ip_input.c revision 1.13
1/* 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. 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 8.2 (Berkeley) 1/4/94 34 * $Id: ip_input.c,v 1.13 1994/05/13 06:06:21 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 59#ifndef IPFORWARDING 60#ifdef GATEWAY 61#define IPFORWARDING 1 /* forward IP packets not for us */ 62#else /* GATEWAY */ 63#define IPFORWARDING 0 /* don't forward IP packets not for us */ 64#endif /* GATEWAY */ 65#endif /* IPFORWARDING */ 66#ifndef IPSENDREDIRECTS 67#define IPSENDREDIRECTS 1 68#endif 69int ipforwarding = IPFORWARDING; 70int ipsendredirects = IPSENDREDIRECTS; 71int ip_defttl = IPDEFTTL; 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 */ 81struct ifqueue ipintrq; 82 83/* 84 * We need to save the IP options in case a protocol wants to respond 85 * to an incoming packet over the same route if the packet got here 86 * using IP source routing. This allows connection establishment and 87 * maintenance when the remote end is on a network that is not known 88 * to us. 89 */ 90int ip_nhops = 0; 91static struct ip_srcrt { 92 struct in_addr dst; /* final destination */ 93 char nop; /* one NOP to align */ 94 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 95 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 96} ip_srcrt; 97 98#ifdef GATEWAY 99extern int if_index; 100u_long *ip_ifmatrix; 101#endif 102 103static void save_rte __P((u_char *, struct in_addr)); 104/* 105 * IP initialization: fill in IP protocol switch table. 106 * All protocols not implemented in kernel go to raw IP protocol handler. 107 */ 108void 109ip_init() 110{ 111 register struct protosw *pr; 112 register int i; 113 114 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 115 if (pr == 0) 116 panic("ip_init"); 117 for (i = 0; i < IPPROTO_MAX; i++) 118 ip_protox[i] = pr - inetsw; 119 for (pr = inetdomain.dom_protosw; 120 pr < inetdomain.dom_protoswNPROTOSW; pr++) 121 if (pr->pr_domain->dom_family == PF_INET && 122 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 123 ip_protox[pr->pr_protocol] = pr - inetsw; 124 ipq.next = ipq.prev = &ipq; 125 ip_id = time.tv_sec & 0xffff; 126 ipintrq.ifq_maxlen = ipqmaxlen; 127#ifdef GATEWAY 128 i = (if_index + 1) * (if_index + 1) * sizeof (u_long); 129 ip_ifmatrix = (u_long *) malloc(i, M_RTABLE, M_WAITOK); 130 bzero((char *)ip_ifmatrix, i); 131#endif 132} 133 134struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 135struct route ipforward_rt; 136 137/* 138 * Ip input routine. Checksum and byte swap header. If fragmented 139 * try to reassemble. Process options. Pass to next level. 140 */ 141void 142ipintr() 143{ 144 register struct ip *ip; 145 register struct mbuf *m; 146 register struct ipq *fp; 147 register struct in_ifaddr *ia; 148 int hlen, s; 149 150next: 151 /* 152 * Get next datagram off input queue and get IP header 153 * in first mbuf. 154 */ 155 s = splimp(); 156 IF_DEQUEUE(&ipintrq, m); 157 splx(s); 158 if (m == 0) 159 return; 160#ifdef DIAGNOSTIC 161 if ((m->m_flags & M_PKTHDR) == 0) 162 panic("ipintr no HDR"); 163#endif 164 /* 165 * If no IP addresses have been set yet but the interfaces 166 * are receiving, can't do anything with incoming packets yet. 167 */ 168 if (in_ifaddr == NULL) 169 goto bad; 170 ipstat.ips_total++; 171 if (m->m_len < sizeof (struct ip) && 172 (m = m_pullup(m, sizeof (struct ip))) == 0) { 173 ipstat.ips_toosmall++; 174 goto next; 175 } 176 ip = mtod(m, struct ip *); 177 if (ip->ip_v != IPVERSION) { 178 ipstat.ips_badvers++; 179 goto bad; 180 } 181 hlen = ip->ip_hl << 2; 182 if (hlen < sizeof(struct ip)) { /* minimum header length */ 183 ipstat.ips_badhlen++; 184 goto bad; 185 } 186 if (hlen > m->m_len) { 187 if ((m = m_pullup(m, hlen)) == 0) { 188 ipstat.ips_badhlen++; 189 goto next; 190 } 191 ip = mtod(m, struct ip *); 192 } 193 if (ip->ip_sum = in_cksum(m, hlen)) { 194 ipstat.ips_badsum++; 195 goto bad; 196 } 197 198 /* 199 * Convert fields to host representation. 200 */ 201 NTOHS(ip->ip_len); 202 if (ip->ip_len < hlen) { 203 ipstat.ips_badlen++; 204 goto bad; 205 } 206 NTOHS(ip->ip_id); 207 NTOHS(ip->ip_off); 208 209 /* 210 * Check that the amount of data in the buffers 211 * is as at least much as the IP header would have us expect. 212 * Trim mbufs if longer than we expect. 213 * Drop packet if shorter than we expect. 214 */ 215 if (m->m_pkthdr.len < ip->ip_len) { 216 ipstat.ips_tooshort++; 217 goto bad; 218 } 219 if (m->m_pkthdr.len > ip->ip_len) { 220 if (m->m_len == m->m_pkthdr.len) { 221 m->m_len = ip->ip_len; 222 m->m_pkthdr.len = ip->ip_len; 223 } else 224 m_adj(m, ip->ip_len - m->m_pkthdr.len); 225 } 226 227 /* 228 * Process options and, if not destined for us, 229 * ship it on. ip_dooptions returns 1 when an 230 * error was detected (causing an icmp message 231 * to be sent and the original packet to be freed). 232 */ 233 ip_nhops = 0; /* for source routed packets */ 234 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 235 goto next; 236 237 /* 238 * Check our list of addresses, to see if the packet is for us. 239 */ 240 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 241#define satosin(sa) ((struct sockaddr_in *)(sa)) 242 243 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 244 goto ours; 245 if ( 246#ifdef DIRECTED_BROADCAST 247 ia->ia_ifp == m->m_pkthdr.rcvif && 248#endif 249 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 250 u_long t; 251 252 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 253 ip->ip_dst.s_addr) 254 goto ours; 255 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 256 goto ours; 257 /* 258 * Look for all-0's host part (old broadcast addr), 259 * either for subnet or net. 260 */ 261 t = ntohl(ip->ip_dst.s_addr); 262 if (t == ia->ia_subnet) 263 goto ours; 264 if (t == ia->ia_net) 265 goto ours; 266 } 267 } 268 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 269 struct in_multi *inm; 270#ifdef MROUTING 271 extern struct socket *ip_mrouter; 272 273 if (m->m_flags & M_EXT) { 274 if ((m = m_pullup(m, hlen)) == 0) { 275 ipstat.ips_toosmall++; 276 goto next; 277 } 278 ip = mtod(m, struct ip *); 279 } 280 281 if (ip_mrouter) { 282 /* 283 * If we are acting as a multicast router, all 284 * incoming multicast packets are passed to the 285 * kernel-level multicast forwarding function. 286 * The packet is returned (relatively) intact; if 287 * ip_mforward() returns a non-zero value, the packet 288 * must be discarded, else it may be accepted below. 289 * 290 * (The IP ident field is put in the same byte order 291 * as expected when ip_mforward() is called from 292 * ip_output().) 293 */ 294 ip->ip_id = htons(ip->ip_id); 295 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 296 ipstat.ips_cantforward++; 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 ipstat.ips_forward++; 310 } 311#endif 312 /* 313 * See if we belong to the destination multicast group on the 314 * arrival interface. 315 */ 316 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 317 if (inm == NULL) { 318 ipstat.ips_cantforward++; 319 m_freem(m); 320 goto next; 321 } 322 goto ours; 323 } 324 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 325 goto ours; 326 if (ip->ip_dst.s_addr == INADDR_ANY) 327 goto ours; 328 329 /* 330 * Not for us; forward if possible and desirable. 331 */ 332 if (ipforwarding == 0) { 333 ipstat.ips_cantforward++; 334 m_freem(m); 335 } else 336 ip_forward(m, 0); 337 goto next; 338 339ours: 340 /* 341 * If offset or IP_MF are set, must reassemble. 342 * Otherwise, nothing need be done. 343 * (We could look in the reassembly queue to see 344 * if the packet was previously fragmented, 345 * but it's not worth the time; just let them time out.) 346 */ 347 if (ip->ip_off &~ IP_DF) { 348 if (m->m_flags & M_EXT) { /* XXX */ 349 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 350 ipstat.ips_toosmall++; 351 goto next; 352 } 353 ip = mtod(m, struct ip *); 354 } 355 /* 356 * Look for queue of fragments 357 * of this datagram. 358 */ 359 for (fp = ipq.next; fp != &ipq; fp = fp->next) 360 if (ip->ip_id == fp->ipq_id && 361 ip->ip_src.s_addr == fp->ipq_src.s_addr && 362 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 363 ip->ip_p == fp->ipq_p) 364 goto found; 365 fp = 0; 366found: 367 368 /* 369 * Adjust ip_len to not reflect header, 370 * set ip_mff if more fragments are expected, 371 * convert offset of this to bytes. 372 */ 373 ip->ip_len -= hlen; 374 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 375 if (ip->ip_off & IP_MF) 376 ((struct ipasfrag *)ip)->ipf_mff |= 1; 377 ip->ip_off <<= 3; 378 379 /* 380 * If datagram marked as having more fragments 381 * or if this is not the first fragment, 382 * attempt reassembly; if it succeeds, proceed. 383 */ 384 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 385 ipstat.ips_fragments++; 386 ip = ip_reass((struct ipasfrag *)ip, fp); 387 if (ip == 0) 388 goto next; 389 ipstat.ips_reassembled++; 390 m = dtom(ip); 391 } else 392 if (fp) 393 ip_freef(fp); 394 } else 395 ip->ip_len -= hlen; 396 397 /* 398 * Switch out to protocol's input routine. 399 */ 400 ipstat.ips_delivered++; 401 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 402 goto next; 403bad: 404 m_freem(m); 405 goto next; 406} 407 408/* 409 * Take incoming datagram fragment and try to 410 * reassemble it into whole datagram. If a chain for 411 * reassembly of this datagram already exists, then it 412 * is given as fp; otherwise have to make a chain. 413 */ 414struct ip * 415ip_reass(ip, fp) 416 register struct ipasfrag *ip; 417 register struct ipq *fp; 418{ 419 register struct mbuf *m = dtom(ip); 420 register struct ipasfrag *q; 421 struct mbuf *t; 422 int hlen = ip->ip_hl << 2; 423 int i, next; 424 425 /* 426 * Presence of header sizes in mbufs 427 * would confuse code below. 428 */ 429 m->m_data += hlen; 430 m->m_len -= hlen; 431 432 /* 433 * If first fragment to arrive, create a reassembly queue. 434 */ 435 if (fp == 0) { 436 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 437 goto dropfrag; 438 fp = mtod(t, struct ipq *); 439 insque(fp, &ipq); 440 fp->ipq_ttl = IPFRAGTTL; 441 fp->ipq_p = ip->ip_p; 442 fp->ipq_id = ip->ip_id; 443 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 444 fp->ipq_src = ((struct ip *)ip)->ip_src; 445 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 446 q = (struct ipasfrag *)fp; 447 goto insert; 448 } 449 450 /* 451 * Find a segment which begins after this one does. 452 */ 453 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 454 if (q->ip_off > ip->ip_off) 455 break; 456 457 /* 458 * If there is a preceding segment, it may provide some of 459 * our data already. If so, drop the data from the incoming 460 * segment. If it provides all of our data, drop us. 461 */ 462 if (q->ipf_prev != (struct ipasfrag *)fp) { 463 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 464 if (i > 0) { 465 if (i >= ip->ip_len) 466 goto dropfrag; 467 m_adj(dtom(ip), i); 468 ip->ip_off += i; 469 ip->ip_len -= i; 470 } 471 } 472 473 /* 474 * While we overlap succeeding segments trim them or, 475 * if they are completely covered, dequeue them. 476 */ 477 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 478 i = (ip->ip_off + ip->ip_len) - q->ip_off; 479 if (i < q->ip_len) { 480 q->ip_len -= i; 481 q->ip_off += i; 482 m_adj(dtom(q), i); 483 break; 484 } 485 q = q->ipf_next; 486 m_freem(dtom(q->ipf_prev)); 487 ip_deq(q->ipf_prev); 488 } 489 490insert: 491 /* 492 * Stick new segment in its place; 493 * check for complete reassembly. 494 */ 495 ip_enq(ip, q->ipf_prev); 496 next = 0; 497 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 498 if (q->ip_off != next) 499 return (0); 500 next += q->ip_len; 501 } 502 if (q->ipf_prev->ipf_mff & 1) 503 return (0); 504 505 /* 506 * Reassembly is complete; concatenate fragments. 507 */ 508 q = fp->ipq_next; 509 m = dtom(q); 510 t = m->m_next; 511 m->m_next = 0; 512 m_cat(m, t); 513 q = q->ipf_next; 514 while (q != (struct ipasfrag *)fp) { 515 t = dtom(q); 516 q = q->ipf_next; 517 m_cat(m, t); 518 } 519 520 /* 521 * Create header for new ip packet by 522 * modifying header of first packet; 523 * dequeue and discard fragment reassembly header. 524 * Make header visible. 525 */ 526 ip = fp->ipq_next; 527 ip->ip_len = next; 528 ip->ipf_mff &= ~1; 529 ((struct ip *)ip)->ip_src = fp->ipq_src; 530 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 531 remque(fp); 532 (void) m_free(dtom(fp)); 533 m = dtom(ip); 534 m->m_len += (ip->ip_hl << 2); 535 m->m_data -= (ip->ip_hl << 2); 536 /* some debugging cruft by sklower, below, will go away soon */ 537 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 538 register int plen = 0; 539 for (t = m; m; m = m->m_next) 540 plen += m->m_len; 541 t->m_pkthdr.len = plen; 542 } 543 return ((struct ip *)ip); 544 545dropfrag: 546 ipstat.ips_fragdropped++; 547 m_freem(m); 548 return (0); 549} 550 551/* 552 * Free a fragment reassembly header and all 553 * associated datagrams. 554 */ 555void 556ip_freef(fp) 557 struct ipq *fp; 558{ 559 register struct ipasfrag *q, *p; 560 561 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 562 p = q->ipf_next; 563 ip_deq(q); 564 m_freem(dtom(q)); 565 } 566 remque(fp); 567 (void) m_free(dtom(fp)); 568} 569 570/* 571 * Put an ip fragment on a reassembly chain. 572 * Like insque, but pointers in middle of structure. 573 */ 574void 575ip_enq(p, prev) 576 register struct ipasfrag *p, *prev; 577{ 578 579 p->ipf_prev = prev; 580 p->ipf_next = prev->ipf_next; 581 prev->ipf_next->ipf_prev = p; 582 prev->ipf_next = p; 583} 584 585/* 586 * To ip_enq as remque is to insque. 587 */ 588void 589ip_deq(p) 590 register struct ipasfrag *p; 591{ 592 593 p->ipf_prev->ipf_next = p->ipf_next; 594 p->ipf_next->ipf_prev = p->ipf_prev; 595} 596 597/* 598 * IP timer processing; 599 * if a timer expires on a reassembly 600 * queue, discard it. 601 */ 602void 603ip_slowtimo() 604{ 605 register struct ipq *fp; 606 int s = splnet(); 607 608 fp = ipq.next; 609 if (fp == 0) { 610 splx(s); 611 return; 612 } 613 while (fp != &ipq) { 614 --fp->ipq_ttl; 615 fp = fp->next; 616 if (fp->prev->ipq_ttl == 0) { 617 ipstat.ips_fragtimeout++; 618 ip_freef(fp->prev); 619 } 620 } 621 splx(s); 622} 623 624/* 625 * Drain off all datagram fragments. 626 */ 627void 628ip_drain() 629{ 630 631 while (ipq.next != &ipq) { 632 ipstat.ips_fragdropped++; 633 ip_freef(ipq.next); 634 } 635} 636 637/* 638 * Do option processing on a datagram, 639 * possibly discarding it if bad options are encountered, 640 * or forwarding it if source-routed. 641 * Returns 1 if packet has been forwarded/freed, 642 * 0 if the packet should be processed further. 643 */ 644int 645ip_dooptions(m) 646 struct mbuf *m; 647{ 648 register struct ip *ip = mtod(m, struct ip *); 649 register u_char *cp; 650 register struct ip_timestamp *ipt; 651 register struct in_ifaddr *ia; 652 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 653 struct in_addr *sin, dst; 654 n_time ntime; 655 656 dst = ip->ip_dst; 657 cp = (u_char *)(ip + 1); 658 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 659 for (; cnt > 0; cnt -= optlen, cp += optlen) { 660 opt = cp[IPOPT_OPTVAL]; 661 if (opt == IPOPT_EOL) 662 break; 663 if (opt == IPOPT_NOP) 664 optlen = 1; 665 else { 666 optlen = cp[IPOPT_OLEN]; 667 if (optlen <= 0 || optlen > cnt) { 668 code = &cp[IPOPT_OLEN] - (u_char *)ip; 669 goto bad; 670 } 671 } 672 switch (opt) { 673 674 default: 675 break; 676 677 /* 678 * Source routing with record. 679 * Find interface with current destination address. 680 * If none on this machine then drop if strictly routed, 681 * or do nothing if loosely routed. 682 * Record interface address and bring up next address 683 * component. If strictly routed make sure next 684 * address is on directly accessible net. 685 */ 686 case IPOPT_LSRR: 687 case IPOPT_SSRR: 688 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 689 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 690 goto bad; 691 } 692 ipaddr.sin_addr = ip->ip_dst; 693 ia = (struct in_ifaddr *) 694 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 695 if (ia == 0) { 696 if (opt == IPOPT_SSRR) { 697 type = ICMP_UNREACH; 698 code = ICMP_UNREACH_SRCFAIL; 699 goto bad; 700 } 701 /* 702 * Loose routing, and not at next destination 703 * yet; nothing to do except forward. 704 */ 705 break; 706 } 707 off--; /* 0 origin */ 708 if (off > optlen - sizeof(struct in_addr)) { 709 /* 710 * End of source route. Should be for us. 711 */ 712 save_rte(cp, ip->ip_src); 713 break; 714 } 715 /* 716 * locate outgoing interface 717 */ 718 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 719 sizeof(ipaddr.sin_addr)); 720 if (opt == IPOPT_SSRR) { 721#define INA struct in_ifaddr * 722#define SA struct sockaddr * 723 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 724 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 725 } else 726 ia = ip_rtaddr(ipaddr.sin_addr); 727 if (ia == 0) { 728 type = ICMP_UNREACH; 729 code = ICMP_UNREACH_SRCFAIL; 730 goto bad; 731 } 732 ip->ip_dst = ipaddr.sin_addr; 733 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 734 (caddr_t)(cp + off), sizeof(struct in_addr)); 735 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 736 /* 737 * Let ip_intr's mcast routing check handle mcast pkts 738 */ 739 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 740 break; 741 742 case IPOPT_RR: 743 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 744 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 745 goto bad; 746 } 747 /* 748 * If no space remains, ignore. 749 */ 750 off--; /* 0 origin */ 751 if (off > optlen - sizeof(struct in_addr)) 752 break; 753 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 754 sizeof(ipaddr.sin_addr)); 755 /* 756 * locate outgoing interface; if we're the destination, 757 * use the incoming interface (should be same). 758 */ 759 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 760 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 761 type = ICMP_UNREACH; 762 code = ICMP_UNREACH_HOST; 763 goto bad; 764 } 765 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 766 (caddr_t)(cp + off), sizeof(struct in_addr)); 767 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 768 break; 769 770 case IPOPT_TS: 771 code = cp - (u_char *)ip; 772 ipt = (struct ip_timestamp *)cp; 773 if (ipt->ipt_len < 5) 774 goto bad; 775 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 776 if (++ipt->ipt_oflw == 0) 777 goto bad; 778 break; 779 } 780 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 781 switch (ipt->ipt_flg) { 782 783 case IPOPT_TS_TSONLY: 784 break; 785 786 case IPOPT_TS_TSANDADDR: 787 if (ipt->ipt_ptr + sizeof(n_time) + 788 sizeof(struct in_addr) > ipt->ipt_len) 789 goto bad; 790 ipaddr.sin_addr = dst; 791 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 792 m->m_pkthdr.rcvif); 793 if (ia == 0) 794 continue; 795 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 796 (caddr_t)sin, sizeof(struct in_addr)); 797 ipt->ipt_ptr += sizeof(struct in_addr); 798 break; 799 800 case IPOPT_TS_PRESPEC: 801 if (ipt->ipt_ptr + sizeof(n_time) + 802 sizeof(struct in_addr) > ipt->ipt_len) 803 goto bad; 804 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 805 sizeof(struct in_addr)); 806 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 807 continue; 808 ipt->ipt_ptr += sizeof(struct in_addr); 809 break; 810 811 default: 812 goto bad; 813 } 814 ntime = iptime(); 815 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 816 sizeof(n_time)); 817 ipt->ipt_ptr += sizeof(n_time); 818 } 819 } 820 if (forward) { 821 ip_forward(m, 1); 822 return (1); 823 } 824 return (0); 825bad: 826 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 827 icmp_error(m, type, code, 0, 0); 828 ipstat.ips_badoptions++; 829 return (1); 830} 831 832/* 833 * Given address of next destination (final or next hop), 834 * return internet address info of interface to be used to get there. 835 */ 836struct in_ifaddr * 837ip_rtaddr(dst) 838 struct in_addr dst; 839{ 840 register struct sockaddr_in *sin; 841 842 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 843 844 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 845 if (ipforward_rt.ro_rt) { 846 RTFREE(ipforward_rt.ro_rt); 847 ipforward_rt.ro_rt = 0; 848 } 849 sin->sin_family = AF_INET; 850 sin->sin_len = sizeof(*sin); 851 sin->sin_addr = dst; 852 853 rtalloc(&ipforward_rt); 854 } 855 if (ipforward_rt.ro_rt == 0) 856 return ((struct in_ifaddr *)0); 857 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 858} 859 860/* 861 * Save incoming source route for use in replies, 862 * to be picked up later by ip_srcroute if the receiver is interested. 863 */ 864void 865save_rte(option, dst) 866 u_char *option; 867 struct in_addr dst; 868{ 869 unsigned olen; 870 871 olen = option[IPOPT_OLEN]; 872#ifdef DIAGNOSTIC 873 if (ipprintfs) 874 printf("save_rte: olen %d\n", olen); 875#endif 876 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 877 return; 878 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 879 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 880 ip_srcrt.dst = dst; 881} 882 883/* 884 * Retrieve incoming source route for use in replies, 885 * in the same form used by setsockopt. 886 * The first hop is placed before the options, will be removed later. 887 */ 888struct mbuf * 889ip_srcroute() 890{ 891 register struct in_addr *p, *q; 892 register struct mbuf *m; 893 894 if (ip_nhops == 0) 895 return ((struct mbuf *)0); 896 m = m_get(M_DONTWAIT, MT_SOOPTS); 897 if (m == 0) 898 return ((struct mbuf *)0); 899 900#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 901 902 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 903 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 904 OPTSIZ; 905#ifdef DIAGNOSTIC 906 if (ipprintfs) 907 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 908#endif 909 910 /* 911 * First save first hop for return route 912 */ 913 p = &ip_srcrt.route[ip_nhops - 1]; 914 *(mtod(m, struct in_addr *)) = *p--; 915#ifdef DIAGNOSTIC 916 if (ipprintfs) 917 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 918#endif 919 920 /* 921 * Copy option fields and padding (nop) to mbuf. 922 */ 923 ip_srcrt.nop = IPOPT_NOP; 924 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 925 bcopy((caddr_t)&ip_srcrt.nop, 926 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 927 q = (struct in_addr *)(mtod(m, caddr_t) + 928 sizeof(struct in_addr) + OPTSIZ); 929#undef OPTSIZ 930 /* 931 * Record return path as an IP source route, 932 * reversing the path (pointers are now aligned). 933 */ 934 while (p >= ip_srcrt.route) { 935#ifdef DIAGNOSTIC 936 if (ipprintfs) 937 printf(" %lx", ntohl(q->s_addr)); 938#endif 939 *q++ = *p--; 940 } 941 /* 942 * Last hop goes to final destination. 943 */ 944 *q = ip_srcrt.dst; 945#ifdef DIAGNOSTIC 946 if (ipprintfs) 947 printf(" %lx\n", ntohl(q->s_addr)); 948#endif 949 return (m); 950} 951 952/* 953 * Strip out IP options, at higher 954 * level protocol in the kernel. 955 * Second argument is buffer to which options 956 * will be moved, and return value is their length. 957 * XXX should be deleted; last arg currently ignored. 958 */ 959void 960ip_stripoptions(m, mopt) 961 register struct mbuf *m; 962 struct mbuf *mopt; 963{ 964 register int i; 965 struct ip *ip = mtod(m, struct ip *); 966 register caddr_t opts; 967 int olen; 968 969 olen = (ip->ip_hl<<2) - sizeof (struct ip); 970 opts = (caddr_t)(ip + 1); 971 i = m->m_len - (sizeof (struct ip) + olen); 972 bcopy(opts + olen, opts, (unsigned)i); 973 m->m_len -= olen; 974 if (m->m_flags & M_PKTHDR) 975 m->m_pkthdr.len -= olen; 976 ip->ip_hl = sizeof(struct ip) >> 2; 977} 978 979u_char inetctlerrmap[PRC_NCMDS] = { 980 0, 0, 0, 0, 981 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 982 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 983 EMSGSIZE, EHOSTUNREACH, 0, 0, 984 0, 0, 0, 0, 985 ENOPROTOOPT 986}; 987 988/* 989 * Forward a packet. If some error occurs return the sender 990 * an icmp packet. Note we can't always generate a meaningful 991 * icmp message because icmp doesn't have a large enough repertoire 992 * of codes and types. 993 * 994 * If not forwarding, just drop the packet. This could be confusing 995 * if ipforwarding was zero but some routing protocol was advancing 996 * us as a gateway to somewhere. However, we must let the routing 997 * protocol deal with that. 998 * 999 * The srcrt parameter indicates whether the packet is being forwarded 1000 * via a source route. 1001 */ 1002void 1003ip_forward(m, srcrt) 1004 struct mbuf *m; 1005 int srcrt; 1006{ 1007 register struct ip *ip = mtod(m, struct ip *); 1008 register struct sockaddr_in *sin; 1009 register struct rtentry *rt; 1010 int error, type = 0, code; 1011 struct mbuf *mcopy; 1012 n_long dest; 1013 struct ifnet *destifp; 1014 1015 dest = 0; 1016#ifdef DIAGNOSTIC 1017 if (ipprintfs) 1018 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1019 ip->ip_dst, ip->ip_ttl); 1020#endif 1021 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1022 ipstat.ips_cantforward++; 1023 m_freem(m); 1024 return; 1025 } 1026 HTONS(ip->ip_id); 1027 if (ip->ip_ttl <= IPTTLDEC) { 1028 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1029 return; 1030 } 1031 ip->ip_ttl -= IPTTLDEC; 1032 1033 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1034 if ((rt = ipforward_rt.ro_rt) == 0 || 1035 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1036 if (ipforward_rt.ro_rt) { 1037 RTFREE(ipforward_rt.ro_rt); 1038 ipforward_rt.ro_rt = 0; 1039 } 1040 sin->sin_family = AF_INET; 1041 sin->sin_len = sizeof(*sin); 1042 sin->sin_addr = ip->ip_dst; 1043 1044 rtalloc(&ipforward_rt); 1045 if (ipforward_rt.ro_rt == 0) { 1046 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1047 return; 1048 } 1049 rt = ipforward_rt.ro_rt; 1050 } 1051 1052 /* 1053 * Save at most 64 bytes of the packet in case 1054 * we need to generate an ICMP message to the src. 1055 */ 1056 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1057 1058#ifdef GATEWAY 1059 ip_ifmatrix[rt->rt_ifp->if_index + 1060 if_index * m->m_pkthdr.rcvif->if_index]++; 1061#endif 1062 /* 1063 * If forwarding packet using same interface that it came in on, 1064 * perhaps should send a redirect to sender to shortcut a hop. 1065 * Only send redirect if source is sending directly to us, 1066 * and if packet was not source routed (or has any options). 1067 * Also, don't send redirect if forwarding using a default route 1068 * or a route modified by a redirect. 1069 */ 1070#define satosin(sa) ((struct sockaddr_in *)(sa)) 1071 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1072 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1073 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1074 ipsendredirects && !srcrt) { 1075#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1076 u_long src = ntohl(ip->ip_src.s_addr); 1077 1078 if (RTA(rt) && 1079 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1080 if (rt->rt_flags & RTF_GATEWAY) 1081 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1082 else 1083 dest = ip->ip_dst.s_addr; 1084 /* Router requirements says to only send host redirects */ 1085 type = ICMP_REDIRECT; 1086 code = ICMP_REDIRECT_HOST; 1087#ifdef DIAGNOSTIC 1088 if (ipprintfs) 1089 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1090#endif 1091 } 1092 } 1093 1094 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1095#ifdef DIRECTED_BROADCAST 1096 | IP_ALLOWBROADCAST 1097#endif 1098 , 0); 1099 if (error) 1100 ipstat.ips_cantforward++; 1101 else { 1102 ipstat.ips_forward++; 1103 if (type) 1104 ipstat.ips_redirectsent++; 1105 else { 1106 if (mcopy) 1107 m_freem(mcopy); 1108 return; 1109 } 1110 } 1111 if (mcopy == NULL) 1112 return; 1113 destifp = NULL; 1114 1115 switch (error) { 1116 1117 case 0: /* forwarded, but need redirect */ 1118 /* type, code set above */ 1119 break; 1120 1121 case ENETUNREACH: /* shouldn't happen, checked above */ 1122 case EHOSTUNREACH: 1123 case ENETDOWN: 1124 case EHOSTDOWN: 1125 default: 1126 type = ICMP_UNREACH; 1127 code = ICMP_UNREACH_HOST; 1128 break; 1129 1130 case EMSGSIZE: 1131 type = ICMP_UNREACH; 1132 code = ICMP_UNREACH_NEEDFRAG; 1133 if (ipforward_rt.ro_rt) 1134 destifp = ipforward_rt.ro_rt->rt_ifp; 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, destifp); 1144} 1145 1146int 1147ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1148 int *name; 1149 u_int namelen; 1150 void *oldp; 1151 size_t *oldlenp; 1152 void *newp; 1153 size_t newlen; 1154{ 1155 /* All sysctl names at this level are terminal. */ 1156 if (namelen != 1) 1157 return (ENOTDIR); 1158 1159 switch (name[0]) { 1160 case IPCTL_FORWARDING: 1161 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1162 case IPCTL_SENDREDIRECTS: 1163 return (sysctl_int(oldp, oldlenp, newp, newlen, 1164 &ipsendredirects)); 1165 case IPCTL_DEFTTL: 1166 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1167#ifdef notyet 1168 case IPCTL_DEFMTU: 1169 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1170#endif 1171 default: 1172 return (EOPNOTSUPP); 1173 } 1174 /* NOTREACHED */ 1175} 1176