ip_input.c revision 1.19
1/* $NetBSD: ip_input.c,v 1.19 1995/06/04 05:07:03 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 (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.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 == 250 satosin(&ia->ia_broadaddr)->sin_addr.s_addr) 251 goto ours; 252 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 253 goto ours; 254 /* 255 * Look for all-0's host part (old broadcast addr), 256 * either for subnet or net. 257 */ 258 if (ip->ip_dst.s_addr == ia->ia_subnet || 259 ip->ip_dst.s_addr == ia->ia_net) 260 goto ours; 261 } 262 } 263 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 264 struct in_multi *inm; 265#ifdef MROUTING 266 extern struct socket *ip_mrouter; 267 268 if (m->m_flags & M_EXT) { 269 if ((m = m_pullup(m, hlen)) == 0) { 270 ipstat.ips_toosmall++; 271 goto next; 272 } 273 ip = mtod(m, struct ip *); 274 } 275 276 if (ip_mrouter) { 277 /* 278 * If we are acting as a multicast router, all 279 * incoming multicast packets are passed to the 280 * kernel-level multicast forwarding function. 281 * The packet is returned (relatively) intact; if 282 * ip_mforward() returns a non-zero value, the packet 283 * must be discarded, else it may be accepted below. 284 * 285 * (The IP ident field is put in the same byte order 286 * as expected when ip_mforward() is called from 287 * ip_output().) 288 */ 289 ip->ip_id = htons(ip->ip_id); 290 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 291 ipstat.ips_cantforward++; 292 m_freem(m); 293 goto next; 294 } 295 ip->ip_id = ntohs(ip->ip_id); 296 297 /* 298 * The process-level routing demon needs to receive 299 * all multicast IGMP packets, whether or not this 300 * host belongs to their destination groups. 301 */ 302 if (ip->ip_p == IPPROTO_IGMP) 303 goto ours; 304 ipstat.ips_forward++; 305 } 306#endif 307 /* 308 * See if we belong to the destination multicast group on the 309 * arrival interface. 310 */ 311 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 312 if (inm == NULL) { 313 ipstat.ips_cantforward++; 314 m_freem(m); 315 goto next; 316 } 317 goto ours; 318 } 319 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 320 ip->ip_dst.s_addr == INADDR_ANY) 321 goto ours; 322 323 /* 324 * Not for us; forward if possible and desirable. 325 */ 326 if (ipforwarding == 0) { 327 ipstat.ips_cantforward++; 328 m_freem(m); 329 } else 330 ip_forward(m, 0); 331 goto next; 332 333ours: 334 /* 335 * If offset or IP_MF are set, must reassemble. 336 * Otherwise, nothing need be done. 337 * (We could look in the reassembly queue to see 338 * if the packet was previously fragmented, 339 * but it's not worth the time; just let them time out.) 340 */ 341 if (ip->ip_off &~ IP_DF) { 342 if (m->m_flags & M_EXT) { /* XXX */ 343 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 344 ipstat.ips_toosmall++; 345 goto next; 346 } 347 ip = mtod(m, struct ip *); 348 } 349 /* 350 * Look for queue of fragments 351 * of this datagram. 352 */ 353 for (fp = ipq.next; fp != &ipq; fp = fp->next) 354 if (ip->ip_id == fp->ipq_id && 355 ip->ip_src.s_addr == fp->ipq_src.s_addr && 356 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 357 ip->ip_p == fp->ipq_p) 358 goto found; 359 fp = 0; 360found: 361 362 /* 363 * Adjust ip_len to not reflect header, 364 * set ip_mff if more fragments are expected, 365 * convert offset of this to bytes. 366 */ 367 ip->ip_len -= hlen; 368 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 369 if (ip->ip_off & IP_MF) { 370 /* 371 * Make sure that fragments have a data length 372 * that's a non-zero multiple of 8 bytes. 373 */ 374 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 375 ipstat.ips_badfrags++; 376 goto bad; 377 } 378 ((struct ipasfrag *)ip)->ipf_mff |= 1; 379 } 380 ip->ip_off <<= 3; 381 382 /* 383 * If datagram marked as having more fragments 384 * or if this is not the first fragment, 385 * attempt reassembly; if it succeeds, proceed. 386 */ 387 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 388 ipstat.ips_fragments++; 389 ip = ip_reass((struct ipasfrag *)ip, fp); 390 if (ip == 0) 391 goto next; 392 ipstat.ips_reassembled++; 393 m = dtom(ip); 394 } else 395 if (fp) 396 ip_freef(fp); 397 } else 398 ip->ip_len -= hlen; 399 400 /* 401 * Switch out to protocol's input routine. 402 */ 403 ipstat.ips_delivered++; 404 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 405 goto next; 406bad: 407 m_freem(m); 408 goto next; 409} 410 411/* 412 * Take incoming datagram fragment and try to 413 * reassemble it into whole datagram. If a chain for 414 * reassembly of this datagram already exists, then it 415 * is given as fp; otherwise have to make a chain. 416 */ 417struct ip * 418ip_reass(ip, fp) 419 register struct ipasfrag *ip; 420 register struct ipq *fp; 421{ 422 register struct mbuf *m = dtom(ip); 423 register struct ipasfrag *q; 424 struct mbuf *t; 425 int hlen = ip->ip_hl << 2; 426 int i, next; 427 428 /* 429 * Presence of header sizes in mbufs 430 * would confuse code below. 431 */ 432 m->m_data += hlen; 433 m->m_len -= hlen; 434 435 /* 436 * If first fragment to arrive, create a reassembly queue. 437 */ 438 if (fp == 0) { 439 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 440 goto dropfrag; 441 fp = mtod(t, struct ipq *); 442 insque(fp, &ipq); 443 fp->ipq_ttl = IPFRAGTTL; 444 fp->ipq_p = ip->ip_p; 445 fp->ipq_id = ip->ip_id; 446 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 447 fp->ipq_src = ((struct ip *)ip)->ip_src; 448 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 449 q = (struct ipasfrag *)fp; 450 goto insert; 451 } 452 453 /* 454 * Find a segment which begins after this one does. 455 */ 456 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 457 if (q->ip_off > ip->ip_off) 458 break; 459 460 /* 461 * If there is a preceding segment, it may provide some of 462 * our data already. If so, drop the data from the incoming 463 * segment. If it provides all of our data, drop us. 464 */ 465 if (q->ipf_prev != (struct ipasfrag *)fp) { 466 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 467 if (i > 0) { 468 if (i >= ip->ip_len) 469 goto dropfrag; 470 m_adj(dtom(ip), i); 471 ip->ip_off += i; 472 ip->ip_len -= i; 473 } 474 } 475 476 /* 477 * While we overlap succeeding segments trim them or, 478 * if they are completely covered, dequeue them. 479 */ 480 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 481 i = (ip->ip_off + ip->ip_len) - q->ip_off; 482 if (i < q->ip_len) { 483 q->ip_len -= i; 484 q->ip_off += i; 485 m_adj(dtom(q), i); 486 break; 487 } 488 q = q->ipf_next; 489 m_freem(dtom(q->ipf_prev)); 490 ip_deq(q->ipf_prev); 491 } 492 493insert: 494 /* 495 * Stick new segment in its place; 496 * check for complete reassembly. 497 */ 498 ip_enq(ip, q->ipf_prev); 499 next = 0; 500 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 501 if (q->ip_off != next) 502 return (0); 503 next += q->ip_len; 504 } 505 if (q->ipf_prev->ipf_mff & 1) 506 return (0); 507 508 /* 509 * Reassembly is complete; concatenate fragments. 510 */ 511 q = fp->ipq_next; 512 m = dtom(q); 513 t = m->m_next; 514 m->m_next = 0; 515 m_cat(m, t); 516 q = q->ipf_next; 517 while (q != (struct ipasfrag *)fp) { 518 t = dtom(q); 519 q = q->ipf_next; 520 m_cat(m, t); 521 } 522 523 /* 524 * Create header for new ip packet by 525 * modifying header of first packet; 526 * dequeue and discard fragment reassembly header. 527 * Make header visible. 528 */ 529 ip = fp->ipq_next; 530 ip->ip_len = next; 531 ip->ipf_mff &= ~1; 532 ((struct ip *)ip)->ip_src = fp->ipq_src; 533 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 534 remque(fp); 535 (void) m_free(dtom(fp)); 536 m = dtom(ip); 537 m->m_len += (ip->ip_hl << 2); 538 m->m_data -= (ip->ip_hl << 2); 539 /* some debugging cruft by sklower, below, will go away soon */ 540 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 541 register int plen = 0; 542 for (t = m; m; m = m->m_next) 543 plen += m->m_len; 544 t->m_pkthdr.len = plen; 545 } 546 return ((struct ip *)ip); 547 548dropfrag: 549 ipstat.ips_fragdropped++; 550 m_freem(m); 551 return (0); 552} 553 554/* 555 * Free a fragment reassembly header and all 556 * associated datagrams. 557 */ 558void 559ip_freef(fp) 560 struct ipq *fp; 561{ 562 register struct ipasfrag *q, *p; 563 564 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 565 p = q->ipf_next; 566 ip_deq(q); 567 m_freem(dtom(q)); 568 } 569 remque(fp); 570 (void) m_free(dtom(fp)); 571} 572 573/* 574 * Put an ip fragment on a reassembly chain. 575 * Like insque, but pointers in middle of structure. 576 */ 577void 578ip_enq(p, prev) 579 register struct ipasfrag *p, *prev; 580{ 581 582 p->ipf_prev = prev; 583 p->ipf_next = prev->ipf_next; 584 prev->ipf_next->ipf_prev = p; 585 prev->ipf_next = p; 586} 587 588/* 589 * To ip_enq as remque is to insque. 590 */ 591void 592ip_deq(p) 593 register struct ipasfrag *p; 594{ 595 596 p->ipf_prev->ipf_next = p->ipf_next; 597 p->ipf_next->ipf_prev = p->ipf_prev; 598} 599 600/* 601 * IP timer processing; 602 * if a timer expires on a reassembly 603 * queue, discard it. 604 */ 605void 606ip_slowtimo() 607{ 608 register struct ipq *fp; 609 int s = splnet(); 610 611 fp = ipq.next; 612 if (fp == 0) { 613 splx(s); 614 return; 615 } 616 while (fp != &ipq) { 617 --fp->ipq_ttl; 618 fp = fp->next; 619 if (fp->prev->ipq_ttl == 0) { 620 ipstat.ips_fragtimeout++; 621 ip_freef(fp->prev); 622 } 623 } 624 splx(s); 625} 626 627/* 628 * Drain off all datagram fragments. 629 */ 630void 631ip_drain() 632{ 633 634 while (ipq.next != &ipq) { 635 ipstat.ips_fragdropped++; 636 ip_freef(ipq.next); 637 } 638} 639 640/* 641 * Do option processing on a datagram, 642 * possibly discarding it if bad options are encountered, 643 * or forwarding it if source-routed. 644 * Returns 1 if packet has been forwarded/freed, 645 * 0 if the packet should be processed further. 646 */ 647int 648ip_dooptions(m) 649 struct mbuf *m; 650{ 651 register struct ip *ip = mtod(m, struct ip *); 652 register u_char *cp; 653 register struct ip_timestamp *ipt; 654 register struct in_ifaddr *ia; 655 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 656 struct in_addr *sin, dst; 657 n_time ntime; 658 659 dst = ip->ip_dst; 660 cp = (u_char *)(ip + 1); 661 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 662 for (; cnt > 0; cnt -= optlen, cp += optlen) { 663 opt = cp[IPOPT_OPTVAL]; 664 if (opt == IPOPT_EOL) 665 break; 666 if (opt == IPOPT_NOP) 667 optlen = 1; 668 else { 669 optlen = cp[IPOPT_OLEN]; 670 if (optlen <= 0 || optlen > cnt) { 671 code = &cp[IPOPT_OLEN] - (u_char *)ip; 672 goto bad; 673 } 674 } 675 switch (opt) { 676 677 default: 678 break; 679 680 /* 681 * Source routing with record. 682 * Find interface with current destination address. 683 * If none on this machine then drop if strictly routed, 684 * or do nothing if loosely routed. 685 * Record interface address and bring up next address 686 * component. If strictly routed make sure next 687 * address is on directly accessible net. 688 */ 689 case IPOPT_LSRR: 690 case IPOPT_SSRR: 691 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 692 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 693 goto bad; 694 } 695 ipaddr.sin_addr = ip->ip_dst; 696 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 697 if (ia == 0) { 698 if (opt == IPOPT_SSRR) { 699 type = ICMP_UNREACH; 700 code = ICMP_UNREACH_SRCFAIL; 701 goto bad; 702 } 703 /* 704 * Loose routing, and not at next destination 705 * yet; nothing to do except forward. 706 */ 707 break; 708 } 709 off--; /* 0 origin */ 710 if (off > optlen - sizeof(struct in_addr)) { 711 /* 712 * End of source route. Should be for us. 713 */ 714 save_rte(cp, ip->ip_src); 715 break; 716 } 717 /* 718 * locate outgoing interface 719 */ 720 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 721 sizeof(ipaddr.sin_addr)); 722 if (opt == IPOPT_SSRR) { 723#define INA struct in_ifaddr * 724#define SA struct sockaddr * 725 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 726 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 727 } else 728 ia = ip_rtaddr(ipaddr.sin_addr); 729 if (ia == 0) { 730 type = ICMP_UNREACH; 731 code = ICMP_UNREACH_SRCFAIL; 732 goto bad; 733 } 734 ip->ip_dst = ipaddr.sin_addr; 735 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 736 (caddr_t)(cp + off), sizeof(struct in_addr)); 737 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 738 /* 739 * Let ip_intr's mcast routing check handle mcast pkts 740 */ 741 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 742 break; 743 744 case IPOPT_RR: 745 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 746 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 747 goto bad; 748 } 749 /* 750 * If no space remains, ignore. 751 */ 752 off--; /* 0 origin */ 753 if (off > optlen - sizeof(struct in_addr)) 754 break; 755 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 756 sizeof(ipaddr.sin_addr)); 757 /* 758 * locate outgoing interface; if we're the destination, 759 * use the incoming interface (should be same). 760 */ 761 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 762 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 763 type = ICMP_UNREACH; 764 code = ICMP_UNREACH_HOST; 765 goto bad; 766 } 767 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 768 (caddr_t)(cp + off), sizeof(struct in_addr)); 769 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 770 break; 771 772 case IPOPT_TS: 773 code = cp - (u_char *)ip; 774 ipt = (struct ip_timestamp *)cp; 775 if (ipt->ipt_len < 5) 776 goto bad; 777 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 778 if (++ipt->ipt_oflw == 0) 779 goto bad; 780 break; 781 } 782 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 783 switch (ipt->ipt_flg) { 784 785 case IPOPT_TS_TSONLY: 786 break; 787 788 case IPOPT_TS_TSANDADDR: 789 if (ipt->ipt_ptr + sizeof(n_time) + 790 sizeof(struct in_addr) > ipt->ipt_len) 791 goto bad; 792 ipaddr.sin_addr = dst; 793 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 794 m->m_pkthdr.rcvif); 795 if (ia == 0) 796 continue; 797 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 798 (caddr_t)sin, sizeof(struct in_addr)); 799 ipt->ipt_ptr += sizeof(struct in_addr); 800 break; 801 802 case IPOPT_TS_PRESPEC: 803 if (ipt->ipt_ptr + sizeof(n_time) + 804 sizeof(struct in_addr) > ipt->ipt_len) 805 goto bad; 806 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 807 sizeof(struct in_addr)); 808 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 809 continue; 810 ipt->ipt_ptr += sizeof(struct in_addr); 811 break; 812 813 default: 814 goto bad; 815 } 816 ntime = iptime(); 817 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 818 sizeof(n_time)); 819 ipt->ipt_ptr += sizeof(n_time); 820 } 821 } 822 if (forward) { 823 ip_forward(m, 1); 824 return (1); 825 } 826 return (0); 827bad: 828 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 829 icmp_error(m, type, code, 0, 0); 830 ipstat.ips_badoptions++; 831 return (1); 832} 833 834/* 835 * Given address of next destination (final or next hop), 836 * return internet address info of interface to be used to get there. 837 */ 838struct in_ifaddr * 839ip_rtaddr(dst) 840 struct in_addr dst; 841{ 842 register struct sockaddr_in *sin; 843 844 sin = satosin(&ipforward_rt.ro_dst); 845 846 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 847 if (ipforward_rt.ro_rt) { 848 RTFREE(ipforward_rt.ro_rt); 849 ipforward_rt.ro_rt = 0; 850 } 851 sin->sin_family = AF_INET; 852 sin->sin_len = sizeof(*sin); 853 sin->sin_addr = dst; 854 855 rtalloc(&ipforward_rt); 856 } 857 if (ipforward_rt.ro_rt == 0) 858 return ((struct in_ifaddr *)0); 859 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 860} 861 862/* 863 * Save incoming source route for use in replies, 864 * to be picked up later by ip_srcroute if the receiver is interested. 865 */ 866void 867save_rte(option, dst) 868 u_char *option; 869 struct in_addr dst; 870{ 871 unsigned olen; 872 873 olen = option[IPOPT_OLEN]; 874#ifdef DIAGNOSTIC 875 if (ipprintfs) 876 printf("save_rte: olen %d\n", olen); 877#endif 878 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 879 return; 880 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 881 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 882 ip_srcrt.dst = dst; 883} 884 885/* 886 * Retrieve incoming source route for use in replies, 887 * in the same form used by setsockopt. 888 * The first hop is placed before the options, will be removed later. 889 */ 890struct mbuf * 891ip_srcroute() 892{ 893 register struct in_addr *p, *q; 894 register struct mbuf *m; 895 896 if (ip_nhops == 0) 897 return ((struct mbuf *)0); 898 m = m_get(M_DONTWAIT, MT_SOOPTS); 899 if (m == 0) 900 return ((struct mbuf *)0); 901 902#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 903 904 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 905 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 906 OPTSIZ; 907#ifdef DIAGNOSTIC 908 if (ipprintfs) 909 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 910#endif 911 912 /* 913 * First save first hop for return route 914 */ 915 p = &ip_srcrt.route[ip_nhops - 1]; 916 *(mtod(m, struct in_addr *)) = *p--; 917#ifdef DIAGNOSTIC 918 if (ipprintfs) 919 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 920#endif 921 922 /* 923 * Copy option fields and padding (nop) to mbuf. 924 */ 925 ip_srcrt.nop = IPOPT_NOP; 926 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 927 bcopy((caddr_t)&ip_srcrt.nop, 928 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 929 q = (struct in_addr *)(mtod(m, caddr_t) + 930 sizeof(struct in_addr) + OPTSIZ); 931#undef OPTSIZ 932 /* 933 * Record return path as an IP source route, 934 * reversing the path (pointers are now aligned). 935 */ 936 while (p >= ip_srcrt.route) { 937#ifdef DIAGNOSTIC 938 if (ipprintfs) 939 printf(" %lx", ntohl(q->s_addr)); 940#endif 941 *q++ = *p--; 942 } 943 /* 944 * Last hop goes to final destination. 945 */ 946 *q = ip_srcrt.dst; 947#ifdef DIAGNOSTIC 948 if (ipprintfs) 949 printf(" %lx\n", ntohl(q->s_addr)); 950#endif 951 return (m); 952} 953 954/* 955 * Strip out IP options, at higher 956 * level protocol in the kernel. 957 * Second argument is buffer to which options 958 * will be moved, and return value is their length. 959 * XXX should be deleted; last arg currently ignored. 960 */ 961void 962ip_stripoptions(m, mopt) 963 register struct mbuf *m; 964 struct mbuf *mopt; 965{ 966 register int i; 967 struct ip *ip = mtod(m, struct ip *); 968 register caddr_t opts; 969 int olen; 970 971 olen = (ip->ip_hl<<2) - sizeof (struct ip); 972 opts = (caddr_t)(ip + 1); 973 i = m->m_len - (sizeof (struct ip) + olen); 974 bcopy(opts + olen, opts, (unsigned)i); 975 m->m_len -= olen; 976 if (m->m_flags & M_PKTHDR) 977 m->m_pkthdr.len -= olen; 978 ip->ip_hl = sizeof(struct ip) >> 2; 979} 980 981u_char inetctlerrmap[PRC_NCMDS] = { 982 0, 0, 0, 0, 983 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 984 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 985 EMSGSIZE, EHOSTUNREACH, 0, 0, 986 0, 0, 0, 0, 987 ENOPROTOOPT 988}; 989 990/* 991 * Forward a packet. If some error occurs return the sender 992 * an icmp packet. Note we can't always generate a meaningful 993 * icmp message because icmp doesn't have a large enough repertoire 994 * of codes and types. 995 * 996 * If not forwarding, just drop the packet. This could be confusing 997 * if ipforwarding was zero but some routing protocol was advancing 998 * us as a gateway to somewhere. However, we must let the routing 999 * protocol deal with that. 1000 * 1001 * The srcrt parameter indicates whether the packet is being forwarded 1002 * via a source route. 1003 */ 1004void 1005ip_forward(m, srcrt) 1006 struct mbuf *m; 1007 int srcrt; 1008{ 1009 register struct ip *ip = mtod(m, struct ip *); 1010 register struct sockaddr_in *sin; 1011 register struct rtentry *rt; 1012 int error, type = 0, code; 1013 struct mbuf *mcopy; 1014 n_long dest; 1015 struct ifnet *destifp; 1016 1017 dest = 0; 1018#ifdef DIAGNOSTIC 1019 if (ipprintfs) 1020 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1021 ip->ip_dst, ip->ip_ttl); 1022#endif 1023 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1024 ipstat.ips_cantforward++; 1025 m_freem(m); 1026 return; 1027 } 1028 HTONS(ip->ip_id); 1029 if (ip->ip_ttl <= IPTTLDEC) { 1030 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1031 return; 1032 } 1033 ip->ip_ttl -= IPTTLDEC; 1034 1035 sin = satosin(&ipforward_rt.ro_dst); 1036 if ((rt = ipforward_rt.ro_rt) == 0 || 1037 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1038 if (ipforward_rt.ro_rt) { 1039 RTFREE(ipforward_rt.ro_rt); 1040 ipforward_rt.ro_rt = 0; 1041 } 1042 sin->sin_family = AF_INET; 1043 sin->sin_len = sizeof(*sin); 1044 sin->sin_addr = ip->ip_dst; 1045 1046 rtalloc(&ipforward_rt); 1047 if (ipforward_rt.ro_rt == 0) { 1048 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1049 return; 1050 } 1051 rt = ipforward_rt.ro_rt; 1052 } 1053 1054 /* 1055 * Save at most 64 bytes of the packet in case 1056 * we need to generate an ICMP message to the src. 1057 */ 1058 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1059 1060#ifdef GATEWAY 1061 ip_ifmatrix[rt->rt_ifp->if_index + 1062 if_index * m->m_pkthdr.rcvif->if_index]++; 1063#endif 1064 /* 1065 * If forwarding packet using same interface that it came in on, 1066 * perhaps should send a redirect to sender to shortcut a hop. 1067 * Only send redirect if source is sending directly to us, 1068 * and if packet was not source routed (or has any options). 1069 * Also, don't send redirect if forwarding using a default route 1070 * or a route modified by a redirect. 1071 */ 1072 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1073 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1074 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1075 ipsendredirects && !srcrt) { 1076 if (rt->rt_ifa && 1077 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1078 ifatoia(rt->rt_ifa)->ia_subnet) { 1079 if (rt->rt_flags & RTF_GATEWAY) 1080 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1081 else 1082 dest = ip->ip_dst.s_addr; 1083 /* Router requirements says to only send host redirects */ 1084 type = ICMP_REDIRECT; 1085 code = ICMP_REDIRECT_HOST; 1086#ifdef DIAGNOSTIC 1087 if (ipprintfs) 1088 printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); 1089#endif 1090 } 1091 } 1092 1093 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1094#ifdef DIRECTED_BROADCAST 1095 | IP_ALLOWBROADCAST 1096#endif 1097 , 0); 1098 if (error) 1099 ipstat.ips_cantforward++; 1100 else { 1101 ipstat.ips_forward++; 1102 if (type) 1103 ipstat.ips_redirectsent++; 1104 else { 1105 if (mcopy) 1106 m_freem(mcopy); 1107 return; 1108 } 1109 } 1110 if (mcopy == NULL) 1111 return; 1112 destifp = NULL; 1113 1114 switch (error) { 1115 1116 case 0: /* forwarded, but need redirect */ 1117 /* type, code set above */ 1118 break; 1119 1120 case ENETUNREACH: /* shouldn't happen, checked above */ 1121 case EHOSTUNREACH: 1122 case ENETDOWN: 1123 case EHOSTDOWN: 1124 default: 1125 type = ICMP_UNREACH; 1126 code = ICMP_UNREACH_HOST; 1127 break; 1128 1129 case EMSGSIZE: 1130 type = ICMP_UNREACH; 1131 code = ICMP_UNREACH_NEEDFRAG; 1132 if (ipforward_rt.ro_rt) 1133 destifp = ipforward_rt.ro_rt->rt_ifp; 1134 ipstat.ips_cantfrag++; 1135 break; 1136 1137 case ENOBUFS: 1138 type = ICMP_SOURCEQUENCH; 1139 code = 0; 1140 break; 1141 } 1142 icmp_error(mcopy, type, code, dest, destifp); 1143} 1144 1145int 1146ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1147 int *name; 1148 u_int namelen; 1149 void *oldp; 1150 size_t *oldlenp; 1151 void *newp; 1152 size_t newlen; 1153{ 1154 /* All sysctl names at this level are terminal. */ 1155 if (namelen != 1) 1156 return (ENOTDIR); 1157 1158 switch (name[0]) { 1159 case IPCTL_FORWARDING: 1160 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1161 case IPCTL_SENDREDIRECTS: 1162 return (sysctl_int(oldp, oldlenp, newp, newlen, 1163 &ipsendredirects)); 1164 case IPCTL_DEFTTL: 1165 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1166#ifdef notyet 1167 case IPCTL_DEFMTU: 1168 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1169#endif 1170 default: 1171 return (EOPNOTSUPP); 1172 } 1173 /* NOTREACHED */ 1174} 1175