ip_input.c revision 1.16
1/* $NetBSD: ip_input.c,v 1.16 1995/05/14 08:23:00 cgd 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#define satosin(sa) ((struct sockaddr_in *)(sa)) 243 244 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 245 goto ours; 246 if ( 247#ifdef DIRECTED_BROADCAST 248 ia->ia_ifp == m->m_pkthdr.rcvif && 249#endif 250 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 251 u_int32_t t; 252 253 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 254 ip->ip_dst.s_addr) 255 goto ours; 256 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 257 goto ours; 258 /* 259 * Look for all-0's host part (old broadcast addr), 260 * either for subnet or net. 261 */ 262 t = ntohl(ip->ip_dst.s_addr); 263 if (t == ia->ia_subnet) 264 goto ours; 265 if (t == ia->ia_net) 266 goto ours; 267 } 268 } 269 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 270 struct in_multi *inm; 271#ifdef MROUTING 272 extern struct socket *ip_mrouter; 273 274 if (m->m_flags & M_EXT) { 275 if ((m = m_pullup(m, hlen)) == 0) { 276 ipstat.ips_toosmall++; 277 goto next; 278 } 279 ip = mtod(m, struct ip *); 280 } 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(m, m->m_pkthdr.rcvif) != 0) { 297 ipstat.ips_cantforward++; 298 m_freem(m); 299 goto next; 300 } 301 ip->ip_id = ntohs(ip->ip_id); 302 303 /* 304 * The process-level routing demon needs to receive 305 * all multicast IGMP packets, whether or not this 306 * host belongs to their destination groups. 307 */ 308 if (ip->ip_p == IPPROTO_IGMP) 309 goto ours; 310 ipstat.ips_forward++; 311 } 312#endif 313 /* 314 * See if we belong to the destination multicast group on the 315 * arrival interface. 316 */ 317 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 318 if (inm == NULL) { 319 ipstat.ips_cantforward++; 320 m_freem(m); 321 goto next; 322 } 323 goto ours; 324 } 325 if (ip->ip_dst.s_addr == (u_int32_t)INADDR_BROADCAST) 326 goto ours; 327 if (ip->ip_dst.s_addr == INADDR_ANY) 328 goto ours; 329 330 /* 331 * Not for us; forward if possible and desirable. 332 */ 333 if (ipforwarding == 0) { 334 ipstat.ips_cantforward++; 335 m_freem(m); 336 } else 337 ip_forward(m, 0); 338 goto next; 339 340ours: 341 /* 342 * If offset or IP_MF are set, must reassemble. 343 * Otherwise, nothing need be done. 344 * (We could look in the reassembly queue to see 345 * if the packet was previously fragmented, 346 * but it's not worth the time; just let them time out.) 347 */ 348 if (ip->ip_off &~ IP_DF) { 349 if (m->m_flags & M_EXT) { /* XXX */ 350 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 351 ipstat.ips_toosmall++; 352 goto next; 353 } 354 ip = mtod(m, struct ip *); 355 } 356 /* 357 * Look for queue of fragments 358 * of this datagram. 359 */ 360 for (fp = ipq.next; fp != &ipq; fp = fp->next) 361 if (ip->ip_id == fp->ipq_id && 362 ip->ip_src.s_addr == fp->ipq_src.s_addr && 363 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 364 ip->ip_p == fp->ipq_p) 365 goto found; 366 fp = 0; 367found: 368 369 /* 370 * Adjust ip_len to not reflect header, 371 * set ip_mff if more fragments are expected, 372 * convert offset of this to bytes. 373 */ 374 ip->ip_len -= hlen; 375 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 376 if (ip->ip_off & IP_MF) { 377 /* 378 * Make sure that fragments have a data length 379 * that's a non-zero multiple of 8 bytes. 380 */ 381 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) 382 ipstat.ips_badfrags++; 383 goto bad; 384 } 385 ((struct ipasfrag *)ip)->ipf_mff |= 1; 386 } 387 ip->ip_off <<= 3; 388 389 /* 390 * If datagram marked as having more fragments 391 * or if this is not the first fragment, 392 * attempt reassembly; if it succeeds, proceed. 393 */ 394 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 395 ipstat.ips_fragments++; 396 ip = ip_reass((struct ipasfrag *)ip, fp); 397 if (ip == 0) 398 goto next; 399 ipstat.ips_reassembled++; 400 m = dtom(ip); 401 } else 402 if (fp) 403 ip_freef(fp); 404 } else 405 ip->ip_len -= hlen; 406 407 /* 408 * Switch out to protocol's input routine. 409 */ 410 ipstat.ips_delivered++; 411 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 412 goto next; 413bad: 414 m_freem(m); 415 goto next; 416} 417 418/* 419 * Take incoming datagram fragment and try to 420 * reassemble it into whole datagram. If a chain for 421 * reassembly of this datagram already exists, then it 422 * is given as fp; otherwise have to make a chain. 423 */ 424struct ip * 425ip_reass(ip, fp) 426 register struct ipasfrag *ip; 427 register struct ipq *fp; 428{ 429 register struct mbuf *m = dtom(ip); 430 register struct ipasfrag *q; 431 struct mbuf *t; 432 int hlen = ip->ip_hl << 2; 433 int i, next; 434 435 /* 436 * Presence of header sizes in mbufs 437 * would confuse code below. 438 */ 439 m->m_data += hlen; 440 m->m_len -= hlen; 441 442 /* 443 * If first fragment to arrive, create a reassembly queue. 444 */ 445 if (fp == 0) { 446 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 447 goto dropfrag; 448 fp = mtod(t, struct ipq *); 449 insque(fp, &ipq); 450 fp->ipq_ttl = IPFRAGTTL; 451 fp->ipq_p = ip->ip_p; 452 fp->ipq_id = ip->ip_id; 453 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 454 fp->ipq_src = ((struct ip *)ip)->ip_src; 455 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 456 q = (struct ipasfrag *)fp; 457 goto insert; 458 } 459 460 /* 461 * Find a segment which begins after this one does. 462 */ 463 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 464 if (q->ip_off > ip->ip_off) 465 break; 466 467 /* 468 * If there is a preceding segment, it may provide some of 469 * our data already. If so, drop the data from the incoming 470 * segment. If it provides all of our data, drop us. 471 */ 472 if (q->ipf_prev != (struct ipasfrag *)fp) { 473 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 474 if (i > 0) { 475 if (i >= ip->ip_len) 476 goto dropfrag; 477 m_adj(dtom(ip), i); 478 ip->ip_off += i; 479 ip->ip_len -= i; 480 } 481 } 482 483 /* 484 * While we overlap succeeding segments trim them or, 485 * if they are completely covered, dequeue them. 486 */ 487 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 488 i = (ip->ip_off + ip->ip_len) - q->ip_off; 489 if (i < q->ip_len) { 490 q->ip_len -= i; 491 q->ip_off += i; 492 m_adj(dtom(q), i); 493 break; 494 } 495 q = q->ipf_next; 496 m_freem(dtom(q->ipf_prev)); 497 ip_deq(q->ipf_prev); 498 } 499 500insert: 501 /* 502 * Stick new segment in its place; 503 * check for complete reassembly. 504 */ 505 ip_enq(ip, q->ipf_prev); 506 next = 0; 507 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 508 if (q->ip_off != next) 509 return (0); 510 next += q->ip_len; 511 } 512 if (q->ipf_prev->ipf_mff & 1) 513 return (0); 514 515 /* 516 * Reassembly is complete; concatenate fragments. 517 */ 518 q = fp->ipq_next; 519 m = dtom(q); 520 t = m->m_next; 521 m->m_next = 0; 522 m_cat(m, t); 523 q = q->ipf_next; 524 while (q != (struct ipasfrag *)fp) { 525 t = dtom(q); 526 q = q->ipf_next; 527 m_cat(m, t); 528 } 529 530 /* 531 * Create header for new ip packet by 532 * modifying header of first packet; 533 * dequeue and discard fragment reassembly header. 534 * Make header visible. 535 */ 536 ip = fp->ipq_next; 537 ip->ip_len = next; 538 ip->ipf_mff &= ~1; 539 ((struct ip *)ip)->ip_src = fp->ipq_src; 540 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 541 remque(fp); 542 (void) m_free(dtom(fp)); 543 m = dtom(ip); 544 m->m_len += (ip->ip_hl << 2); 545 m->m_data -= (ip->ip_hl << 2); 546 /* some debugging cruft by sklower, below, will go away soon */ 547 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 548 register int plen = 0; 549 for (t = m; m; m = m->m_next) 550 plen += m->m_len; 551 t->m_pkthdr.len = plen; 552 } 553 return ((struct ip *)ip); 554 555dropfrag: 556 ipstat.ips_fragdropped++; 557 m_freem(m); 558 return (0); 559} 560 561/* 562 * Free a fragment reassembly header and all 563 * associated datagrams. 564 */ 565void 566ip_freef(fp) 567 struct ipq *fp; 568{ 569 register struct ipasfrag *q, *p; 570 571 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 572 p = q->ipf_next; 573 ip_deq(q); 574 m_freem(dtom(q)); 575 } 576 remque(fp); 577 (void) m_free(dtom(fp)); 578} 579 580/* 581 * Put an ip fragment on a reassembly chain. 582 * Like insque, but pointers in middle of structure. 583 */ 584void 585ip_enq(p, prev) 586 register struct ipasfrag *p, *prev; 587{ 588 589 p->ipf_prev = prev; 590 p->ipf_next = prev->ipf_next; 591 prev->ipf_next->ipf_prev = p; 592 prev->ipf_next = p; 593} 594 595/* 596 * To ip_enq as remque is to insque. 597 */ 598void 599ip_deq(p) 600 register struct ipasfrag *p; 601{ 602 603 p->ipf_prev->ipf_next = p->ipf_next; 604 p->ipf_next->ipf_prev = p->ipf_prev; 605} 606 607/* 608 * IP timer processing; 609 * if a timer expires on a reassembly 610 * queue, discard it. 611 */ 612void 613ip_slowtimo() 614{ 615 register struct ipq *fp; 616 int s = splnet(); 617 618 fp = ipq.next; 619 if (fp == 0) { 620 splx(s); 621 return; 622 } 623 while (fp != &ipq) { 624 --fp->ipq_ttl; 625 fp = fp->next; 626 if (fp->prev->ipq_ttl == 0) { 627 ipstat.ips_fragtimeout++; 628 ip_freef(fp->prev); 629 } 630 } 631 splx(s); 632} 633 634/* 635 * Drain off all datagram fragments. 636 */ 637void 638ip_drain() 639{ 640 641 while (ipq.next != &ipq) { 642 ipstat.ips_fragdropped++; 643 ip_freef(ipq.next); 644 } 645} 646 647/* 648 * Do option processing on a datagram, 649 * possibly discarding it if bad options are encountered, 650 * or forwarding it if source-routed. 651 * Returns 1 if packet has been forwarded/freed, 652 * 0 if the packet should be processed further. 653 */ 654int 655ip_dooptions(m) 656 struct mbuf *m; 657{ 658 register struct ip *ip = mtod(m, struct ip *); 659 register u_char *cp; 660 register struct ip_timestamp *ipt; 661 register struct in_ifaddr *ia; 662 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 663 struct in_addr *sin, dst; 664 n_time ntime; 665 666 dst = ip->ip_dst; 667 cp = (u_char *)(ip + 1); 668 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 669 for (; cnt > 0; cnt -= optlen, cp += optlen) { 670 opt = cp[IPOPT_OPTVAL]; 671 if (opt == IPOPT_EOL) 672 break; 673 if (opt == IPOPT_NOP) 674 optlen = 1; 675 else { 676 optlen = cp[IPOPT_OLEN]; 677 if (optlen <= 0 || optlen > cnt) { 678 code = &cp[IPOPT_OLEN] - (u_char *)ip; 679 goto bad; 680 } 681 } 682 switch (opt) { 683 684 default: 685 break; 686 687 /* 688 * Source routing with record. 689 * Find interface with current destination address. 690 * If none on this machine then drop if strictly routed, 691 * or do nothing if loosely routed. 692 * Record interface address and bring up next address 693 * component. If strictly routed make sure next 694 * address is on directly accessible net. 695 */ 696 case IPOPT_LSRR: 697 case IPOPT_SSRR: 698 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 699 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 700 goto bad; 701 } 702 ipaddr.sin_addr = ip->ip_dst; 703 ia = (struct in_ifaddr *) 704 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 705 if (ia == 0) { 706 if (opt == IPOPT_SSRR) { 707 type = ICMP_UNREACH; 708 code = ICMP_UNREACH_SRCFAIL; 709 goto bad; 710 } 711 /* 712 * Loose routing, and not at next destination 713 * yet; nothing to do except forward. 714 */ 715 break; 716 } 717 off--; /* 0 origin */ 718 if (off > optlen - sizeof(struct in_addr)) { 719 /* 720 * End of source route. Should be for us. 721 */ 722 save_rte(cp, ip->ip_src); 723 break; 724 } 725 /* 726 * locate outgoing interface 727 */ 728 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 729 sizeof(ipaddr.sin_addr)); 730 if (opt == IPOPT_SSRR) { 731#define INA struct in_ifaddr * 732#define SA struct sockaddr * 733 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 734 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 735 } else 736 ia = ip_rtaddr(ipaddr.sin_addr); 737 if (ia == 0) { 738 type = ICMP_UNREACH; 739 code = ICMP_UNREACH_SRCFAIL; 740 goto bad; 741 } 742 ip->ip_dst = ipaddr.sin_addr; 743 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 744 (caddr_t)(cp + off), sizeof(struct in_addr)); 745 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 746 /* 747 * Let ip_intr's mcast routing check handle mcast pkts 748 */ 749 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 750 break; 751 752 case IPOPT_RR: 753 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 754 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 755 goto bad; 756 } 757 /* 758 * If no space remains, ignore. 759 */ 760 off--; /* 0 origin */ 761 if (off > optlen - sizeof(struct in_addr)) 762 break; 763 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 764 sizeof(ipaddr.sin_addr)); 765 /* 766 * locate outgoing interface; if we're the destination, 767 * use the incoming interface (should be same). 768 */ 769 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 770 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 771 type = ICMP_UNREACH; 772 code = ICMP_UNREACH_HOST; 773 goto bad; 774 } 775 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 776 (caddr_t)(cp + off), sizeof(struct in_addr)); 777 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 778 break; 779 780 case IPOPT_TS: 781 code = cp - (u_char *)ip; 782 ipt = (struct ip_timestamp *)cp; 783 if (ipt->ipt_len < 5) 784 goto bad; 785 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 786 if (++ipt->ipt_oflw == 0) 787 goto bad; 788 break; 789 } 790 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 791 switch (ipt->ipt_flg) { 792 793 case IPOPT_TS_TSONLY: 794 break; 795 796 case IPOPT_TS_TSANDADDR: 797 if (ipt->ipt_ptr + sizeof(n_time) + 798 sizeof(struct in_addr) > ipt->ipt_len) 799 goto bad; 800 ipaddr.sin_addr = dst; 801 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 802 m->m_pkthdr.rcvif); 803 if (ia == 0) 804 continue; 805 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 806 (caddr_t)sin, sizeof(struct in_addr)); 807 ipt->ipt_ptr += sizeof(struct in_addr); 808 break; 809 810 case IPOPT_TS_PRESPEC: 811 if (ipt->ipt_ptr + sizeof(n_time) + 812 sizeof(struct in_addr) > ipt->ipt_len) 813 goto bad; 814 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 815 sizeof(struct in_addr)); 816 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 817 continue; 818 ipt->ipt_ptr += sizeof(struct in_addr); 819 break; 820 821 default: 822 goto bad; 823 } 824 ntime = iptime(); 825 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 826 sizeof(n_time)); 827 ipt->ipt_ptr += sizeof(n_time); 828 } 829 } 830 if (forward) { 831 ip_forward(m, 1); 832 return (1); 833 } 834 return (0); 835bad: 836 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 837 icmp_error(m, type, code, 0, 0); 838 ipstat.ips_badoptions++; 839 return (1); 840} 841 842/* 843 * Given address of next destination (final or next hop), 844 * return internet address info of interface to be used to get there. 845 */ 846struct in_ifaddr * 847ip_rtaddr(dst) 848 struct in_addr dst; 849{ 850 register struct sockaddr_in *sin; 851 852 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 853 854 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 855 if (ipforward_rt.ro_rt) { 856 RTFREE(ipforward_rt.ro_rt); 857 ipforward_rt.ro_rt = 0; 858 } 859 sin->sin_family = AF_INET; 860 sin->sin_len = sizeof(*sin); 861 sin->sin_addr = dst; 862 863 rtalloc(&ipforward_rt); 864 } 865 if (ipforward_rt.ro_rt == 0) 866 return ((struct in_ifaddr *)0); 867 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 868} 869 870/* 871 * Save incoming source route for use in replies, 872 * to be picked up later by ip_srcroute if the receiver is interested. 873 */ 874void 875save_rte(option, dst) 876 u_char *option; 877 struct in_addr dst; 878{ 879 unsigned olen; 880 881 olen = option[IPOPT_OLEN]; 882#ifdef DIAGNOSTIC 883 if (ipprintfs) 884 printf("save_rte: olen %d\n", olen); 885#endif 886 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 887 return; 888 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 889 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 890 ip_srcrt.dst = dst; 891} 892 893/* 894 * Retrieve incoming source route for use in replies, 895 * in the same form used by setsockopt. 896 * The first hop is placed before the options, will be removed later. 897 */ 898struct mbuf * 899ip_srcroute() 900{ 901 register struct in_addr *p, *q; 902 register struct mbuf *m; 903 904 if (ip_nhops == 0) 905 return ((struct mbuf *)0); 906 m = m_get(M_DONTWAIT, MT_SOOPTS); 907 if (m == 0) 908 return ((struct mbuf *)0); 909 910#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 911 912 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 913 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 914 OPTSIZ; 915#ifdef DIAGNOSTIC 916 if (ipprintfs) 917 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 918#endif 919 920 /* 921 * First save first hop for return route 922 */ 923 p = &ip_srcrt.route[ip_nhops - 1]; 924 *(mtod(m, struct in_addr *)) = *p--; 925#ifdef DIAGNOSTIC 926 if (ipprintfs) 927 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 928#endif 929 930 /* 931 * Copy option fields and padding (nop) to mbuf. 932 */ 933 ip_srcrt.nop = IPOPT_NOP; 934 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 935 bcopy((caddr_t)&ip_srcrt.nop, 936 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 937 q = (struct in_addr *)(mtod(m, caddr_t) + 938 sizeof(struct in_addr) + OPTSIZ); 939#undef OPTSIZ 940 /* 941 * Record return path as an IP source route, 942 * reversing the path (pointers are now aligned). 943 */ 944 while (p >= ip_srcrt.route) { 945#ifdef DIAGNOSTIC 946 if (ipprintfs) 947 printf(" %lx", ntohl(q->s_addr)); 948#endif 949 *q++ = *p--; 950 } 951 /* 952 * Last hop goes to final destination. 953 */ 954 *q = ip_srcrt.dst; 955#ifdef DIAGNOSTIC 956 if (ipprintfs) 957 printf(" %lx\n", ntohl(q->s_addr)); 958#endif 959 return (m); 960} 961 962/* 963 * Strip out IP options, at higher 964 * level protocol in the kernel. 965 * Second argument is buffer to which options 966 * will be moved, and return value is their length. 967 * XXX should be deleted; last arg currently ignored. 968 */ 969void 970ip_stripoptions(m, mopt) 971 register struct mbuf *m; 972 struct mbuf *mopt; 973{ 974 register int i; 975 struct ip *ip = mtod(m, struct ip *); 976 register caddr_t opts; 977 int olen; 978 979 olen = (ip->ip_hl<<2) - sizeof (struct ip); 980 opts = (caddr_t)(ip + 1); 981 i = m->m_len - (sizeof (struct ip) + olen); 982 bcopy(opts + olen, opts, (unsigned)i); 983 m->m_len -= olen; 984 if (m->m_flags & M_PKTHDR) 985 m->m_pkthdr.len -= olen; 986 ip->ip_hl = sizeof(struct ip) >> 2; 987} 988 989u_char inetctlerrmap[PRC_NCMDS] = { 990 0, 0, 0, 0, 991 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 992 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 993 EMSGSIZE, EHOSTUNREACH, 0, 0, 994 0, 0, 0, 0, 995 ENOPROTOOPT 996}; 997 998/* 999 * Forward a packet. If some error occurs return the sender 1000 * an icmp packet. Note we can't always generate a meaningful 1001 * icmp message because icmp doesn't have a large enough repertoire 1002 * of codes and types. 1003 * 1004 * If not forwarding, just drop the packet. This could be confusing 1005 * if ipforwarding was zero but some routing protocol was advancing 1006 * us as a gateway to somewhere. However, we must let the routing 1007 * protocol deal with that. 1008 * 1009 * The srcrt parameter indicates whether the packet is being forwarded 1010 * via a source route. 1011 */ 1012void 1013ip_forward(m, srcrt) 1014 struct mbuf *m; 1015 int srcrt; 1016{ 1017 register struct ip *ip = mtod(m, struct ip *); 1018 register struct sockaddr_in *sin; 1019 register struct rtentry *rt; 1020 int error, type = 0, code; 1021 struct mbuf *mcopy; 1022 n_long dest; 1023 struct ifnet *destifp; 1024 1025 dest = 0; 1026#ifdef DIAGNOSTIC 1027 if (ipprintfs) 1028 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1029 ip->ip_dst, ip->ip_ttl); 1030#endif 1031 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1032 ipstat.ips_cantforward++; 1033 m_freem(m); 1034 return; 1035 } 1036 HTONS(ip->ip_id); 1037 if (ip->ip_ttl <= IPTTLDEC) { 1038 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1039 return; 1040 } 1041 ip->ip_ttl -= IPTTLDEC; 1042 1043 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1044 if ((rt = ipforward_rt.ro_rt) == 0 || 1045 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1046 if (ipforward_rt.ro_rt) { 1047 RTFREE(ipforward_rt.ro_rt); 1048 ipforward_rt.ro_rt = 0; 1049 } 1050 sin->sin_family = AF_INET; 1051 sin->sin_len = sizeof(*sin); 1052 sin->sin_addr = ip->ip_dst; 1053 1054 rtalloc(&ipforward_rt); 1055 if (ipforward_rt.ro_rt == 0) { 1056 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1057 return; 1058 } 1059 rt = ipforward_rt.ro_rt; 1060 } 1061 1062 /* 1063 * Save at most 64 bytes of the packet in case 1064 * we need to generate an ICMP message to the src. 1065 */ 1066 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1067 1068#ifdef GATEWAY 1069 ip_ifmatrix[rt->rt_ifp->if_index + 1070 if_index * m->m_pkthdr.rcvif->if_index]++; 1071#endif 1072 /* 1073 * If forwarding packet using same interface that it came in on, 1074 * perhaps should send a redirect to sender to shortcut a hop. 1075 * Only send redirect if source is sending directly to us, 1076 * and if packet was not source routed (or has any options). 1077 * Also, don't send redirect if forwarding using a default route 1078 * or a route modified by a redirect. 1079 */ 1080#define satosin(sa) ((struct sockaddr_in *)(sa)) 1081 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1082 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1083 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1084 ipsendredirects && !srcrt) { 1085#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1086 u_int32_t src = ntohl(ip->ip_src.s_addr); 1087 1088 if (RTA(rt) && 1089 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1090 if (rt->rt_flags & RTF_GATEWAY) 1091 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1092 else 1093 dest = ip->ip_dst.s_addr; 1094 /* Router requirements says to only send host redirects */ 1095 type = ICMP_REDIRECT; 1096 code = ICMP_REDIRECT_HOST; 1097#ifdef DIAGNOSTIC 1098 if (ipprintfs) 1099 printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); 1100#endif 1101 } 1102 } 1103 1104 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1105#ifdef DIRECTED_BROADCAST 1106 | IP_ALLOWBROADCAST 1107#endif 1108 , 0); 1109 if (error) 1110 ipstat.ips_cantforward++; 1111 else { 1112 ipstat.ips_forward++; 1113 if (type) 1114 ipstat.ips_redirectsent++; 1115 else { 1116 if (mcopy) 1117 m_freem(mcopy); 1118 return; 1119 } 1120 } 1121 if (mcopy == NULL) 1122 return; 1123 destifp = NULL; 1124 1125 switch (error) { 1126 1127 case 0: /* forwarded, but need redirect */ 1128 /* type, code set above */ 1129 break; 1130 1131 case ENETUNREACH: /* shouldn't happen, checked above */ 1132 case EHOSTUNREACH: 1133 case ENETDOWN: 1134 case EHOSTDOWN: 1135 default: 1136 type = ICMP_UNREACH; 1137 code = ICMP_UNREACH_HOST; 1138 break; 1139 1140 case EMSGSIZE: 1141 type = ICMP_UNREACH; 1142 code = ICMP_UNREACH_NEEDFRAG; 1143 if (ipforward_rt.ro_rt) 1144 destifp = ipforward_rt.ro_rt->rt_ifp; 1145 ipstat.ips_cantfrag++; 1146 break; 1147 1148 case ENOBUFS: 1149 type = ICMP_SOURCEQUENCH; 1150 code = 0; 1151 break; 1152 } 1153 icmp_error(mcopy, type, code, dest, destifp); 1154} 1155 1156int 1157ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1158 int *name; 1159 u_int namelen; 1160 void *oldp; 1161 size_t *oldlenp; 1162 void *newp; 1163 size_t newlen; 1164{ 1165 /* All sysctl names at this level are terminal. */ 1166 if (namelen != 1) 1167 return (ENOTDIR); 1168 1169 switch (name[0]) { 1170 case IPCTL_FORWARDING: 1171 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1172 case IPCTL_SENDREDIRECTS: 1173 return (sysctl_int(oldp, oldlenp, newp, newlen, 1174 &ipsendredirects)); 1175 case IPCTL_DEFTTL: 1176 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1177#ifdef notyet 1178 case IPCTL_DEFMTU: 1179 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1180#endif 1181 default: 1182 return (EOPNOTSUPP); 1183 } 1184 /* NOTREACHED */ 1185} 1186