ip_input.c revision 1.8
1/* 2 * Copyright (c) 1982, 1986, 1988 Regents of the University of California. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * from: @(#)ip_input.c 7.19 (Berkeley) 5/25/91 34 * $Id: ip_input.c,v 1.8 1994/01/09 01:06:13 mycroft Exp $ 35 */ 36 37#include <sys/param.h> 38#include <sys/systm.h> 39#include <sys/malloc.h> 40#include <sys/mbuf.h> 41#include <sys/domain.h> 42#include <sys/protosw.h> 43#include <sys/socket.h> 44#include <sys/errno.h> 45#include <sys/time.h> 46#include <sys/kernel.h> 47 48#include <net/if.h> 49#include <net/route.h> 50 51#include <netinet/in.h> 52#include <netinet/in_systm.h> 53#include <netinet/ip.h> 54#include <netinet/in_pcb.h> 55#include <netinet/in_var.h> 56#include <netinet/ip_var.h> 57#include <netinet/ip_icmp.h> 58#include <netinet/ip_mroute.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; 72#ifdef DIAGNOSTIC 73int ipprintfs = 0; 74#endif 75 76extern struct domain inetdomain; 77extern struct protosw inetsw[]; 78u_char ip_protox[IPPROTO_MAX]; 79int ipqmaxlen = IFQ_MAXLEN; 80struct in_ifaddr *in_ifaddr; /* first inet address */ 81 82/* 83 * We need to save the IP options in case a protocol wants to respond 84 * to an incoming packet over the same route if the packet got here 85 * using IP source routing. This allows connection establishment and 86 * maintenance when the remote end is on a network that is not known 87 * to us. 88 */ 89int ip_nhops = 0; 90static struct ip_srcrt { 91 struct in_addr dst; /* final destination */ 92 char nop; /* one NOP to align */ 93 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 94 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 95} ip_srcrt; 96 97#ifdef GATEWAY 98extern int if_index; 99u_long *ip_ifmatrix; 100#endif 101 102static void ip_forward __P((struct mbuf *, int)); 103static void save_rte __P((u_char *, struct in_addr)); 104 105/* 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_long); 130 if ((ip_ifmatrix = (u_long *) malloc(i, M_RTABLE, M_WAITOK)) == 0) 131 panic("no memory for ip_ifmatrix"); 132#endif 133} 134 135struct ip *ip_reass(); 136struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 137struct route ipforward_rt; 138 139/* 140 * Ip input routine. Checksum and byte swap header. If fragmented 141 * try to reassemble. Process options. Pass to next level. 142 */ 143void 144ipintr() 145{ 146 register struct ip *ip; 147 register struct mbuf *m; 148 register struct ipq *fp; 149 register struct in_ifaddr *ia; 150 int hlen, s; 151#ifdef PARANOID 152 static int busy = 0; 153 154 if (busy) 155 panic("ipintr: called recursively\n"); 156 ++busy; 157#endif 158next: 159 /* 160 * Get next datagram off input queue and get IP header 161 * in first mbuf. 162 */ 163 s = splimp(); 164 IF_DEQUEUE(&ipintrq, m); 165 splx(s); 166 if (m == 0) { 167#ifdef PARANOID 168 --busy; 169#endif 170 return; 171 } 172#ifdef DIAGNOSTIC 173 if ((m->m_flags & M_PKTHDR) == 0) 174 panic("ipintr no HDR"); 175#endif 176 /* 177 * If no IP addresses have been set yet but the interfaces 178 * are receiving, can't do anything with incoming packets yet. 179 */ 180 if (in_ifaddr == NULL) 181 goto bad; 182 ipstat.ips_total++; 183 if (m->m_len < sizeof (struct ip) && 184 (m = m_pullup(m, sizeof (struct ip))) == 0) { 185 ipstat.ips_toosmall++; 186 goto next; 187 } 188 ip = mtod(m, struct ip *); 189 hlen = ip->ip_hl << 2; 190 if (hlen < sizeof(struct ip)) { /* minimum header length */ 191 ipstat.ips_badhlen++; 192 goto bad; 193 } 194 if (hlen > m->m_len) { 195 if ((m = m_pullup(m, hlen)) == 0) { 196 ipstat.ips_badhlen++; 197 goto next; 198 } 199 ip = mtod(m, struct ip *); 200 } 201 if (ip->ip_sum = in_cksum(m, hlen)) { 202 ipstat.ips_badsum++; 203 goto bad; 204 } 205 206 /* 207 * Convert fields to host representation. 208 */ 209 NTOHS(ip->ip_len); 210 if (ip->ip_len < hlen) { 211 ipstat.ips_badlen++; 212 goto bad; 213 } 214 NTOHS(ip->ip_id); 215 NTOHS(ip->ip_off); 216 217 /* 218 * Check that the amount of data in the buffers 219 * is as at least much as the IP header would have us expect. 220 * Trim mbufs if longer than we expect. 221 * Drop packet if shorter than we expect. 222 */ 223 if (m->m_pkthdr.len < ip->ip_len) { 224 ipstat.ips_tooshort++; 225 goto bad; 226 } 227 if (m->m_pkthdr.len > ip->ip_len) { 228 if (m->m_len == m->m_pkthdr.len) { 229 m->m_len = ip->ip_len; 230 m->m_pkthdr.len = ip->ip_len; 231 } else 232 m_adj(m, ip->ip_len - m->m_pkthdr.len); 233 } 234 235 /* 236 * Process options and, if not destined for us, 237 * ship it on. ip_dooptions returns 1 when an 238 * error was detected (causing an icmp message 239 * to be sent and the original packet to be freed). 240 */ 241 ip_nhops = 0; /* for source routed packets */ 242 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 243 goto next; 244 245 /* 246 * Check our list of addresses, to see if the packet is for us. 247 */ 248 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 249#define satosin(sa) ((struct sockaddr_in *)(sa)) 250 251 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 252 goto ours; 253 if ( 254#ifdef DIRECTED_BROADCAST 255 ia->ia_ifp == m->m_pkthdr.rcvif && 256#endif 257 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 258 u_long t; 259 260 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 261 ip->ip_dst.s_addr) 262 goto ours; 263 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 264 goto ours; 265 /* 266 * Look for all-0's host part (old broadcast addr), 267 * either for subnet or net. 268 */ 269 t = ntohl(ip->ip_dst.s_addr); 270 if (t == ia->ia_subnet) 271 goto ours; 272 if (t == ia->ia_net) 273 goto ours; 274 } 275 } 276#ifdef MULTICAST 277 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 278 struct in_multi *inm; 279#ifdef MROUTING 280 extern struct socket *ip_mrouter; 281 282 if (ip_mrouter) { 283 /* 284 * If we are acting as a multicast router, all 285 * incoming multicast packets are passed to the 286 * kernel-level multicast forwarding function. 287 * The packet is returned (relatively) intact; if 288 * ip_mforward() returns a non-zero value, the packet 289 * must be discarded, else it may be accepted below. 290 * 291 * (The IP ident field is put in the same byte order 292 * as expected when ip_mforward() is called from 293 * ip_output().) 294 */ 295 ip->ip_id = htons(ip->ip_id); 296 if (ip_mforward(ip, m->m_pkthdr.rcvif, m) != 0) { 297 m_freem(m); 298 goto next; 299 } 300 ip->ip_id = ntohs(ip->ip_id); 301 302 /* 303 * The process-level routing demon needs to receive 304 * all multicast IGMP packets, whether or not this 305 * host belongs to their destination groups. 306 */ 307 if (ip->ip_p == IPPROTO_IGMP) 308 goto ours; 309 } 310#endif 311 /* 312 * See if we belong to the destination multicast group on the 313 * arrival interface. 314 */ 315 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 316 if (inm == NULL) { 317 m_freem(m); 318 goto next; 319 } 320 goto ours; 321 } 322#endif 323 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 324 goto ours; 325 if (ip->ip_dst.s_addr == INADDR_ANY) 326 goto ours; 327 328 /* 329 * Not for us; forward if possible and desirable. 330 */ 331 if (ipforwarding == 0) { 332 ipstat.ips_cantforward++; 333 m_freem(m); 334 } else 335 ip_forward(m, 0); 336 goto next; 337 338ours: 339 /* 340 * If offset or IP_MF are set, must reassemble. 341 * Otherwise, nothing need be done. 342 * (We could look in the reassembly queue to see 343 * if the packet was previously fragmented, 344 * but it's not worth the time; just let them time out.) 345 */ 346 if (ip->ip_off &~ IP_DF) { 347 if (m->m_flags & M_EXT) { /* XXX */ 348 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 349 ipstat.ips_toosmall++; 350 goto next; 351 } 352 ip = mtod(m, struct ip *); 353 } 354 /* 355 * Look for queue of fragments 356 * of this datagram. 357 */ 358 for (fp = ipq.next; fp != &ipq; fp = fp->next) 359 if (ip->ip_id == fp->ipq_id && 360 ip->ip_src.s_addr == fp->ipq_src.s_addr && 361 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 362 ip->ip_p == fp->ipq_p) 363 goto found; 364 fp = 0; 365found: 366 367 /* 368 * Adjust ip_len to not reflect header, 369 * set ip_mff if more fragments are expected, 370 * convert offset of this to bytes. 371 */ 372 ip->ip_len -= hlen; 373 ((struct ipasfrag *)ip)->ipf_mff = 0; 374 if (ip->ip_off & IP_MF) 375 ((struct ipasfrag *)ip)->ipf_mff = 1; 376 ip->ip_off <<= 3; 377 378 /* 379 * If datagram marked as having more fragments 380 * or if this is not the first fragment, 381 * attempt reassembly; if it succeeds, proceed. 382 */ 383 if (((struct ipasfrag *)ip)->ipf_mff || ip->ip_off) { 384 ipstat.ips_fragments++; 385 ip = ip_reass((struct ipasfrag *)ip, fp); 386 if (ip == 0) 387 goto next; 388 else 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) 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 ((struct ip *)ip)->ip_src = fp->ipq_src; 529 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 530 remque(fp); 531 (void) m_free(dtom(fp)); 532 m = dtom(ip); 533 m->m_len += (ip->ip_hl << 2); 534 m->m_data -= (ip->ip_hl << 2); 535 /* some debugging cruft by sklower, below, will go away soon */ 536 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 537 register int plen = 0; 538 for (t = m; m; m = m->m_next) 539 plen += m->m_len; 540 t->m_pkthdr.len = plen; 541 } 542 return ((struct ip *)ip); 543 544dropfrag: 545 ipstat.ips_fragdropped++; 546 m_freem(m); 547 return (0); 548} 549 550/* 551 * Free a fragment reassembly header and all 552 * associated datagrams. 553 */ 554void 555ip_freef(fp) 556 struct ipq *fp; 557{ 558 register struct ipasfrag *q, *p; 559 560 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 561 p = q->ipf_next; 562 ip_deq(q); 563 m_freem(dtom(q)); 564 } 565 remque(fp); 566 (void) m_free(dtom(fp)); 567} 568 569/* 570 * Put an ip fragment on a reassembly chain. 571 * Like insque, but pointers in middle of structure. 572 */ 573void 574ip_enq(p, prev) 575 register struct ipasfrag *p, *prev; 576{ 577 578 p->ipf_prev = prev; 579 p->ipf_next = prev->ipf_next; 580 prev->ipf_next->ipf_prev = p; 581 prev->ipf_next = p; 582} 583 584/* 585 * To ip_enq as remque is to insque. 586 */ 587void 588ip_deq(p) 589 register struct ipasfrag *p; 590{ 591 592 p->ipf_prev->ipf_next = p->ipf_next; 593 p->ipf_next->ipf_prev = p->ipf_prev; 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; 605 int s = splnet(); 606 607 fp = ipq.next; 608 if (fp == 0) { 609 splx(s); 610 return; 611 } 612 while (fp != &ipq) { 613 --fp->ipq_ttl; 614 fp = fp->next; 615 if (fp->prev->ipq_ttl == 0) { 616 ipstat.ips_fragtimeout++; 617 ip_freef(fp->prev); 618 } 619 } 620 splx(s); 621} 622 623/* 624 * Drain off all datagram fragments. 625 */ 626void 627ip_drain() 628{ 629 630 while (ipq.next != &ipq) { 631 ipstat.ips_fragdropped++; 632 ip_freef(ipq.next); 633 } 634} 635 636extern struct in_ifaddr *ifptoia(); 637struct in_ifaddr *ip_rtaddr(); 638 639/* 640 * Do option processing on a datagram, 641 * possibly discarding it if bad options are encountered, 642 * or forwarding it if source-routed. 643 * Returns 1 if packet has been forwarded/freed, 644 * 0 if the packet should be processed further. 645 */ 646int 647ip_dooptions(m) 648 struct mbuf *m; 649{ 650 register struct ip *ip = mtod(m, struct ip *); 651 register u_char *cp; 652 register struct ip_timestamp *ipt; 653 register struct in_ifaddr *ia; 654 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 655 struct in_addr *sin; 656 n_time ntime; 657 658 cp = (u_char *)(ip + 1); 659 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 660 for (; cnt > 0; cnt -= optlen, cp += optlen) { 661 opt = cp[IPOPT_OPTVAL]; 662 if (opt == IPOPT_EOL) 663 break; 664 if (opt == IPOPT_NOP) 665 optlen = 1; 666 else { 667 optlen = cp[IPOPT_OLEN]; 668 if (optlen <= 0 || optlen > cnt) { 669 code = &cp[IPOPT_OLEN] - (u_char *)ip; 670 goto bad; 671 } 672 } 673 switch (opt) { 674 675 default: 676 break; 677 678 /* 679 * Source routing with record. 680 * Find interface with current destination address. 681 * If none on this machine then drop if strictly routed, 682 * or do nothing if loosely routed. 683 * Record interface address and bring up next address 684 * component. If strictly routed make sure next 685 * address is on directly accessible net. 686 */ 687 case IPOPT_LSRR: 688 case IPOPT_SSRR: 689 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 690 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 691 goto bad; 692 } 693 ipaddr.sin_addr = ip->ip_dst; 694 ia = (struct in_ifaddr *) 695 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 696 if (ia == 0) { 697 if (opt == IPOPT_SSRR) { 698 type = ICMP_UNREACH; 699 code = ICMP_UNREACH_SRCFAIL; 700 goto bad; 701 } 702 /* 703 * Loose routing, and not at next destination 704 * yet; nothing to do except forward. 705 */ 706 break; 707 } 708 off--; /* 0 origin */ 709 if (off > optlen - sizeof(struct in_addr)) { 710 /* 711 * End of source route. Should be for us. 712 */ 713 save_rte(cp, ip->ip_src); 714 break; 715 } 716 /* 717 * locate outgoing interface 718 */ 719 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 720 sizeof(ipaddr.sin_addr)); 721 if (opt == IPOPT_SSRR) { 722#define INA struct in_ifaddr * 723#define SA struct sockaddr * 724 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 725 ia = in_iaonnetof(in_netof(ipaddr.sin_addr)); 726 } else 727 ia = ip_rtaddr(ipaddr.sin_addr); 728 if (ia == 0) { 729 type = ICMP_UNREACH; 730 code = ICMP_UNREACH_SRCFAIL; 731 goto bad; 732 } 733 ip->ip_dst = ipaddr.sin_addr; 734 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 735 (caddr_t)(cp + off), sizeof(struct in_addr)); 736 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 737 forward = 1; 738 break; 739 740 case IPOPT_RR: 741 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 742 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 743 goto bad; 744 } 745 /* 746 * If no space remains, ignore. 747 */ 748 off--; /* 0 origin */ 749 if (off > optlen - sizeof(struct in_addr)) 750 break; 751 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 752 sizeof(ipaddr.sin_addr)); 753 /* 754 * locate outgoing interface; if we're the destination, 755 * use the incoming interface (should be same). 756 */ 757 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 758 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 759 type = ICMP_UNREACH; 760 code = ICMP_UNREACH_HOST; 761 goto bad; 762 } 763 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 764 (caddr_t)(cp + off), sizeof(struct in_addr)); 765 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 766 break; 767 768 case IPOPT_TS: 769 code = cp - (u_char *)ip; 770 ipt = (struct ip_timestamp *)cp; 771 if (ipt->ipt_len < 5) 772 goto bad; 773 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 774 if (++ipt->ipt_oflw == 0) 775 goto bad; 776 break; 777 } 778 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 779 switch (ipt->ipt_flg) { 780 781 case IPOPT_TS_TSONLY: 782 break; 783 784 case IPOPT_TS_TSANDADDR: 785 if (ipt->ipt_ptr + sizeof(n_time) + 786 sizeof(struct in_addr) > ipt->ipt_len) 787 goto bad; 788 ia = ifptoia(m->m_pkthdr.rcvif); 789 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 790 (caddr_t)sin, sizeof(struct in_addr)); 791 ipt->ipt_ptr += sizeof(struct in_addr); 792 break; 793 794 case IPOPT_TS_PRESPEC: 795 if (ipt->ipt_ptr + sizeof(n_time) + 796 sizeof(struct in_addr) > ipt->ipt_len) 797 goto bad; 798 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 799 sizeof(struct in_addr)); 800 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 801 continue; 802 ipt->ipt_ptr += sizeof(struct in_addr); 803 break; 804 805 default: 806 goto bad; 807 } 808 ntime = iptime(); 809 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 810 sizeof(n_time)); 811 ipt->ipt_ptr += sizeof(n_time); 812 } 813 } 814 if (forward) { 815 ip_forward(m, 1); 816 return (1); 817 } else 818 return (0); 819bad: 820 { 821 register struct in_addr foo = {}; 822 icmp_error(m, type, code, foo); 823 } 824 return (1); 825} 826 827/* 828 * Given address of next destination (final or next hop), 829 * return internet address info of interface to be used to get there. 830 */ 831struct in_ifaddr * 832ip_rtaddr(dst) 833 struct in_addr dst; 834{ 835 register struct sockaddr_in *sin; 836 837 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 838 839 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 840 if (ipforward_rt.ro_rt) { 841 RTFREE(ipforward_rt.ro_rt); 842 ipforward_rt.ro_rt = 0; 843 } 844 sin->sin_family = AF_INET; 845 sin->sin_len = sizeof(*sin); 846 sin->sin_addr = dst; 847 848 rtalloc(&ipforward_rt); 849 } 850 if (ipforward_rt.ro_rt == 0) 851 return ((struct in_ifaddr *)0); 852 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 853} 854 855/* 856 * Save incoming source route for use in replies, 857 * to be picked up later by ip_srcroute if the receiver is interested. 858 */ 859static void 860save_rte(option, dst) 861 u_char *option; 862 struct in_addr dst; 863{ 864 unsigned olen; 865 866 olen = option[IPOPT_OLEN]; 867#ifdef DIAGNOSTIC 868 if (ipprintfs) 869 printf("save_rte: olen %d\n", olen); 870#endif 871 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 872 return; 873 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 874 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 875 ip_srcrt.dst = dst; 876} 877 878/* 879 * Retrieve incoming source route for use in replies, 880 * in the same form used by setsockopt. 881 * The first hop is placed before the options, will be removed later. 882 */ 883struct mbuf * 884ip_srcroute() 885{ 886 register struct in_addr *p, *q; 887 register struct mbuf *m; 888 889 if (ip_nhops == 0) 890 return ((struct mbuf *)0); 891 m = m_get(M_DONTWAIT, MT_SOOPTS); 892 if (m == 0) 893 return ((struct mbuf *)0); 894 895#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 896 897 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 898 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 899 OPTSIZ; 900#ifdef DIAGNOSTIC 901 if (ipprintfs) 902 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 903#endif 904 905 /* 906 * First save first hop for return route 907 */ 908 p = &ip_srcrt.route[ip_nhops - 1]; 909 *(mtod(m, struct in_addr *)) = *p--; 910#ifdef DIAGNOSTIC 911 if (ipprintfs) 912 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 913#endif 914 915 /* 916 * Copy option fields and padding (nop) to mbuf. 917 */ 918 ip_srcrt.nop = IPOPT_NOP; 919 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 920 bcopy((caddr_t)&ip_srcrt.nop, 921 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 922 q = (struct in_addr *)(mtod(m, caddr_t) + 923 sizeof(struct in_addr) + OPTSIZ); 924#undef OPTSIZ 925 /* 926 * Record return path as an IP source route, 927 * reversing the path (pointers are now aligned). 928 */ 929 while (p >= ip_srcrt.route) { 930#ifdef DIAGNOSTIC 931 if (ipprintfs) 932 printf(" %lx", ntohl(q->s_addr)); 933#endif 934 *q++ = *p--; 935 } 936 /* 937 * Last hop goes to final destination. 938 */ 939 *q = ip_srcrt.dst; 940#ifdef DIAGNOSTIC 941 if (ipprintfs) 942 printf(" %lx\n", ntohl(q->s_addr)); 943#endif 944 return (m); 945} 946 947/* 948 * Strip out IP options, at higher 949 * level protocol in the kernel. 950 * Second argument is buffer to which options 951 * will be moved, and return value is their length. 952 * XXX should be deleted; last arg currently ignored. 953 */ 954void 955ip_stripoptions(m, mopt) 956 register struct mbuf *m; 957 struct mbuf *mopt; 958{ 959 register int i; 960 struct ip *ip = mtod(m, struct ip *); 961 register caddr_t opts; 962 int olen; 963 964 olen = (ip->ip_hl<<2) - sizeof (struct ip); 965 opts = (caddr_t)(ip + 1); 966 i = m->m_len - (sizeof (struct ip) + olen); 967 bcopy(opts + olen, opts, (unsigned)i); 968 m->m_len -= olen; 969 if (m->m_flags & M_PKTHDR) 970 m->m_pkthdr.len -= olen; 971 ip->ip_hl = sizeof(struct ip) >> 2; 972} 973 974u_char inetctlerrmap[PRC_NCMDS] = { 975 0, 0, 0, 0, 976 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 977 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 978 EMSGSIZE, EHOSTUNREACH, 0, 0, 979 0, 0, 0, 0, 980 ENOPROTOOPT 981}; 982 983/* 984 * Forward a packet. If some error occurs return the sender 985 * an icmp packet. Note we can't always generate a meaningful 986 * icmp message because icmp doesn't have a large enough repertoire 987 * of codes and types. 988 * 989 * If not forwarding, just drop the packet. This could be confusing 990 * if ipforwarding was zero but some routing protocol was advancing 991 * us as a gateway to somewhere. However, we must let the routing 992 * protocol deal with that. 993 * 994 * The srcrt parameter indicates whether the packet is being forwarded 995 * via a source route. 996 */ 997static void 998ip_forward(m, srcrt) 999 struct mbuf *m; 1000 int srcrt; 1001{ 1002 register struct ip *ip = mtod(m, struct ip *); 1003 register struct sockaddr_in *sin; 1004 register struct rtentry *rt; 1005 int error, type = 0, code; 1006 struct mbuf *mcopy; 1007 struct in_addr dest; 1008 1009 dest.s_addr = 0; 1010#ifdef DIAGNOSTIC 1011 if (ipprintfs) 1012 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1013 ip->ip_dst, ip->ip_ttl); 1014#endif 1015 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1016 ipstat.ips_cantforward++; 1017 m_freem(m); 1018 return; 1019 } 1020 HTONS(ip->ip_id); 1021 if (ip->ip_ttl <= IPTTLDEC) { 1022 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest); 1023 return; 1024 } 1025 ip->ip_ttl -= IPTTLDEC; 1026 1027 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1028 if ((rt = ipforward_rt.ro_rt) == 0 || 1029 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1030 if (ipforward_rt.ro_rt) { 1031 RTFREE(ipforward_rt.ro_rt); 1032 ipforward_rt.ro_rt = 0; 1033 } 1034 sin->sin_family = AF_INET; 1035 sin->sin_len = sizeof(*sin); 1036 sin->sin_addr = ip->ip_dst; 1037 1038 rtalloc(&ipforward_rt); 1039 if (ipforward_rt.ro_rt == 0) { 1040 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest); 1041 return; 1042 } 1043 rt = ipforward_rt.ro_rt; 1044 } 1045 1046 /* 1047 * Save at most 64 bytes of the packet in case 1048 * we need to generate an ICMP message to the src. 1049 */ 1050 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1051 1052#ifdef GATEWAY 1053 ip_ifmatrix[rt->rt_ifp->if_index + 1054 if_index * m->m_pkthdr.rcvif->if_index]++; 1055#endif 1056 /* 1057 * If forwarding packet using same interface that it came in on, 1058 * perhaps should send a redirect to sender to shortcut a hop. 1059 * Only send redirect if source is sending directly to us, 1060 * and if packet was not source routed (or has any options). 1061 * Also, don't send redirect if forwarding using a default route 1062 * or a route modified by a redirect. 1063 */ 1064#define satosin(sa) ((struct sockaddr_in *)(sa)) 1065 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1066 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1067 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1068 ipsendredirects && !srcrt) { 1069 struct in_ifaddr *ia; 1070 u_long src = ntohl(ip->ip_src.s_addr); 1071 u_long dst = ntohl(ip->ip_dst.s_addr); 1072 1073 if ((ia = ifptoia(m->m_pkthdr.rcvif)) && 1074 (src & ia->ia_subnetmask) == ia->ia_subnet) { 1075 if (rt->rt_flags & RTF_GATEWAY) 1076 dest = satosin(rt->rt_gateway)->sin_addr; 1077 else 1078 dest = ip->ip_dst; 1079 /* 1080 * If the destination is reached by a route to host, 1081 * is on a subnet of a local net, or is directly 1082 * on the attached net (!), use host redirect. 1083 * (We may be the correct first hop for other subnets.) 1084 */ 1085#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1086 type = ICMP_REDIRECT; 1087 if ((rt->rt_flags & RTF_HOST) || 1088 (rt->rt_flags & RTF_GATEWAY) == 0) 1089 code = ICMP_REDIRECT_HOST; 1090 else if (RTA(rt)->ia_subnetmask != RTA(rt)->ia_netmask && 1091 (dst & RTA(rt)->ia_netmask) == RTA(rt)->ia_net) 1092 code = ICMP_REDIRECT_HOST; 1093 else 1094 code = ICMP_REDIRECT_NET; 1095#ifdef DIAGNOSTIC 1096 if (ipprintfs) 1097 printf("redirect (%d) to %x\n", code, dest.s_addr); 1098#endif 1099 } 1100 } 1101 1102 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING); 1103 if (error) 1104 ipstat.ips_cantforward++; 1105 else { 1106 ipstat.ips_forward++; 1107 if (type) 1108 ipstat.ips_redirectsent++; 1109 else { 1110 if (mcopy) 1111 m_freem(mcopy); 1112 return; 1113 } 1114 } 1115 if (mcopy == NULL) 1116 return; 1117 switch (error) { 1118 1119 case 0: /* forwarded, but need redirect */ 1120 /* type, code set above */ 1121 break; 1122 1123 case ENETUNREACH: /* shouldn't happen, checked above */ 1124 case EHOSTUNREACH: 1125 case ENETDOWN: 1126 case EHOSTDOWN: 1127 default: 1128 type = ICMP_UNREACH; 1129 code = ICMP_UNREACH_HOST; 1130 break; 1131 1132 case EMSGSIZE: 1133 type = ICMP_UNREACH; 1134 code = ICMP_UNREACH_NEEDFRAG; 1135 ipstat.ips_cantfrag++; 1136 break; 1137 1138 case ENOBUFS: 1139 type = ICMP_SOURCEQUENCH; 1140 code = 0; 1141 break; 1142 } 1143 icmp_error(mcopy, type, code, dest); 1144} 1145