ip_input.c revision 7090
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 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 34 * $Id: ip_input.c,v 1.17 1995/02/14 23:04:52 wollman 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#include <vm/vm.h> 48#include <sys/sysctl.h> 49 50#include <net/if.h> 51#include <net/route.h> 52 53#include <netinet/in.h> 54#include <netinet/in_systm.h> 55#include <netinet/in_var.h> 56#include <netinet/ip.h> 57#include <netinet/in_pcb.h> 58#include <netinet/in_var.h> 59#include <netinet/ip_var.h> 60#include <netinet/ip_icmp.h> 61 62#include <netinet/ip_fw.h> 63 64#include <sys/socketvar.h> 65struct socket *ip_rsvpd; 66 67#ifndef IPFORWARDING 68#ifdef GATEWAY 69#define IPFORWARDING 1 /* forward IP packets not for us */ 70#else /* GATEWAY */ 71#define IPFORWARDING 0 /* don't forward IP packets not for us */ 72#endif /* GATEWAY */ 73#endif /* IPFORWARDING */ 74#ifndef IPSENDREDIRECTS 75#define IPSENDREDIRECTS 1 76#endif 77int ipforwarding = IPFORWARDING; 78int ipsendredirects = IPSENDREDIRECTS; 79int ip_defttl = IPDEFTTL; 80#ifdef DIAGNOSTIC 81int ipprintfs = 0; 82#endif 83 84extern struct domain inetdomain; 85extern struct protosw inetsw[]; 86u_char ip_protox[IPPROTO_MAX]; 87int ipqmaxlen = IFQ_MAXLEN; 88struct in_ifaddr *in_ifaddr; /* first inet address */ 89struct ifqueue ipintrq; 90 91struct ipstat ipstat; 92struct ipq ipq; 93 94/* 95 * We need to save the IP options in case a protocol wants to respond 96 * to an incoming packet over the same route if the packet got here 97 * using IP source routing. This allows connection establishment and 98 * maintenance when the remote end is on a network that is not known 99 * to us. 100 */ 101int ip_nhops = 0; 102static struct ip_srcrt { 103 struct in_addr dst; /* final destination */ 104 char nop; /* one NOP to align */ 105 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 106 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 107} ip_srcrt; 108 109#ifdef GATEWAY 110extern int if_index; 111u_long *ip_ifmatrix; 112#endif 113 114static void save_rte __P((u_char *, struct in_addr)); 115/* 116 * IP initialization: fill in IP protocol switch table. 117 * All protocols not implemented in kernel go to raw IP protocol handler. 118 */ 119void 120ip_init() 121{ 122 register struct protosw *pr; 123 register int i; 124 125 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 126 if (pr == 0) 127 panic("ip_init"); 128 for (i = 0; i < IPPROTO_MAX; i++) 129 ip_protox[i] = pr - inetsw; 130 for (pr = inetdomain.dom_protosw; 131 pr < inetdomain.dom_protoswNPROTOSW; pr++) 132 if (pr->pr_domain->dom_family == PF_INET && 133 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 134 ip_protox[pr->pr_protocol] = pr - inetsw; 135 ipq.next = ipq.prev = &ipq; 136 ip_id = time.tv_sec & 0xffff; 137 ipintrq.ifq_maxlen = ipqmaxlen; 138#ifdef GATEWAY 139 i = (if_index + 1) * (if_index + 1) * sizeof (u_long); 140 ip_ifmatrix = (u_long *) malloc(i, M_RTABLE, M_WAITOK); 141 bzero((char *)ip_ifmatrix, i); 142#endif 143} 144 145struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 146struct route ipforward_rt; 147 148/* 149 * Ip input routine. Checksum and byte swap header. If fragmented 150 * try to reassemble. Process options. Pass to next level. 151 */ 152void 153ipintr() 154{ 155 register struct ip *ip; 156 register struct mbuf *m; 157 register struct ipq *fp; 158 register struct in_ifaddr *ia; 159 int hlen, s; 160 161next: 162 /* 163 * Get next datagram off input queue and get IP header 164 * in first mbuf. 165 */ 166 s = splimp(); 167 IF_DEQUEUE(&ipintrq, m); 168 splx(s); 169 if (m == 0) 170 return; 171#ifdef DIAGNOSTIC 172 if ((m->m_flags & M_PKTHDR) == 0) 173 panic("ipintr no HDR"); 174#endif 175 /* 176 * If no IP addresses have been set yet but the interfaces 177 * are receiving, can't do anything with incoming packets yet. 178 */ 179 if (in_ifaddr == NULL) 180 goto bad; 181 ipstat.ips_total++; 182 if (m->m_len < sizeof (struct ip) && 183 (m = m_pullup(m, sizeof (struct ip))) == 0) { 184 ipstat.ips_toosmall++; 185 goto next; 186 } 187 ip = mtod(m, struct ip *); 188 if (ip->ip_v != IPVERSION) { 189 ipstat.ips_badvers++; 190 goto bad; 191 } 192 hlen = ip->ip_hl << 2; 193 if (hlen < sizeof(struct ip)) { /* minimum header length */ 194 ipstat.ips_badhlen++; 195 goto bad; 196 } 197 if (hlen > m->m_len) { 198 if ((m = m_pullup(m, hlen)) == 0) { 199 ipstat.ips_badhlen++; 200 goto next; 201 } 202 ip = mtod(m, struct ip *); 203 } 204 ip->ip_sum = in_cksum(m, hlen); 205 if (ip->ip_sum) { 206 ipstat.ips_badsum++; 207 goto bad; 208 } 209 210 /* 211 * Convert fields to host representation. 212 */ 213 NTOHS(ip->ip_len); 214 if (ip->ip_len < hlen) { 215 ipstat.ips_badlen++; 216 goto bad; 217 } 218 NTOHS(ip->ip_id); 219 NTOHS(ip->ip_off); 220 221 /* 222 * Check that the amount of data in the buffers 223 * is as at least much as the IP header would have us expect. 224 * Trim mbufs if longer than we expect. 225 * Drop packet if shorter than we expect. 226 */ 227 if (m->m_pkthdr.len < ip->ip_len) { 228 ipstat.ips_tooshort++; 229 goto bad; 230 } 231 if (m->m_pkthdr.len > ip->ip_len) { 232 if (m->m_len == m->m_pkthdr.len) { 233 m->m_len = ip->ip_len; 234 m->m_pkthdr.len = ip->ip_len; 235 } else 236 m_adj(m, ip->ip_len - m->m_pkthdr.len); 237 } 238 /* 239 * IpHack's section. 240 * Right now when no processing on packet has done 241 * and it is still fresh out of network we do our black 242 * deals with it. 243 * - Firewall: deny/allow 244 * - Wrap: fake packet's addr/port <unimpl.> 245 * - Encapsulate: put it in another IP and send out. <unimp.> 246 */ 247 248 if (ip_fw_chk_ptr!=NULL) 249 if (!(*ip_fw_chk_ptr)(ip,m->m_pkthdr.rcvif,ip_fw_chain) ) { 250 goto bad; 251 } 252 253 /* 254 * Process options and, if not destined for us, 255 * ship it on. ip_dooptions returns 1 when an 256 * error was detected (causing an icmp message 257 * to be sent and the original packet to be freed). 258 */ 259 ip_nhops = 0; /* for source routed packets */ 260 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 261 goto next; 262 263 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no 264 * matter if it is destined to another node, or whether it is 265 * a multicast one, RSVP wants it! and prevents it from being forwarded 266 * anywhere else. Also checks if the rsvp daemon is running before 267 * grabbing the packet. 268 */ 269 if (ip_rsvpd != NULL && ip->ip_p==IPPROTO_RSVP) 270 goto ours; 271 272 /* 273 * Check our list of addresses, to see if the packet is for us. 274 */ 275 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 276#define satosin(sa) ((struct sockaddr_in *)(sa)) 277 278 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 279 goto ours; 280 if ( 281#ifdef DIRECTED_BROADCAST 282 ia->ia_ifp == m->m_pkthdr.rcvif && 283#endif 284 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 285 u_long t; 286 287 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 288 ip->ip_dst.s_addr) 289 goto ours; 290 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 291 goto ours; 292 /* 293 * Look for all-0's host part (old broadcast addr), 294 * either for subnet or net. 295 */ 296 t = ntohl(ip->ip_dst.s_addr); 297 if (t == ia->ia_subnet) 298 goto ours; 299 if (t == ia->ia_net) 300 goto ours; 301 } 302 } 303 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 304 struct in_multi *inm; 305 if (ip_mrouter) { 306 /* 307 * If we are acting as a multicast router, all 308 * incoming multicast packets are passed to the 309 * kernel-level multicast forwarding function. 310 * The packet is returned (relatively) intact; if 311 * ip_mforward() returns a non-zero value, the packet 312 * must be discarded, else it may be accepted below. 313 * 314 * (The IP ident field is put in the same byte order 315 * as expected when ip_mforward() is called from 316 * ip_output().) 317 */ 318 ip->ip_id = htons(ip->ip_id); 319 if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) { 320 ipstat.ips_cantforward++; 321 m_freem(m); 322 goto next; 323 } 324 ip->ip_id = ntohs(ip->ip_id); 325 326 /* 327 * The process-level routing demon needs to receive 328 * all multicast IGMP packets, whether or not this 329 * host belongs to their destination groups. 330 */ 331 if (ip->ip_p == IPPROTO_IGMP) 332 goto ours; 333 ipstat.ips_forward++; 334 } 335 /* 336 * See if we belong to the destination multicast group on the 337 * arrival interface. 338 */ 339 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 340 if (inm == NULL) { 341 ipstat.ips_cantforward++; 342 m_freem(m); 343 goto next; 344 } 345 goto ours; 346 } 347 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 348 goto ours; 349 if (ip->ip_dst.s_addr == INADDR_ANY) 350 goto ours; 351 352 /* 353 * Not for us; forward if possible and desirable. 354 */ 355 if (ipforwarding == 0) { 356 ipstat.ips_cantforward++; 357 m_freem(m); 358 } else 359 ip_forward(m, 0); 360 goto next; 361 362ours: 363 364 /* 365 * If packet came to us we count it... 366 * This way we count all incoming packets which has 367 * not been forwarded... 368 * Do not convert ip_len to host byte order when 369 * counting,ppl already made it for us before.. 370 */ 371 if (ip_acct_cnt_ptr!=NULL) 372 (*ip_acct_cnt_ptr)(ip,m->m_pkthdr.rcvif,ip_acct_chain,0); 373 374 /* 375 * If offset or IP_MF are set, must reassemble. 376 * Otherwise, nothing need be done. 377 * (We could look in the reassembly queue to see 378 * if the packet was previously fragmented, 379 * but it's not worth the time; just let them time out.) 380 */ 381 if (ip->ip_off &~ IP_DF) { 382 if (m->m_flags & M_EXT) { /* XXX */ 383 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 384 ipstat.ips_toosmall++; 385 goto next; 386 } 387 ip = mtod(m, struct ip *); 388 } 389 /* 390 * Look for queue of fragments 391 * of this datagram. 392 */ 393 for (fp = ipq.next; fp != &ipq; fp = fp->next) 394 if (ip->ip_id == fp->ipq_id && 395 ip->ip_src.s_addr == fp->ipq_src.s_addr && 396 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 397 ip->ip_p == fp->ipq_p) 398 goto found; 399 fp = 0; 400found: 401 402 /* 403 * Adjust ip_len to not reflect header, 404 * set ip_mff if more fragments are expected, 405 * convert offset of this to bytes. 406 */ 407 ip->ip_len -= hlen; 408 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 409 if (ip->ip_off & IP_MF) 410 ((struct ipasfrag *)ip)->ipf_mff |= 1; 411 ip->ip_off <<= 3; 412 413 /* 414 * If datagram marked as having more fragments 415 * or if this is not the first fragment, 416 * attempt reassembly; if it succeeds, proceed. 417 */ 418 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 419 ipstat.ips_fragments++; 420 ip = ip_reass((struct ipasfrag *)ip, fp); 421 if (ip == 0) 422 goto next; 423 ipstat.ips_reassembled++; 424 m = dtom(ip); 425 } else 426 if (fp) 427 ip_freef(fp); 428 } else 429 ip->ip_len -= hlen; 430 431 /* 432 * Switch out to protocol's input routine. 433 */ 434 ipstat.ips_delivered++; 435 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 436 goto next; 437bad: 438 m_freem(m); 439 goto next; 440} 441 442/* 443 * Take incoming datagram fragment and try to 444 * reassemble it into whole datagram. If a chain for 445 * reassembly of this datagram already exists, then it 446 * is given as fp; otherwise have to make a chain. 447 */ 448struct ip * 449ip_reass(ip, fp) 450 register struct ipasfrag *ip; 451 register struct ipq *fp; 452{ 453 register struct mbuf *m = dtom(ip); 454 register struct ipasfrag *q; 455 struct mbuf *t; 456 int hlen = ip->ip_hl << 2; 457 int i, next; 458 459 /* 460 * Presence of header sizes in mbufs 461 * would confuse code below. 462 */ 463 m->m_data += hlen; 464 m->m_len -= hlen; 465 466 /* 467 * If first fragment to arrive, create a reassembly queue. 468 */ 469 if (fp == 0) { 470 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 471 goto dropfrag; 472 fp = mtod(t, struct ipq *); 473 insque(fp, &ipq); 474 fp->ipq_ttl = IPFRAGTTL; 475 fp->ipq_p = ip->ip_p; 476 fp->ipq_id = ip->ip_id; 477 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 478 fp->ipq_src = ((struct ip *)ip)->ip_src; 479 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 480 q = (struct ipasfrag *)fp; 481 goto insert; 482 } 483 484 /* 485 * Find a segment which begins after this one does. 486 */ 487 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 488 if (q->ip_off > ip->ip_off) 489 break; 490 491 /* 492 * If there is a preceding segment, it may provide some of 493 * our data already. If so, drop the data from the incoming 494 * segment. If it provides all of our data, drop us. 495 */ 496 if (q->ipf_prev != (struct ipasfrag *)fp) { 497 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 498 if (i > 0) { 499 if (i >= ip->ip_len) 500 goto dropfrag; 501 m_adj(dtom(ip), i); 502 ip->ip_off += i; 503 ip->ip_len -= i; 504 } 505 } 506 507 /* 508 * While we overlap succeeding segments trim them or, 509 * if they are completely covered, dequeue them. 510 */ 511 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 512 i = (ip->ip_off + ip->ip_len) - q->ip_off; 513 if (i < q->ip_len) { 514 q->ip_len -= i; 515 q->ip_off += i; 516 m_adj(dtom(q), i); 517 break; 518 } 519 q = q->ipf_next; 520 m_freem(dtom(q->ipf_prev)); 521 ip_deq(q->ipf_prev); 522 } 523 524insert: 525 /* 526 * Stick new segment in its place; 527 * check for complete reassembly. 528 */ 529 ip_enq(ip, q->ipf_prev); 530 next = 0; 531 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 532 if (q->ip_off != next) 533 return (0); 534 next += q->ip_len; 535 } 536 if (q->ipf_prev->ipf_mff & 1) 537 return (0); 538 539 /* 540 * Reassembly is complete; concatenate fragments. 541 */ 542 q = fp->ipq_next; 543 m = dtom(q); 544 t = m->m_next; 545 m->m_next = 0; 546 m_cat(m, t); 547 q = q->ipf_next; 548 while (q != (struct ipasfrag *)fp) { 549 t = dtom(q); 550 q = q->ipf_next; 551 m_cat(m, t); 552 } 553 554 /* 555 * Create header for new ip packet by 556 * modifying header of first packet; 557 * dequeue and discard fragment reassembly header. 558 * Make header visible. 559 */ 560 ip = fp->ipq_next; 561 ip->ip_len = next; 562 ip->ipf_mff &= ~1; 563 ((struct ip *)ip)->ip_src = fp->ipq_src; 564 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 565 remque(fp); 566 (void) m_free(dtom(fp)); 567 m = dtom(ip); 568 m->m_len += (ip->ip_hl << 2); 569 m->m_data -= (ip->ip_hl << 2); 570 /* some debugging cruft by sklower, below, will go away soon */ 571 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 572 register int plen = 0; 573 for (t = m; m; m = m->m_next) 574 plen += m->m_len; 575 t->m_pkthdr.len = plen; 576 } 577 return ((struct ip *)ip); 578 579dropfrag: 580 ipstat.ips_fragdropped++; 581 m_freem(m); 582 return (0); 583} 584 585/* 586 * Free a fragment reassembly header and all 587 * associated datagrams. 588 */ 589void 590ip_freef(fp) 591 struct ipq *fp; 592{ 593 register struct ipasfrag *q, *p; 594 595 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 596 p = q->ipf_next; 597 ip_deq(q); 598 m_freem(dtom(q)); 599 } 600 remque(fp); 601 (void) m_free(dtom(fp)); 602} 603 604/* 605 * Put an ip fragment on a reassembly chain. 606 * Like insque, but pointers in middle of structure. 607 */ 608void 609ip_enq(p, prev) 610 register struct ipasfrag *p, *prev; 611{ 612 613 p->ipf_prev = prev; 614 p->ipf_next = prev->ipf_next; 615 prev->ipf_next->ipf_prev = p; 616 prev->ipf_next = p; 617} 618 619/* 620 * To ip_enq as remque is to insque. 621 */ 622void 623ip_deq(p) 624 register struct ipasfrag *p; 625{ 626 627 p->ipf_prev->ipf_next = p->ipf_next; 628 p->ipf_next->ipf_prev = p->ipf_prev; 629} 630 631/* 632 * IP timer processing; 633 * if a timer expires on a reassembly 634 * queue, discard it. 635 */ 636void 637ip_slowtimo() 638{ 639 register struct ipq *fp; 640 int s = splnet(); 641 642 fp = ipq.next; 643 if (fp == 0) { 644 splx(s); 645 return; 646 } 647 while (fp != &ipq) { 648 --fp->ipq_ttl; 649 fp = fp->next; 650 if (fp->prev->ipq_ttl == 0) { 651 ipstat.ips_fragtimeout++; 652 ip_freef(fp->prev); 653 } 654 } 655 splx(s); 656} 657 658/* 659 * Drain off all datagram fragments. 660 */ 661void 662ip_drain() 663{ 664 665 while (ipq.next != &ipq) { 666 ipstat.ips_fragdropped++; 667 ip_freef(ipq.next); 668 } 669} 670 671/* 672 * Do option processing on a datagram, 673 * possibly discarding it if bad options are encountered, 674 * or forwarding it if source-routed. 675 * Returns 1 if packet has been forwarded/freed, 676 * 0 if the packet should be processed further. 677 */ 678int 679ip_dooptions(m) 680 struct mbuf *m; 681{ 682 register struct ip *ip = mtod(m, struct ip *); 683 register u_char *cp; 684 register struct ip_timestamp *ipt; 685 register struct in_ifaddr *ia; 686 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 687 struct in_addr *sin, dst; 688 n_time ntime; 689 690 dst = ip->ip_dst; 691 cp = (u_char *)(ip + 1); 692 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 693 for (; cnt > 0; cnt -= optlen, cp += optlen) { 694 opt = cp[IPOPT_OPTVAL]; 695 if (opt == IPOPT_EOL) 696 break; 697 if (opt == IPOPT_NOP) 698 optlen = 1; 699 else { 700 optlen = cp[IPOPT_OLEN]; 701 if (optlen <= 0 || optlen > cnt) { 702 code = &cp[IPOPT_OLEN] - (u_char *)ip; 703 goto bad; 704 } 705 } 706 switch (opt) { 707 708 default: 709 break; 710 711 /* 712 * Source routing with record. 713 * Find interface with current destination address. 714 * If none on this machine then drop if strictly routed, 715 * or do nothing if loosely routed. 716 * Record interface address and bring up next address 717 * component. If strictly routed make sure next 718 * address is on directly accessible net. 719 */ 720 case IPOPT_LSRR: 721 case IPOPT_SSRR: 722 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 723 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 724 goto bad; 725 } 726 ipaddr.sin_addr = ip->ip_dst; 727 ia = (struct in_ifaddr *) 728 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 729 if (ia == 0) { 730 if (opt == IPOPT_SSRR) { 731 type = ICMP_UNREACH; 732 code = ICMP_UNREACH_SRCFAIL; 733 goto bad; 734 } 735 /* 736 * Loose routing, and not at next destination 737 * yet; nothing to do except forward. 738 */ 739 break; 740 } 741 off--; /* 0 origin */ 742 if (off > optlen - sizeof(struct in_addr)) { 743 /* 744 * End of source route. Should be for us. 745 */ 746 save_rte(cp, ip->ip_src); 747 break; 748 } 749 /* 750 * locate outgoing interface 751 */ 752 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 753 sizeof(ipaddr.sin_addr)); 754 if (opt == IPOPT_SSRR) { 755#define INA struct in_ifaddr * 756#define SA struct sockaddr * 757 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 758 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 759 } else 760 ia = ip_rtaddr(ipaddr.sin_addr); 761 if (ia == 0) { 762 type = ICMP_UNREACH; 763 code = ICMP_UNREACH_SRCFAIL; 764 goto bad; 765 } 766 ip->ip_dst = ipaddr.sin_addr; 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 /* 771 * Let ip_intr's mcast routing check handle mcast pkts 772 */ 773 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 774 break; 775 776 case IPOPT_RR: 777 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 778 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 779 goto bad; 780 } 781 /* 782 * If no space remains, ignore. 783 */ 784 off--; /* 0 origin */ 785 if (off > optlen - sizeof(struct in_addr)) 786 break; 787 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 788 sizeof(ipaddr.sin_addr)); 789 /* 790 * locate outgoing interface; if we're the destination, 791 * use the incoming interface (should be same). 792 */ 793 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 794 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 795 type = ICMP_UNREACH; 796 code = ICMP_UNREACH_HOST; 797 goto bad; 798 } 799 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 800 (caddr_t)(cp + off), sizeof(struct in_addr)); 801 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 802 break; 803 804 case IPOPT_TS: 805 code = cp - (u_char *)ip; 806 ipt = (struct ip_timestamp *)cp; 807 if (ipt->ipt_len < 5) 808 goto bad; 809 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 810 if (++ipt->ipt_oflw == 0) 811 goto bad; 812 break; 813 } 814 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 815 switch (ipt->ipt_flg) { 816 817 case IPOPT_TS_TSONLY: 818 break; 819 820 case IPOPT_TS_TSANDADDR: 821 if (ipt->ipt_ptr + sizeof(n_time) + 822 sizeof(struct in_addr) > ipt->ipt_len) 823 goto bad; 824 ipaddr.sin_addr = dst; 825 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 826 m->m_pkthdr.rcvif); 827 if (ia == 0) 828 continue; 829 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 830 (caddr_t)sin, sizeof(struct in_addr)); 831 ipt->ipt_ptr += sizeof(struct in_addr); 832 break; 833 834 case IPOPT_TS_PRESPEC: 835 if (ipt->ipt_ptr + sizeof(n_time) + 836 sizeof(struct in_addr) > ipt->ipt_len) 837 goto bad; 838 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 839 sizeof(struct in_addr)); 840 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 841 continue; 842 ipt->ipt_ptr += sizeof(struct in_addr); 843 break; 844 845 default: 846 goto bad; 847 } 848 ntime = iptime(); 849 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 850 sizeof(n_time)); 851 ipt->ipt_ptr += sizeof(n_time); 852 } 853 } 854 if (forward) { 855 ip_forward(m, 1); 856 return (1); 857 } 858 return (0); 859bad: 860 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 861 icmp_error(m, type, code, 0, 0); 862 ipstat.ips_badoptions++; 863 return (1); 864} 865 866/* 867 * Given address of next destination (final or next hop), 868 * return internet address info of interface to be used to get there. 869 */ 870struct in_ifaddr * 871ip_rtaddr(dst) 872 struct in_addr dst; 873{ 874 register struct sockaddr_in *sin; 875 876 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 877 878 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 879 if (ipforward_rt.ro_rt) { 880 RTFREE(ipforward_rt.ro_rt); 881 ipforward_rt.ro_rt = 0; 882 } 883 sin->sin_family = AF_INET; 884 sin->sin_len = sizeof(*sin); 885 sin->sin_addr = dst; 886 887 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 888 } 889 if (ipforward_rt.ro_rt == 0) 890 return ((struct in_ifaddr *)0); 891 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 892} 893 894/* 895 * Save incoming source route for use in replies, 896 * to be picked up later by ip_srcroute if the receiver is interested. 897 */ 898void 899save_rte(option, dst) 900 u_char *option; 901 struct in_addr dst; 902{ 903 unsigned olen; 904 905 olen = option[IPOPT_OLEN]; 906#ifdef DIAGNOSTIC 907 if (ipprintfs) 908 printf("save_rte: olen %d\n", olen); 909#endif 910 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 911 return; 912 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 913 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 914 ip_srcrt.dst = dst; 915} 916 917/* 918 * Retrieve incoming source route for use in replies, 919 * in the same form used by setsockopt. 920 * The first hop is placed before the options, will be removed later. 921 */ 922struct mbuf * 923ip_srcroute() 924{ 925 register struct in_addr *p, *q; 926 register struct mbuf *m; 927 928 if (ip_nhops == 0) 929 return ((struct mbuf *)0); 930 m = m_get(M_DONTWAIT, MT_SOOPTS); 931 if (m == 0) 932 return ((struct mbuf *)0); 933 934#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 935 936 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 937 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 938 OPTSIZ; 939#ifdef DIAGNOSTIC 940 if (ipprintfs) 941 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 942#endif 943 944 /* 945 * First save first hop for return route 946 */ 947 p = &ip_srcrt.route[ip_nhops - 1]; 948 *(mtod(m, struct in_addr *)) = *p--; 949#ifdef DIAGNOSTIC 950 if (ipprintfs) 951 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 952#endif 953 954 /* 955 * Copy option fields and padding (nop) to mbuf. 956 */ 957 ip_srcrt.nop = IPOPT_NOP; 958 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 959 bcopy((caddr_t)&ip_srcrt.nop, 960 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 961 q = (struct in_addr *)(mtod(m, caddr_t) + 962 sizeof(struct in_addr) + OPTSIZ); 963#undef OPTSIZ 964 /* 965 * Record return path as an IP source route, 966 * reversing the path (pointers are now aligned). 967 */ 968 while (p >= ip_srcrt.route) { 969#ifdef DIAGNOSTIC 970 if (ipprintfs) 971 printf(" %lx", ntohl(q->s_addr)); 972#endif 973 *q++ = *p--; 974 } 975 /* 976 * Last hop goes to final destination. 977 */ 978 *q = ip_srcrt.dst; 979#ifdef DIAGNOSTIC 980 if (ipprintfs) 981 printf(" %lx\n", ntohl(q->s_addr)); 982#endif 983 return (m); 984} 985 986/* 987 * Strip out IP options, at higher 988 * level protocol in the kernel. 989 * Second argument is buffer to which options 990 * will be moved, and return value is their length. 991 * XXX should be deleted; last arg currently ignored. 992 */ 993void 994ip_stripoptions(m, mopt) 995 register struct mbuf *m; 996 struct mbuf *mopt; 997{ 998 register int i; 999 struct ip *ip = mtod(m, struct ip *); 1000 register caddr_t opts; 1001 int olen; 1002 1003 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1004 opts = (caddr_t)(ip + 1); 1005 i = m->m_len - (sizeof (struct ip) + olen); 1006 bcopy(opts + olen, opts, (unsigned)i); 1007 m->m_len -= olen; 1008 if (m->m_flags & M_PKTHDR) 1009 m->m_pkthdr.len -= olen; 1010 ip->ip_hl = sizeof(struct ip) >> 2; 1011} 1012 1013u_char inetctlerrmap[PRC_NCMDS] = { 1014 0, 0, 0, 0, 1015 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1016 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1017 EMSGSIZE, EHOSTUNREACH, 0, 0, 1018 0, 0, 0, 0, 1019 ENOPROTOOPT 1020}; 1021 1022/* 1023 * Forward a packet. If some error occurs return the sender 1024 * an icmp packet. Note we can't always generate a meaningful 1025 * icmp message because icmp doesn't have a large enough repertoire 1026 * of codes and types. 1027 * 1028 * If not forwarding, just drop the packet. This could be confusing 1029 * if ipforwarding was zero but some routing protocol was advancing 1030 * us as a gateway to somewhere. However, we must let the routing 1031 * protocol deal with that. 1032 * 1033 * The srcrt parameter indicates whether the packet is being forwarded 1034 * via a source route. 1035 */ 1036void 1037ip_forward(m, srcrt) 1038 struct mbuf *m; 1039 int srcrt; 1040{ 1041 register struct ip *ip = mtod(m, struct ip *); 1042 register struct sockaddr_in *sin; 1043 register struct rtentry *rt; 1044 int error, type = 0, code = 0; 1045 struct mbuf *mcopy; 1046 n_long dest; 1047 struct ifnet *destifp; 1048 1049 dest = 0; 1050#ifdef DIAGNOSTIC 1051 if (ipprintfs) 1052 printf("forward: src %lx dst %lx ttl %x\n", 1053 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1054#endif 1055 1056 1057 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1058 ipstat.ips_cantforward++; 1059 m_freem(m); 1060 return; 1061 } 1062 HTONS(ip->ip_id); 1063 if (ip->ip_ttl <= IPTTLDEC) { 1064 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1065 return; 1066 } 1067 ip->ip_ttl -= IPTTLDEC; 1068 1069 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1070 if ((rt = ipforward_rt.ro_rt) == 0 || 1071 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1072 if (ipforward_rt.ro_rt) { 1073 RTFREE(ipforward_rt.ro_rt); 1074 ipforward_rt.ro_rt = 0; 1075 } 1076 sin->sin_family = AF_INET; 1077 sin->sin_len = sizeof(*sin); 1078 sin->sin_addr = ip->ip_dst; 1079 1080 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1081 if (ipforward_rt.ro_rt == 0) { 1082 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1083 return; 1084 } 1085 rt = ipforward_rt.ro_rt; 1086 } 1087 1088 /* 1089 * Save at most 64 bytes of the packet in case 1090 * we need to generate an ICMP message to the src. 1091 */ 1092 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1093 1094#ifdef bogus 1095#ifdef GATEWAY 1096 ip_ifmatrix[rt->rt_ifp->if_index + 1097 if_index * m->m_pkthdr.rcvif->if_index]++; 1098#endif 1099#endif 1100 /* 1101 * If forwarding packet using same interface that it came in on, 1102 * perhaps should send a redirect to sender to shortcut a hop. 1103 * Only send redirect if source is sending directly to us, 1104 * and if packet was not source routed (or has any options). 1105 * Also, don't send redirect if forwarding using a default route 1106 * or a route modified by a redirect. 1107 */ 1108#define satosin(sa) ((struct sockaddr_in *)(sa)) 1109 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1110 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1111 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1112 ipsendredirects && !srcrt) { 1113#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1114 u_long src = ntohl(ip->ip_src.s_addr); 1115 1116 if (RTA(rt) && 1117 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1118 if (rt->rt_flags & RTF_GATEWAY) 1119 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1120 else 1121 dest = ip->ip_dst.s_addr; 1122 /* Router requirements says to only send host redirects */ 1123 type = ICMP_REDIRECT; 1124 code = ICMP_REDIRECT_HOST; 1125#ifdef DIAGNOSTIC 1126 if (ipprintfs) 1127 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1128#endif 1129 } 1130 } 1131 1132 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1133#ifdef DIRECTED_BROADCAST 1134 | IP_ALLOWBROADCAST 1135#endif 1136 , 0); 1137 if (error) 1138 ipstat.ips_cantforward++; 1139 else { 1140 ipstat.ips_forward++; 1141 if (type) 1142 ipstat.ips_redirectsent++; 1143 else { 1144 if (mcopy) 1145 m_freem(mcopy); 1146 return; 1147 } 1148 } 1149 if (mcopy == NULL) 1150 return; 1151 destifp = NULL; 1152 1153 switch (error) { 1154 1155 case 0: /* forwarded, but need redirect */ 1156 /* type, code set above */ 1157 break; 1158 1159 case ENETUNREACH: /* shouldn't happen, checked above */ 1160 case EHOSTUNREACH: 1161 case ENETDOWN: 1162 case EHOSTDOWN: 1163 default: 1164 type = ICMP_UNREACH; 1165 code = ICMP_UNREACH_HOST; 1166 break; 1167 1168 case EMSGSIZE: 1169 type = ICMP_UNREACH; 1170 code = ICMP_UNREACH_NEEDFRAG; 1171 if (ipforward_rt.ro_rt) 1172 destifp = ipforward_rt.ro_rt->rt_ifp; 1173 ipstat.ips_cantfrag++; 1174 break; 1175 1176 case ENOBUFS: 1177 type = ICMP_SOURCEQUENCH; 1178 code = 0; 1179 break; 1180 } 1181 icmp_error(mcopy, type, code, dest, destifp); 1182} 1183 1184int 1185ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1186 int *name; 1187 u_int namelen; 1188 void *oldp; 1189 size_t *oldlenp; 1190 void *newp; 1191 size_t newlen; 1192{ 1193 /* All sysctl names at this level are terminal. */ 1194 if (namelen != 1) 1195 return (ENOTDIR); 1196 1197 switch (name[0]) { 1198 case IPCTL_FORWARDING: 1199 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1200 case IPCTL_SENDREDIRECTS: 1201 return (sysctl_int(oldp, oldlenp, newp, newlen, 1202 &ipsendredirects)); 1203 case IPCTL_DEFTTL: 1204 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1205#ifdef notyet 1206 case IPCTL_DEFMTU: 1207 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1208#endif 1209 case IPCTL_RTEXPIRE: 1210 return (sysctl_int(oldp, oldlenp, newp, newlen, 1211 &rtq_reallyold)); 1212 case IPCTL_RTMINEXPIRE: 1213 return (sysctl_int(oldp, oldlenp, newp, newlen, 1214 &rtq_minreallyold)); 1215 case IPCTL_RTMAXCACHE: 1216 return (sysctl_int(oldp, oldlenp, newp, newlen, 1217 &rtq_toomany)); 1218 default: 1219 return (EOPNOTSUPP); 1220 } 1221 /* NOTREACHED */ 1222} 1223 1224int 1225ip_rsvp_init(struct socket *so) 1226{ 1227 if (so->so_type != SOCK_RAW || 1228 so->so_proto->pr_protocol != IPPROTO_RSVP) 1229 return EOPNOTSUPP; 1230 1231 if (ip_rsvpd != NULL) 1232 return EADDRINUSE; 1233 1234 ip_rsvpd = so; 1235 1236 return 0; 1237} 1238 1239int 1240ip_rsvp_done(void) 1241{ 1242 ip_rsvpd = NULL; 1243 return 0; 1244} 1245