ip_input.c revision 1.20
1/* $NetBSD: ip_input.c,v 1.20 1995/06/04 05:58:26 mycroft 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 70int ipforwarding = IPFORWARDING; 71int ipsendredirects = IPSENDREDIRECTS; 72int ip_defttl = IPDEFTTL; 73#ifdef DIAGNOSTIC 74int ipprintfs = 0; 75#endif 76 77extern struct domain inetdomain; 78extern struct protosw inetsw[]; 79u_char ip_protox[IPPROTO_MAX]; 80int ipqmaxlen = IFQ_MAXLEN; 81struct in_ifaddr *in_ifaddr; /* first inet address */ 82struct ifqueue ipintrq; 83 84/* 85 * We need to save the IP options in case a protocol wants to respond 86 * to an incoming packet over the same route if the packet got here 87 * using IP source routing. This allows connection establishment and 88 * maintenance when the remote end is on a network that is not known 89 * to us. 90 */ 91int ip_nhops = 0; 92static struct ip_srcrt { 93 struct in_addr dst; /* final destination */ 94 char nop; /* one NOP to align */ 95 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 96 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 97} ip_srcrt; 98 99#ifdef GATEWAY 100extern int if_index; 101u_int32_t *ip_ifmatrix; 102#endif 103 104static void save_rte __P((u_char *, struct in_addr)); 105/* 106 * IP initialization: fill in IP protocol switch table. 107 * All protocols not implemented in kernel go to raw IP protocol handler. 108 */ 109void 110ip_init() 111{ 112 register struct protosw *pr; 113 register int i; 114 115 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 116 if (pr == 0) 117 panic("ip_init"); 118 for (i = 0; i < IPPROTO_MAX; i++) 119 ip_protox[i] = pr - inetsw; 120 for (pr = inetdomain.dom_protosw; 121 pr < inetdomain.dom_protoswNPROTOSW; pr++) 122 if (pr->pr_domain->dom_family == PF_INET && 123 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 124 ip_protox[pr->pr_protocol] = pr - inetsw; 125 ipq.next = ipq.prev = &ipq; 126 ip_id = time.tv_sec & 0xffff; 127 ipintrq.ifq_maxlen = ipqmaxlen; 128#ifdef GATEWAY 129 i = (if_index + 1) * (if_index + 1) * sizeof (u_int32_t); 130 ip_ifmatrix = (u_int32_t *) malloc(i, M_RTABLE, M_WAITOK); 131 bzero((char *)ip_ifmatrix, i); 132#endif 133} 134 135struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 136struct route ipforward_rt; 137 138/* 139 * Ip input routine. Checksum and byte swap header. If fragmented 140 * try to reassemble. Process options. Pass to next level. 141 */ 142void 143ipintr() 144{ 145 register struct ip *ip; 146 register struct mbuf *m; 147 register struct ipq *fp; 148 register struct in_ifaddr *ia; 149 int hlen, s; 150 151next: 152 /* 153 * Get next datagram off input queue and get IP header 154 * in first mbuf. 155 */ 156 s = splimp(); 157 IF_DEQUEUE(&ipintrq, m); 158 splx(s); 159 if (m == 0) 160 return; 161#ifdef DIAGNOSTIC 162 if ((m->m_flags & M_PKTHDR) == 0) 163 panic("ipintr no HDR"); 164#endif 165 /* 166 * If no IP addresses have been set yet but the interfaces 167 * are receiving, can't do anything with incoming packets yet. 168 */ 169 if (in_ifaddr == NULL) 170 goto bad; 171 ipstat.ips_total++; 172 if (m->m_len < sizeof (struct ip) && 173 (m = m_pullup(m, sizeof (struct ip))) == 0) { 174 ipstat.ips_toosmall++; 175 goto next; 176 } 177 ip = mtod(m, struct ip *); 178 if (ip->ip_v != IPVERSION) { 179 ipstat.ips_badvers++; 180 goto bad; 181 } 182 hlen = ip->ip_hl << 2; 183 if (hlen < sizeof(struct ip)) { /* minimum header length */ 184 ipstat.ips_badhlen++; 185 goto bad; 186 } 187 if (hlen > m->m_len) { 188 if ((m = m_pullup(m, hlen)) == 0) { 189 ipstat.ips_badhlen++; 190 goto next; 191 } 192 ip = mtod(m, struct ip *); 193 } 194 if (ip->ip_sum = in_cksum(m, hlen)) { 195 ipstat.ips_badsum++; 196 goto bad; 197 } 198 199 /* 200 * Convert fields to host representation. 201 */ 202 NTOHS(ip->ip_len); 203 if (ip->ip_len < hlen) { 204 ipstat.ips_badlen++; 205 goto bad; 206 } 207 NTOHS(ip->ip_id); 208 NTOHS(ip->ip_off); 209 210 /* 211 * Check that the amount of data in the buffers 212 * is as at least much as the IP header would have us expect. 213 * Trim mbufs if longer than we expect. 214 * Drop packet if shorter than we expect. 215 */ 216 if (m->m_pkthdr.len < ip->ip_len) { 217 ipstat.ips_tooshort++; 218 goto bad; 219 } 220 if (m->m_pkthdr.len > ip->ip_len) { 221 if (m->m_len == m->m_pkthdr.len) { 222 m->m_len = ip->ip_len; 223 m->m_pkthdr.len = ip->ip_len; 224 } else 225 m_adj(m, ip->ip_len - m->m_pkthdr.len); 226 } 227 228 /* 229 * Process options and, if not destined for us, 230 * ship it on. ip_dooptions returns 1 when an 231 * error was detected (causing an icmp message 232 * to be sent and the original packet to be freed). 233 */ 234 ip_nhops = 0; /* for source routed packets */ 235 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 236 goto next; 237 238 /* 239 * Check our list of addresses, to see if the packet is for us. 240 */ 241 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 242 if (ip->ip_dst.s_addr == ia->ia_addr.sin_addr.s_addr) 243 goto ours; 244 if ( 245#ifdef DIRECTED_BROADCAST 246 ia->ia_ifp == m->m_pkthdr.rcvif && 247#endif 248 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 249 if (ip->ip_dst.s_addr == ia->ia_broadaddr.sin_addr.s_addr || 250 ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr || 251 /* 252 * Look for all-0's host part (old broadcast addr), 253 * either for subnet or net. 254 */ 255 ip->ip_dst.s_addr == ia->ia_subnet || 256 ip->ip_dst.s_addr == ia->ia_net) 257 goto ours; 258 } 259 } 260 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 261 struct in_multi *inm; 262#ifdef MROUTING 263 extern struct socket *ip_mrouter; 264 265 if (m->m_flags & M_EXT) { 266 if ((m = m_pullup(m, hlen)) == 0) { 267 ipstat.ips_toosmall++; 268 goto next; 269 } 270 ip = mtod(m, struct ip *); 271 } 272 273 if (ip_mrouter) { 274 /* 275 * If we are acting as a multicast router, all 276 * incoming multicast packets are passed to the 277 * kernel-level multicast forwarding function. 278 * The packet is returned (relatively) intact; if 279 * ip_mforward() returns a non-zero value, the packet 280 * must be discarded, else it may be accepted below. 281 * 282 * (The IP ident field is put in the same byte order 283 * as expected when ip_mforward() is called from 284 * ip_output().) 285 */ 286 ip->ip_id = htons(ip->ip_id); 287 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 288 ipstat.ips_cantforward++; 289 m_freem(m); 290 goto next; 291 } 292 ip->ip_id = ntohs(ip->ip_id); 293 294 /* 295 * The process-level routing demon needs to receive 296 * all multicast IGMP packets, whether or not this 297 * host belongs to their destination groups. 298 */ 299 if (ip->ip_p == IPPROTO_IGMP) 300 goto ours; 301 ipstat.ips_forward++; 302 } 303#endif 304 /* 305 * See if we belong to the destination multicast group on the 306 * arrival interface. 307 */ 308 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 309 if (inm == NULL) { 310 ipstat.ips_cantforward++; 311 m_freem(m); 312 goto next; 313 } 314 goto ours; 315 } 316 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 317 ip->ip_dst.s_addr == INADDR_ANY) 318 goto ours; 319 320 /* 321 * Not for us; forward if possible and desirable. 322 */ 323 if (ipforwarding == 0) { 324 ipstat.ips_cantforward++; 325 m_freem(m); 326 } else 327 ip_forward(m, 0); 328 goto next; 329 330ours: 331 /* 332 * If offset or IP_MF are set, must reassemble. 333 * Otherwise, nothing need be done. 334 * (We could look in the reassembly queue to see 335 * if the packet was previously fragmented, 336 * but it's not worth the time; just let them time out.) 337 */ 338 if (ip->ip_off &~ IP_DF) { 339 if (m->m_flags & M_EXT) { /* XXX */ 340 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 341 ipstat.ips_toosmall++; 342 goto next; 343 } 344 ip = mtod(m, struct ip *); 345 } 346 /* 347 * Look for queue of fragments 348 * of this datagram. 349 */ 350 for (fp = ipq.next; fp != &ipq; fp = fp->next) 351 if (ip->ip_id == fp->ipq_id && 352 ip->ip_src.s_addr == fp->ipq_src.s_addr && 353 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 354 ip->ip_p == fp->ipq_p) 355 goto found; 356 fp = 0; 357found: 358 359 /* 360 * Adjust ip_len to not reflect header, 361 * set ip_mff if more fragments are expected, 362 * convert offset of this to bytes. 363 */ 364 ip->ip_len -= hlen; 365 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 366 if (ip->ip_off & IP_MF) { 367 /* 368 * Make sure that fragments have a data length 369 * that's a non-zero multiple of 8 bytes. 370 */ 371 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 372 ipstat.ips_badfrags++; 373 goto bad; 374 } 375 ((struct ipasfrag *)ip)->ipf_mff |= 1; 376 } 377 ip->ip_off <<= 3; 378 379 /* 380 * If datagram marked as having more fragments 381 * or if this is not the first fragment, 382 * attempt reassembly; if it succeeds, proceed. 383 */ 384 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 385 ipstat.ips_fragments++; 386 ip = ip_reass((struct ipasfrag *)ip, fp); 387 if (ip == 0) 388 goto next; 389 ipstat.ips_reassembled++; 390 m = dtom(ip); 391 } else 392 if (fp) 393 ip_freef(fp); 394 } else 395 ip->ip_len -= hlen; 396 397 /* 398 * Switch out to protocol's input routine. 399 */ 400 ipstat.ips_delivered++; 401 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 402 goto next; 403bad: 404 m_freem(m); 405 goto next; 406} 407 408/* 409 * Take incoming datagram fragment and try to 410 * reassemble it into whole datagram. If a chain for 411 * reassembly of this datagram already exists, then it 412 * is given as fp; otherwise have to make a chain. 413 */ 414struct ip * 415ip_reass(ip, fp) 416 register struct ipasfrag *ip; 417 register struct ipq *fp; 418{ 419 register struct mbuf *m = dtom(ip); 420 register struct ipasfrag *q; 421 struct mbuf *t; 422 int hlen = ip->ip_hl << 2; 423 int i, next; 424 425 /* 426 * Presence of header sizes in mbufs 427 * would confuse code below. 428 */ 429 m->m_data += hlen; 430 m->m_len -= hlen; 431 432 /* 433 * If first fragment to arrive, create a reassembly queue. 434 */ 435 if (fp == 0) { 436 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 437 goto dropfrag; 438 fp = mtod(t, struct ipq *); 439 insque(fp, &ipq); 440 fp->ipq_ttl = IPFRAGTTL; 441 fp->ipq_p = ip->ip_p; 442 fp->ipq_id = ip->ip_id; 443 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 444 fp->ipq_src = ((struct ip *)ip)->ip_src; 445 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 446 q = (struct ipasfrag *)fp; 447 goto insert; 448 } 449 450 /* 451 * Find a segment which begins after this one does. 452 */ 453 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 454 if (q->ip_off > ip->ip_off) 455 break; 456 457 /* 458 * If there is a preceding segment, it may provide some of 459 * our data already. If so, drop the data from the incoming 460 * segment. If it provides all of our data, drop us. 461 */ 462 if (q->ipf_prev != (struct ipasfrag *)fp) { 463 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 464 if (i > 0) { 465 if (i >= ip->ip_len) 466 goto dropfrag; 467 m_adj(dtom(ip), i); 468 ip->ip_off += i; 469 ip->ip_len -= i; 470 } 471 } 472 473 /* 474 * While we overlap succeeding segments trim them or, 475 * if they are completely covered, dequeue them. 476 */ 477 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 478 i = (ip->ip_off + ip->ip_len) - q->ip_off; 479 if (i < q->ip_len) { 480 q->ip_len -= i; 481 q->ip_off += i; 482 m_adj(dtom(q), i); 483 break; 484 } 485 q = q->ipf_next; 486 m_freem(dtom(q->ipf_prev)); 487 ip_deq(q->ipf_prev); 488 } 489 490insert: 491 /* 492 * Stick new segment in its place; 493 * check for complete reassembly. 494 */ 495 ip_enq(ip, q->ipf_prev); 496 next = 0; 497 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 498 if (q->ip_off != next) 499 return (0); 500 next += q->ip_len; 501 } 502 if (q->ipf_prev->ipf_mff & 1) 503 return (0); 504 505 /* 506 * Reassembly is complete; concatenate fragments. 507 */ 508 q = fp->ipq_next; 509 m = dtom(q); 510 t = m->m_next; 511 m->m_next = 0; 512 m_cat(m, t); 513 q = q->ipf_next; 514 while (q != (struct ipasfrag *)fp) { 515 t = dtom(q); 516 q = q->ipf_next; 517 m_cat(m, t); 518 } 519 520 /* 521 * Create header for new ip packet by 522 * modifying header of first packet; 523 * dequeue and discard fragment reassembly header. 524 * Make header visible. 525 */ 526 ip = fp->ipq_next; 527 ip->ip_len = next; 528 ip->ipf_mff &= ~1; 529 ((struct ip *)ip)->ip_src = fp->ipq_src; 530 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 531 remque(fp); 532 (void) m_free(dtom(fp)); 533 m = dtom(ip); 534 m->m_len += (ip->ip_hl << 2); 535 m->m_data -= (ip->ip_hl << 2); 536 /* some debugging cruft by sklower, below, will go away soon */ 537 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 538 register int plen = 0; 539 for (t = m; m; m = m->m_next) 540 plen += m->m_len; 541 t->m_pkthdr.len = plen; 542 } 543 return ((struct ip *)ip); 544 545dropfrag: 546 ipstat.ips_fragdropped++; 547 m_freem(m); 548 return (0); 549} 550 551/* 552 * Free a fragment reassembly header and all 553 * associated datagrams. 554 */ 555void 556ip_freef(fp) 557 struct ipq *fp; 558{ 559 register struct ipasfrag *q, *p; 560 561 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 562 p = q->ipf_next; 563 ip_deq(q); 564 m_freem(dtom(q)); 565 } 566 remque(fp); 567 (void) m_free(dtom(fp)); 568} 569 570/* 571 * Put an ip fragment on a reassembly chain. 572 * Like insque, but pointers in middle of structure. 573 */ 574void 575ip_enq(p, prev) 576 register struct ipasfrag *p, *prev; 577{ 578 579 p->ipf_prev = prev; 580 p->ipf_next = prev->ipf_next; 581 prev->ipf_next->ipf_prev = p; 582 prev->ipf_next = p; 583} 584 585/* 586 * To ip_enq as remque is to insque. 587 */ 588void 589ip_deq(p) 590 register struct ipasfrag *p; 591{ 592 593 p->ipf_prev->ipf_next = p->ipf_next; 594 p->ipf_next->ipf_prev = p->ipf_prev; 595} 596 597/* 598 * IP timer processing; 599 * if a timer expires on a reassembly 600 * queue, discard it. 601 */ 602void 603ip_slowtimo() 604{ 605 register struct ipq *fp; 606 int s = splnet(); 607 608 fp = ipq.next; 609 if (fp == 0) { 610 splx(s); 611 return; 612 } 613 while (fp != &ipq) { 614 --fp->ipq_ttl; 615 fp = fp->next; 616 if (fp->prev->ipq_ttl == 0) { 617 ipstat.ips_fragtimeout++; 618 ip_freef(fp->prev); 619 } 620 } 621 splx(s); 622} 623 624/* 625 * Drain off all datagram fragments. 626 */ 627void 628ip_drain() 629{ 630 631 while (ipq.next != &ipq) { 632 ipstat.ips_fragdropped++; 633 ip_freef(ipq.next); 634 } 635} 636 637/* 638 * Do option processing on a datagram, 639 * possibly discarding it if bad options are encountered, 640 * or forwarding it if source-routed. 641 * Returns 1 if packet has been forwarded/freed, 642 * 0 if the packet should be processed further. 643 */ 644int 645ip_dooptions(m) 646 struct mbuf *m; 647{ 648 register struct ip *ip = mtod(m, struct ip *); 649 register u_char *cp; 650 register struct ip_timestamp *ipt; 651 register struct in_ifaddr *ia; 652 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 653 struct in_addr *sin, dst; 654 n_time ntime; 655 656 dst = ip->ip_dst; 657 cp = (u_char *)(ip + 1); 658 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 659 for (; cnt > 0; cnt -= optlen, cp += optlen) { 660 opt = cp[IPOPT_OPTVAL]; 661 if (opt == IPOPT_EOL) 662 break; 663 if (opt == IPOPT_NOP) 664 optlen = 1; 665 else { 666 optlen = cp[IPOPT_OLEN]; 667 if (optlen <= 0 || optlen > cnt) { 668 code = &cp[IPOPT_OLEN] - (u_char *)ip; 669 goto bad; 670 } 671 } 672 switch (opt) { 673 674 default: 675 break; 676 677 /* 678 * Source routing with record. 679 * Find interface with current destination address. 680 * If none on this machine then drop if strictly routed, 681 * or do nothing if loosely routed. 682 * Record interface address and bring up next address 683 * component. If strictly routed make sure next 684 * address is on directly accessible net. 685 */ 686 case IPOPT_LSRR: 687 case IPOPT_SSRR: 688 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 689 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 690 goto bad; 691 } 692 ipaddr.sin_addr = ip->ip_dst; 693 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 694 if (ia == 0) { 695 if (opt == IPOPT_SSRR) { 696 type = ICMP_UNREACH; 697 code = ICMP_UNREACH_SRCFAIL; 698 goto bad; 699 } 700 /* 701 * Loose routing, and not at next destination 702 * yet; nothing to do except forward. 703 */ 704 break; 705 } 706 off--; /* 0 origin */ 707 if (off > optlen - sizeof(struct in_addr)) { 708 /* 709 * End of source route. Should be for us. 710 */ 711 save_rte(cp, ip->ip_src); 712 break; 713 } 714 /* 715 * locate outgoing interface 716 */ 717 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 718 sizeof(ipaddr.sin_addr)); 719 if (opt == IPOPT_SSRR) { 720#define INA struct in_ifaddr * 721#define SA struct sockaddr * 722 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 723 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 724 } else 725 ia = ip_rtaddr(ipaddr.sin_addr); 726 if (ia == 0) { 727 type = ICMP_UNREACH; 728 code = ICMP_UNREACH_SRCFAIL; 729 goto bad; 730 } 731 ip->ip_dst = ipaddr.sin_addr; 732 bcopy((caddr_t)&ia->ia_addr.sin_addr, 733 (caddr_t)(cp + off), sizeof(struct in_addr)); 734 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 735 /* 736 * Let ip_intr's mcast routing check handle mcast pkts 737 */ 738 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 739 break; 740 741 case IPOPT_RR: 742 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 743 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 744 goto bad; 745 } 746 /* 747 * If no space remains, ignore. 748 */ 749 off--; /* 0 origin */ 750 if (off > optlen - sizeof(struct in_addr)) 751 break; 752 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 753 sizeof(ipaddr.sin_addr)); 754 /* 755 * locate outgoing interface; if we're the destination, 756 * use the incoming interface (should be same). 757 */ 758 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 759 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 760 type = ICMP_UNREACH; 761 code = ICMP_UNREACH_HOST; 762 goto bad; 763 } 764 bcopy((caddr_t)&ia->ia_addr.sin_addr, 765 (caddr_t)(cp + off), sizeof(struct in_addr)); 766 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 767 break; 768 769 case IPOPT_TS: 770 code = cp - (u_char *)ip; 771 ipt = (struct ip_timestamp *)cp; 772 if (ipt->ipt_len < 5) 773 goto bad; 774 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 775 if (++ipt->ipt_oflw == 0) 776 goto bad; 777 break; 778 } 779 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 780 switch (ipt->ipt_flg) { 781 782 case IPOPT_TS_TSONLY: 783 break; 784 785 case IPOPT_TS_TSANDADDR: 786 if (ipt->ipt_ptr + sizeof(n_time) + 787 sizeof(struct in_addr) > ipt->ipt_len) 788 goto bad; 789 ipaddr.sin_addr = dst; 790 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 791 m->m_pkthdr.rcvif); 792 if (ia == 0) 793 continue; 794 bcopy((caddr_t)&ia->ia_addr.sin_addr, 795 (caddr_t)sin, sizeof(struct in_addr)); 796 ipt->ipt_ptr += sizeof(struct in_addr); 797 break; 798 799 case IPOPT_TS_PRESPEC: 800 if (ipt->ipt_ptr + sizeof(n_time) + 801 sizeof(struct in_addr) > ipt->ipt_len) 802 goto bad; 803 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 804 sizeof(struct in_addr)); 805 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 806 continue; 807 ipt->ipt_ptr += sizeof(struct in_addr); 808 break; 809 810 default: 811 goto bad; 812 } 813 ntime = iptime(); 814 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 815 sizeof(n_time)); 816 ipt->ipt_ptr += sizeof(n_time); 817 } 818 } 819 if (forward) { 820 ip_forward(m, 1); 821 return (1); 822 } 823 return (0); 824bad: 825 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 826 icmp_error(m, type, code, 0, 0); 827 ipstat.ips_badoptions++; 828 return (1); 829} 830 831/* 832 * Given address of next destination (final or next hop), 833 * return internet address info of interface to be used to get there. 834 */ 835struct in_ifaddr * 836ip_rtaddr(dst) 837 struct in_addr dst; 838{ 839 register struct sockaddr_in *sin; 840 841 sin = satosin(&ipforward_rt.ro_dst); 842 843 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 844 if (ipforward_rt.ro_rt) { 845 RTFREE(ipforward_rt.ro_rt); 846 ipforward_rt.ro_rt = 0; 847 } 848 sin->sin_family = AF_INET; 849 sin->sin_len = sizeof(*sin); 850 sin->sin_addr = dst; 851 852 rtalloc(&ipforward_rt); 853 } 854 if (ipforward_rt.ro_rt == 0) 855 return ((struct in_ifaddr *)0); 856 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 857} 858 859/* 860 * Save incoming source route for use in replies, 861 * to be picked up later by ip_srcroute if the receiver is interested. 862 */ 863void 864save_rte(option, dst) 865 u_char *option; 866 struct in_addr dst; 867{ 868 unsigned olen; 869 870 olen = option[IPOPT_OLEN]; 871#ifdef DIAGNOSTIC 872 if (ipprintfs) 873 printf("save_rte: olen %d\n", olen); 874#endif 875 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 876 return; 877 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 878 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 879 ip_srcrt.dst = dst; 880} 881 882/* 883 * Retrieve incoming source route for use in replies, 884 * in the same form used by setsockopt. 885 * The first hop is placed before the options, will be removed later. 886 */ 887struct mbuf * 888ip_srcroute() 889{ 890 register struct in_addr *p, *q; 891 register struct mbuf *m; 892 893 if (ip_nhops == 0) 894 return ((struct mbuf *)0); 895 m = m_get(M_DONTWAIT, MT_SOOPTS); 896 if (m == 0) 897 return ((struct mbuf *)0); 898 899#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 900 901 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 902 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 903 OPTSIZ; 904#ifdef DIAGNOSTIC 905 if (ipprintfs) 906 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 907#endif 908 909 /* 910 * First save first hop for return route 911 */ 912 p = &ip_srcrt.route[ip_nhops - 1]; 913 *(mtod(m, struct in_addr *)) = *p--; 914#ifdef DIAGNOSTIC 915 if (ipprintfs) 916 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 917#endif 918 919 /* 920 * Copy option fields and padding (nop) to mbuf. 921 */ 922 ip_srcrt.nop = IPOPT_NOP; 923 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 924 bcopy((caddr_t)&ip_srcrt.nop, 925 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 926 q = (struct in_addr *)(mtod(m, caddr_t) + 927 sizeof(struct in_addr) + OPTSIZ); 928#undef OPTSIZ 929 /* 930 * Record return path as an IP source route, 931 * reversing the path (pointers are now aligned). 932 */ 933 while (p >= ip_srcrt.route) { 934#ifdef DIAGNOSTIC 935 if (ipprintfs) 936 printf(" %lx", ntohl(q->s_addr)); 937#endif 938 *q++ = *p--; 939 } 940 /* 941 * Last hop goes to final destination. 942 */ 943 *q = ip_srcrt.dst; 944#ifdef DIAGNOSTIC 945 if (ipprintfs) 946 printf(" %lx\n", ntohl(q->s_addr)); 947#endif 948 return (m); 949} 950 951/* 952 * Strip out IP options, at higher 953 * level protocol in the kernel. 954 * Second argument is buffer to which options 955 * will be moved, and return value is their length. 956 * XXX should be deleted; last arg currently ignored. 957 */ 958void 959ip_stripoptions(m, mopt) 960 register struct mbuf *m; 961 struct mbuf *mopt; 962{ 963 register int i; 964 struct ip *ip = mtod(m, struct ip *); 965 register caddr_t opts; 966 int olen; 967 968 olen = (ip->ip_hl<<2) - sizeof (struct ip); 969 opts = (caddr_t)(ip + 1); 970 i = m->m_len - (sizeof (struct ip) + olen); 971 bcopy(opts + olen, opts, (unsigned)i); 972 m->m_len -= olen; 973 if (m->m_flags & M_PKTHDR) 974 m->m_pkthdr.len -= olen; 975 ip->ip_hl = sizeof(struct ip) >> 2; 976} 977 978u_char inetctlerrmap[PRC_NCMDS] = { 979 0, 0, 0, 0, 980 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 981 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 982 EMSGSIZE, EHOSTUNREACH, 0, 0, 983 0, 0, 0, 0, 984 ENOPROTOOPT 985}; 986 987/* 988 * Forward a packet. If some error occurs return the sender 989 * an icmp packet. Note we can't always generate a meaningful 990 * icmp message because icmp doesn't have a large enough repertoire 991 * of codes and types. 992 * 993 * If not forwarding, just drop the packet. This could be confusing 994 * if ipforwarding was zero but some routing protocol was advancing 995 * us as a gateway to somewhere. However, we must let the routing 996 * protocol deal with that. 997 * 998 * The srcrt parameter indicates whether the packet is being forwarded 999 * via a source route. 1000 */ 1001void 1002ip_forward(m, srcrt) 1003 struct mbuf *m; 1004 int srcrt; 1005{ 1006 register struct ip *ip = mtod(m, struct ip *); 1007 register struct sockaddr_in *sin; 1008 register struct rtentry *rt; 1009 int error, type = 0, code; 1010 struct mbuf *mcopy; 1011 n_long dest; 1012 struct ifnet *destifp; 1013 1014 dest = 0; 1015#ifdef DIAGNOSTIC 1016 if (ipprintfs) 1017 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1018 ip->ip_dst, ip->ip_ttl); 1019#endif 1020 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1021 ipstat.ips_cantforward++; 1022 m_freem(m); 1023 return; 1024 } 1025 HTONS(ip->ip_id); 1026 if (ip->ip_ttl <= IPTTLDEC) { 1027 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1028 return; 1029 } 1030 ip->ip_ttl -= IPTTLDEC; 1031 1032 sin = satosin(&ipforward_rt.ro_dst); 1033 if ((rt = ipforward_rt.ro_rt) == 0 || 1034 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1035 if (ipforward_rt.ro_rt) { 1036 RTFREE(ipforward_rt.ro_rt); 1037 ipforward_rt.ro_rt = 0; 1038 } 1039 sin->sin_family = AF_INET; 1040 sin->sin_len = sizeof(*sin); 1041 sin->sin_addr = ip->ip_dst; 1042 1043 rtalloc(&ipforward_rt); 1044 if (ipforward_rt.ro_rt == 0) { 1045 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1046 return; 1047 } 1048 rt = ipforward_rt.ro_rt; 1049 } 1050 1051 /* 1052 * Save at most 64 bytes of the packet in case 1053 * we need to generate an ICMP message to the src. 1054 */ 1055 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1056 1057#ifdef GATEWAY 1058 ip_ifmatrix[rt->rt_ifp->if_index + 1059 if_index * m->m_pkthdr.rcvif->if_index]++; 1060#endif 1061 /* 1062 * If forwarding packet using same interface that it came in on, 1063 * perhaps should send a redirect to sender to shortcut a hop. 1064 * Only send redirect if source is sending directly to us, 1065 * and if packet was not source routed (or has any options). 1066 * Also, don't send redirect if forwarding using a default route 1067 * or a route modified by a redirect. 1068 */ 1069 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1070 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1071 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1072 ipsendredirects && !srcrt) { 1073 if (rt->rt_ifa && 1074 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1075 ifatoia(rt->rt_ifa)->ia_subnet) { 1076 if (rt->rt_flags & RTF_GATEWAY) 1077 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1078 else 1079 dest = ip->ip_dst.s_addr; 1080 /* Router requirements says to only send host redirects */ 1081 type = ICMP_REDIRECT; 1082 code = ICMP_REDIRECT_HOST; 1083#ifdef DIAGNOSTIC 1084 if (ipprintfs) 1085 printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); 1086#endif 1087 } 1088 } 1089 1090 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1091#ifdef DIRECTED_BROADCAST 1092 | IP_ALLOWBROADCAST 1093#endif 1094 , 0); 1095 if (error) 1096 ipstat.ips_cantforward++; 1097 else { 1098 ipstat.ips_forward++; 1099 if (type) 1100 ipstat.ips_redirectsent++; 1101 else { 1102 if (mcopy) 1103 m_freem(mcopy); 1104 return; 1105 } 1106 } 1107 if (mcopy == NULL) 1108 return; 1109 destifp = NULL; 1110 1111 switch (error) { 1112 1113 case 0: /* forwarded, but need redirect */ 1114 /* type, code set above */ 1115 break; 1116 1117 case ENETUNREACH: /* shouldn't happen, checked above */ 1118 case EHOSTUNREACH: 1119 case ENETDOWN: 1120 case EHOSTDOWN: 1121 default: 1122 type = ICMP_UNREACH; 1123 code = ICMP_UNREACH_HOST; 1124 break; 1125 1126 case EMSGSIZE: 1127 type = ICMP_UNREACH; 1128 code = ICMP_UNREACH_NEEDFRAG; 1129 if (ipforward_rt.ro_rt) 1130 destifp = ipforward_rt.ro_rt->rt_ifp; 1131 ipstat.ips_cantfrag++; 1132 break; 1133 1134 case ENOBUFS: 1135 type = ICMP_SOURCEQUENCH; 1136 code = 0; 1137 break; 1138 } 1139 icmp_error(mcopy, type, code, dest, destifp); 1140} 1141 1142int 1143ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1144 int *name; 1145 u_int namelen; 1146 void *oldp; 1147 size_t *oldlenp; 1148 void *newp; 1149 size_t newlen; 1150{ 1151 /* All sysctl names at this level are terminal. */ 1152 if (namelen != 1) 1153 return (ENOTDIR); 1154 1155 switch (name[0]) { 1156 case IPCTL_FORWARDING: 1157 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1158 case IPCTL_SENDREDIRECTS: 1159 return (sysctl_int(oldp, oldlenp, newp, newlen, 1160 &ipsendredirects)); 1161 case IPCTL_DEFTTL: 1162 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1163#ifdef notyet 1164 case IPCTL_DEFMTU: 1165 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1166#endif 1167 default: 1168 return (EOPNOTSUPP); 1169 } 1170 /* NOTREACHED */ 1171} 1172