= IPTTLDEC) { 1025 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1026 return; 1027 } 1028 ip->ip_ttl -= IPTTLDEC; 1029 1030 sin = satosin(&ipforward_rt.ro_dst); 1031 if ((rt = ipforward_rt.ro_rt) == 0 || 1032 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1033 if (ipforward_rt.ro_rt) { 1034 RTFREE(ipforward_rt.ro_rt); 1035 ipforward_rt.ro_rt = 0; 1036 } 1037 sin->sin_family = AF_INET; 1038 sin->sin_len = sizeof(*sin); 1039 sin->sin_addr = ip->ip_dst; 1040 1041 rtalloc(&ipforward_rt); 1042 if (ipforward_rt.ro_rt == 0) { 1043 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1044 return; 1045 } 1046 rt = ipforward_rt.ro_rt; 1047 } 1048 1049 /* 1050 * Save at most 64 bytes of the packet in case 1051 * we need to generate an ICMP message to the src. 1052 */ 1053 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1054 1055 /* 1056 * If forwarding packet using same interface that it came in on, 1057 * perhaps should send a redirect to sender to shortcut a hop. 1058 * Only send redirect if source is sending directly to us, 1059 * and if packet was not source routed (or has any options). 1060 * Also, don't send redirect if forwarding using a default route 1061 * or a route modified by a redirect. 1062 */ 1063 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1064 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1065 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1066 ipsendredirects && !srcrt) { 1067 if (rt->rt_ifa && 1068 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1069 ifatoia(rt->rt_ifa)->ia_subnet) { 1070 if (rt->rt_flags & RTF_GATEWAY) 1071 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1072 else 1073 dest = ip->ip_dst.s_addr; 1074 /* Router requirements says to only send host redirects */ 1075 type = ICMP_REDIRECT; 1076 code = ICMP_REDIRECT_HOST; 1077#ifdef DIAGNOSTIC 1078 if (ipprintfs) 1079 printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); 1080#endif 1081 } 1082 } 1083 1084 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1085 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1086 if (error) 1087 ipstat.ips_cantforward++; 1088 else { 1089 ipstat.ips_forward++; 1090 if (type) 1091 ipstat.ips_redirectsent++; 1092 else { 1093 if (mcopy) 1094 m_freem(mcopy); 1095 return; 1096 } 1097 } 1098 if (mcopy == NULL) 1099 return; 1100 destifp = NULL; 1101 1102 switch (error) { 1103 1104 case 0: /* forwarded, but need redirect */ 1105 /* type, code set above */ 1106 break; 1107 1108 case ENETUNREACH: /* shouldn't happen, checked above */ 1109 case EHOSTUNREACH: 1110 case ENETDOWN: 1111 case EHOSTDOWN: 1112 default: 1113 type = ICMP_UNREACH; 1114 code = ICMP_UNREACH_HOST; 1115 break; 1116 1117 case EMSGSIZE: 1118 type = ICMP_UNREACH; 1119 code = ICMP_UNREACH_NEEDFRAG; 1120 if (ipforward_rt.ro_rt) 1121 destifp = ipforward_rt.ro_rt->rt_ifp; 1122 ipstat.ips_cantfrag++; 1123 break; 1124 1125 case ENOBUFS: 1126 type = ICMP_SOURCEQUENCH; 1127 code = 0; 1128 break; 1129 } 1130 icmp_error(mcopy, type, code, dest, destifp); 1131} 1132 1133int 1134ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1135 int *name; 1136 u_int namelen; 1137 void *oldp; 1138 size_t *oldlenp; 1139 void *newp; 1140 size_t newlen; 1141{ 1142 int temp; 1143 1144 /* All sysctl names at this level are terminal. */ 1145 if (namelen != 1) 1146 return (ENOTDIR); 1147 1148 switch (name[0]) { 1149 case IPCTL_FORWARDING: 1150 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1151 case IPCTL_SENDREDIRECTS: 1152 return (sysctl_int(oldp, oldlenp, newp, newlen, 1153 &ipsendredirects)); 1154 case IPCTL_DEFTTL: 1155 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1156#ifdef notyet 1157 case IPCTL_DEFMTU: 1158 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1159#endif 1160 case IPCTL_FORWSRCRT: 1161 /* 1162 * Don't allow this to change in a secure environment. 1163 */ 1164 if (securelevel > 0) 1165 return (EPERM); 1166 return (sysctl_int(oldp, oldlenp, newp, newlen, 1167 &ip_forwsrcrt)); 1168 case IPCTL_DIRECTEDBCAST: 1169 return (sysctl_int(oldp, oldlenp, newp, newlen, 1170 &ip_directedbcast)); 1171 default: 1172 return (EOPNOTSUPP); 1173 } 1174 /* NOTREACHED */ 1175} 1176