ip_input.c revision 1.56
1/* $NetBSD: ip_input.c,v 1.56 1998/01/28 02:36:10 thorpej 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 "opt_mrouting.h" 39 40#include <sys/param.h> 41#include <sys/systm.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/domain.h> 45#include <sys/protosw.h> 46#include <sys/socket.h> 47#include <sys/socketvar.h> 48#include <sys/errno.h> 49#include <sys/time.h> 50#include <sys/kernel.h> 51#include <sys/proc.h> 52 53#include <vm/vm.h> 54#include <sys/sysctl.h> 55 56#include <net/if.h> 57#include <net/if_dl.h> 58#include <net/route.h> 59#include <net/pfil.h> 60 61#include <netinet/in.h> 62#include <netinet/in_systm.h> 63#include <netinet/ip.h> 64#include <netinet/in_pcb.h> 65#include <netinet/in_var.h> 66#include <netinet/ip_var.h> 67#include <netinet/ip_icmp.h> 68 69#ifndef IPFORWARDING 70#ifdef GATEWAY 71#define IPFORWARDING 1 /* forward IP packets not for us */ 72#else /* GATEWAY */ 73#define IPFORWARDING 0 /* don't forward IP packets not for us */ 74#endif /* GATEWAY */ 75#endif /* IPFORWARDING */ 76#ifndef IPSENDREDIRECTS 77#define IPSENDREDIRECTS 1 78#endif 79#ifndef IPFORWSRCRT 80#define IPFORWSRCRT 1 /* forward source-routed packets */ 81#endif 82#ifndef IPALLOWSRCRT 83#define IPALLOWSRCRT 1 /* allow source-routed packets */ 84#endif 85#ifndef IPMTUDISC 86#define IPMTUDISC 0 87#endif 88 89/* 90 * Note: DIRECTED_BROADCAST is handled this way so that previous 91 * configuration using this option will Just Work. 92 */ 93#ifndef IPDIRECTEDBCAST 94#ifdef DIRECTED_BROADCAST 95#define IPDIRECTEDBCAST 1 96#else 97#define IPDIRECTEDBCAST 0 98#endif /* DIRECTED_BROADCAST */ 99#endif /* IPDIRECTEDBCAST */ 100int ipforwarding = IPFORWARDING; 101int ipsendredirects = IPSENDREDIRECTS; 102int ip_defttl = IPDEFTTL; 103int ip_forwsrcrt = IPFORWSRCRT; 104int ip_directedbcast = IPDIRECTEDBCAST; 105int ip_allowsrcrt = IPALLOWSRCRT; 106int ip_mtudisc = IPMTUDISC; 107#ifdef DIAGNOSTIC 108int ipprintfs = 0; 109#endif 110 111extern struct domain inetdomain; 112extern struct protosw inetsw[]; 113u_char ip_protox[IPPROTO_MAX]; 114int ipqmaxlen = IFQ_MAXLEN; 115struct in_ifaddrhead in_ifaddr; 116struct ifqueue ipintrq; 117 118/* 119 * We need to save the IP options in case a protocol wants to respond 120 * to an incoming packet over the same route if the packet got here 121 * using IP source routing. This allows connection establishment and 122 * maintenance when the remote end is on a network that is not known 123 * to us. 124 */ 125int ip_nhops = 0; 126static struct ip_srcrt { 127 struct in_addr dst; /* final destination */ 128 char nop; /* one NOP to align */ 129 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 130 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 131} ip_srcrt; 132 133static void save_rte __P((u_char *, struct in_addr)); 134 135/* 136 * IP initialization: fill in IP protocol switch table. 137 * All protocols not implemented in kernel go to raw IP protocol handler. 138 */ 139void 140ip_init() 141{ 142 register struct protosw *pr; 143 register int i; 144 145 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 146 if (pr == 0) 147 panic("ip_init"); 148 for (i = 0; i < IPPROTO_MAX; i++) 149 ip_protox[i] = pr - inetsw; 150 for (pr = inetdomain.dom_protosw; 151 pr < inetdomain.dom_protoswNPROTOSW; pr++) 152 if (pr->pr_domain->dom_family == PF_INET && 153 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 154 ip_protox[pr->pr_protocol] = pr - inetsw; 155 LIST_INIT(&ipq); 156 ip_id = time.tv_sec & 0xffff; 157 ipintrq.ifq_maxlen = ipqmaxlen; 158 TAILQ_INIT(&in_ifaddr); 159} 160 161struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 162struct route ipforward_rt; 163 164/* 165 * Ip input routine. Checksum and byte swap header. If fragmented 166 * try to reassemble. Process options. Pass to next level. 167 */ 168void 169ipintr() 170{ 171 register struct ip *ip = NULL; 172 register struct mbuf *m; 173 register struct ipq *fp; 174 register struct in_ifaddr *ia; 175 struct ipqent *ipqe; 176 int hlen = 0, mff, len, s; 177#ifdef PFIL_HOOKS 178 struct packet_filter_hook *pfh; 179 struct mbuf *m0; 180 int rv; 181#endif /* PFIL_HOOKS */ 182 183next: 184 /* 185 * Get next datagram off input queue and get IP header 186 * in first mbuf. 187 */ 188 s = splimp(); 189 IF_DEQUEUE(&ipintrq, m); 190 splx(s); 191 if (m == 0) 192 return; 193#ifdef DIAGNOSTIC 194 if ((m->m_flags & M_PKTHDR) == 0) 195 panic("ipintr no HDR"); 196#endif 197 /* 198 * If no IP addresses have been set yet but the interfaces 199 * are receiving, can't do anything with incoming packets yet. 200 */ 201 if (in_ifaddr.tqh_first == 0) 202 goto bad; 203 ipstat.ips_total++; 204 if (m->m_len < sizeof (struct ip) && 205 (m = m_pullup(m, sizeof (struct ip))) == 0) { 206 ipstat.ips_toosmall++; 207 goto next; 208 } 209 ip = mtod(m, struct ip *); 210 if (ip->ip_v != IPVERSION) { 211 ipstat.ips_badvers++; 212 goto bad; 213 } 214 hlen = ip->ip_hl << 2; 215 if (hlen < sizeof(struct ip)) { /* minimum header length */ 216 ipstat.ips_badhlen++; 217 goto bad; 218 } 219 if (hlen > m->m_len) { 220 if ((m = m_pullup(m, hlen)) == 0) { 221 ipstat.ips_badhlen++; 222 goto next; 223 } 224 ip = mtod(m, struct ip *); 225 } 226 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 227 ipstat.ips_badsum++; 228 goto bad; 229 } 230 231 /* 232 * Convert fields to host representation. 233 */ 234 NTOHS(ip->ip_len); 235 NTOHS(ip->ip_id); 236 NTOHS(ip->ip_off); 237 len = ip->ip_len; 238 239 /* 240 * Check that the amount of data in the buffers 241 * is as at least much as the IP header would have us expect. 242 * Trim mbufs if longer than we expect. 243 * Drop packet if shorter than we expect. 244 */ 245 if (m->m_pkthdr.len < len) { 246 ipstat.ips_tooshort++; 247 goto bad; 248 } 249 if (m->m_pkthdr.len > len) { 250 if (m->m_len == m->m_pkthdr.len) { 251 m->m_len = len; 252 m->m_pkthdr.len = len; 253 } else 254 m_adj(m, len - m->m_pkthdr.len); 255 } 256 257#ifdef PFIL_HOOKS 258 /* 259 * Run through list of hooks for input packets. 260 */ 261 m0 = m; 262 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 263 if (pfh->pfil_func) { 264 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 265 if (rv) 266 goto next; 267 ip = mtod(m = m0, struct ip *); 268 } 269#endif /* PFIL_HOOKS */ 270 271 /* 272 * Process options and, if not destined for us, 273 * ship it on. ip_dooptions returns 1 when an 274 * error was detected (causing an icmp message 275 * to be sent and the original packet to be freed). 276 */ 277 ip_nhops = 0; /* for source routed packets */ 278 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 279 goto next; 280 281 /* 282 * Check our list of addresses, to see if the packet is for us. 283 */ 284 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 285 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 286 goto ours; 287 if (((ip_directedbcast == 0) || (ip_directedbcast && 288 ia->ia_ifp == m->m_pkthdr.rcvif)) && 289 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 290 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 291 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 292 /* 293 * Look for all-0's host part (old broadcast addr), 294 * either for subnet or net. 295 */ 296 ip->ip_dst.s_addr == ia->ia_subnet || 297 ip->ip_dst.s_addr == ia->ia_net) 298 goto ours; 299 } 300 /* 301 * An interface with IP address zero accepts 302 * all packets that arrive on that interface. 303 */ 304 if ((ia->ia_ifp == m->m_pkthdr.rcvif) && 305 in_nullhost(ia->ia_addr.sin_addr)) 306 goto ours; 307 } 308 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 309 struct in_multi *inm; 310#ifdef MROUTING 311 extern struct socket *ip_mrouter; 312 313 if (m->m_flags & M_EXT) { 314 if ((m = m_pullup(m, hlen)) == 0) { 315 ipstat.ips_toosmall++; 316 goto next; 317 } 318 ip = mtod(m, struct ip *); 319 } 320 321 if (ip_mrouter) { 322 /* 323 * If we are acting as a multicast router, all 324 * incoming multicast packets are passed to the 325 * kernel-level multicast forwarding function. 326 * The packet is returned (relatively) intact; if 327 * ip_mforward() returns a non-zero value, the packet 328 * must be discarded, else it may be accepted below. 329 * 330 * (The IP ident field is put in the same byte order 331 * as expected when ip_mforward() is called from 332 * ip_output().) 333 */ 334 ip->ip_id = htons(ip->ip_id); 335 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 336 ipstat.ips_cantforward++; 337 m_freem(m); 338 goto next; 339 } 340 ip->ip_id = ntohs(ip->ip_id); 341 342 /* 343 * The process-level routing demon needs to receive 344 * all multicast IGMP packets, whether or not this 345 * host belongs to their destination groups. 346 */ 347 if (ip->ip_p == IPPROTO_IGMP) 348 goto ours; 349 ipstat.ips_forward++; 350 } 351#endif 352 /* 353 * See if we belong to the destination multicast group on the 354 * arrival interface. 355 */ 356 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 357 if (inm == NULL) { 358 ipstat.ips_cantforward++; 359 m_freem(m); 360 goto next; 361 } 362 goto ours; 363 } 364 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 365 in_nullhost(ip->ip_dst)) 366 goto ours; 367 368 /* 369 * Not for us; forward if possible and desirable. 370 */ 371 if (ipforwarding == 0) { 372 ipstat.ips_cantforward++; 373 m_freem(m); 374 } else 375 ip_forward(m, 0); 376 goto next; 377 378ours: 379 /* 380 * If offset or IP_MF are set, must reassemble. 381 * Otherwise, nothing need be done. 382 * (We could look in the reassembly queue to see 383 * if the packet was previously fragmented, 384 * but it's not worth the time; just let them time out.) 385 */ 386 if (ip->ip_off & ~(IP_DF|IP_RF)) { 387 /* 388 * Look for queue of fragments 389 * of this datagram. 390 */ 391 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 392 if (ip->ip_id == fp->ipq_id && 393 in_hosteq(ip->ip_src, fp->ipq_src) && 394 in_hosteq(ip->ip_dst, fp->ipq_dst) && 395 ip->ip_p == fp->ipq_p) 396 goto found; 397 fp = 0; 398found: 399 400 /* 401 * Adjust ip_len to not reflect header, 402 * set ipqe_mff if more fragments are expected, 403 * convert offset of this to bytes. 404 */ 405 ip->ip_len -= hlen; 406 mff = (ip->ip_off & IP_MF) != 0; 407 if (mff) { 408 /* 409 * Make sure that fragments have a data length 410 * that's a non-zero multiple of 8 bytes. 411 */ 412 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 413 ipstat.ips_badfrags++; 414 goto bad; 415 } 416 } 417 ip->ip_off <<= 3; 418 419 /* 420 * If datagram marked as having more fragments 421 * or if this is not the first fragment, 422 * attempt reassembly; if it succeeds, proceed. 423 */ 424 if (mff || ip->ip_off) { 425 ipstat.ips_fragments++; 426 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 427 M_IPQ, M_NOWAIT); 428 if (ipqe == NULL) { 429 ipstat.ips_rcvmemdrop++; 430 goto bad; 431 } 432 ipqe->ipqe_mff = mff; 433 ipqe->ipqe_m = m; 434 ipqe->ipqe_ip = ip; 435 m = ip_reass(ipqe, fp); 436 if (m == 0) 437 goto next; 438 ipstat.ips_reassembled++; 439 ip = mtod(m, struct ip *); 440 } else 441 if (fp) 442 ip_freef(fp); 443 } else 444 ip->ip_len -= hlen; 445 446 /* 447 * Switch out to protocol's input routine. 448 */ 449 ipstat.ips_delivered++; 450 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 451 goto next; 452bad: 453 m_freem(m); 454 goto next; 455} 456 457/* 458 * Take incoming datagram fragment and try to 459 * reassemble it into whole datagram. If a chain for 460 * reassembly of this datagram already exists, then it 461 * is given as fp; otherwise have to make a chain. 462 */ 463struct mbuf * 464ip_reass(ipqe, fp) 465 register struct ipqent *ipqe; 466 register struct ipq *fp; 467{ 468 register struct mbuf *m = ipqe->ipqe_m; 469 register struct ipqent *nq, *p, *q; 470 struct ip *ip; 471 struct mbuf *t; 472 int hlen = ipqe->ipqe_ip->ip_hl << 2; 473 int i, next; 474 475 /* 476 * Presence of header sizes in mbufs 477 * would confuse code below. 478 */ 479 m->m_data += hlen; 480 m->m_len -= hlen; 481 482 /* 483 * If first fragment to arrive, create a reassembly queue. 484 */ 485 if (fp == 0) { 486 MALLOC(fp, struct ipq *, sizeof (struct ipq), 487 M_FTABLE, M_NOWAIT); 488 if (fp == NULL) 489 goto dropfrag; 490 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 491 fp->ipq_ttl = IPFRAGTTL; 492 fp->ipq_p = ipqe->ipqe_ip->ip_p; 493 fp->ipq_id = ipqe->ipqe_ip->ip_id; 494 LIST_INIT(&fp->ipq_fragq); 495 fp->ipq_src = ipqe->ipqe_ip->ip_src; 496 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 497 p = NULL; 498 goto insert; 499 } 500 501 /* 502 * Find a segment which begins after this one does. 503 */ 504 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 505 p = q, q = q->ipqe_q.le_next) 506 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 507 break; 508 509 /* 510 * If there is a preceding segment, it may provide some of 511 * our data already. If so, drop the data from the incoming 512 * segment. If it provides all of our data, drop us. 513 */ 514 if (p != NULL) { 515 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 516 ipqe->ipqe_ip->ip_off; 517 if (i > 0) { 518 if (i >= ipqe->ipqe_ip->ip_len) 519 goto dropfrag; 520 m_adj(ipqe->ipqe_m, i); 521 ipqe->ipqe_ip->ip_off += i; 522 ipqe->ipqe_ip->ip_len -= i; 523 } 524 } 525 526 /* 527 * While we overlap succeeding segments trim them or, 528 * if they are completely covered, dequeue them. 529 */ 530 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 531 q->ipqe_ip->ip_off; q = nq) { 532 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 533 q->ipqe_ip->ip_off; 534 if (i < q->ipqe_ip->ip_len) { 535 q->ipqe_ip->ip_len -= i; 536 q->ipqe_ip->ip_off += i; 537 m_adj(q->ipqe_m, i); 538 break; 539 } 540 nq = q->ipqe_q.le_next; 541 m_freem(q->ipqe_m); 542 LIST_REMOVE(q, ipqe_q); 543 FREE(q, M_IPQ); 544 } 545 546insert: 547 /* 548 * Stick new segment in its place; 549 * check for complete reassembly. 550 */ 551 if (p == NULL) { 552 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 553 } else { 554 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 555 } 556 next = 0; 557 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 558 p = q, q = q->ipqe_q.le_next) { 559 if (q->ipqe_ip->ip_off != next) 560 return (0); 561 next += q->ipqe_ip->ip_len; 562 } 563 if (p->ipqe_mff) 564 return (0); 565 566 /* 567 * Reassembly is complete. Check for a bogus message size and 568 * concatenate fragments. 569 */ 570 q = fp->ipq_fragq.lh_first; 571 ip = q->ipqe_ip; 572 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 573 ipstat.ips_toolong++; 574 ip_freef(fp); 575 return (0); 576 } 577 m = q->ipqe_m; 578 t = m->m_next; 579 m->m_next = 0; 580 m_cat(m, t); 581 nq = q->ipqe_q.le_next; 582 FREE(q, M_IPQ); 583 for (q = nq; q != NULL; q = nq) { 584 t = q->ipqe_m; 585 nq = q->ipqe_q.le_next; 586 FREE(q, M_IPQ); 587 m_cat(m, t); 588 } 589 590 /* 591 * Create header for new ip packet by 592 * modifying header of first packet; 593 * dequeue and discard fragment reassembly header. 594 * Make header visible. 595 */ 596 ip->ip_len = next; 597 ip->ip_src = fp->ipq_src; 598 ip->ip_dst = fp->ipq_dst; 599 LIST_REMOVE(fp, ipq_q); 600 FREE(fp, M_FTABLE); 601 m->m_len += (ip->ip_hl << 2); 602 m->m_data -= (ip->ip_hl << 2); 603 /* some debugging cruft by sklower, below, will go away soon */ 604 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 605 register int plen = 0; 606 for (t = m; t; t = t->m_next) 607 plen += t->m_len; 608 m->m_pkthdr.len = plen; 609 } 610 return (m); 611 612dropfrag: 613 ipstat.ips_fragdropped++; 614 m_freem(m); 615 FREE(ipqe, M_IPQ); 616 return (0); 617} 618 619/* 620 * Free a fragment reassembly header and all 621 * associated datagrams. 622 */ 623void 624ip_freef(fp) 625 struct ipq *fp; 626{ 627 register struct ipqent *q, *p; 628 629 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 630 p = q->ipqe_q.le_next; 631 m_freem(q->ipqe_m); 632 LIST_REMOVE(q, ipqe_q); 633 FREE(q, M_IPQ); 634 } 635 LIST_REMOVE(fp, ipq_q); 636 FREE(fp, M_FTABLE); 637} 638 639/* 640 * IP timer processing; 641 * if a timer expires on a reassembly 642 * queue, discard it. 643 */ 644void 645ip_slowtimo() 646{ 647 register struct ipq *fp, *nfp; 648 int s = splsoftnet(); 649 650 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 651 nfp = fp->ipq_q.le_next; 652 if (--fp->ipq_ttl == 0) { 653 ipstat.ips_fragtimeout++; 654 ip_freef(fp); 655 } 656 } 657 splx(s); 658} 659 660/* 661 * Drain off all datagram fragments. 662 */ 663void 664ip_drain() 665{ 666 667 while (ipq.lh_first != NULL) { 668 ipstat.ips_fragdropped++; 669 ip_freef(ipq.lh_first); 670 } 671} 672 673/* 674 * Do option processing on a datagram, 675 * possibly discarding it if bad options are encountered, 676 * or forwarding it if source-routed. 677 * Returns 1 if packet has been forwarded/freed, 678 * 0 if the packet should be processed further. 679 */ 680int 681ip_dooptions(m) 682 struct mbuf *m; 683{ 684 register struct ip *ip = mtod(m, struct ip *); 685 register u_char *cp; 686 register struct ip_timestamp *ipt; 687 register struct in_ifaddr *ia; 688 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 689 struct in_addr *sin, dst; 690 n_time ntime; 691 692 dst = ip->ip_dst; 693 cp = (u_char *)(ip + 1); 694 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 695 for (; cnt > 0; cnt -= optlen, cp += optlen) { 696 opt = cp[IPOPT_OPTVAL]; 697 if (opt == IPOPT_EOL) 698 break; 699 if (opt == IPOPT_NOP) 700 optlen = 1; 701 else { 702 optlen = cp[IPOPT_OLEN]; 703 if (optlen <= 0 || optlen > cnt) { 704 code = &cp[IPOPT_OLEN] - (u_char *)ip; 705 goto bad; 706 } 707 } 708 switch (opt) { 709 710 default: 711 break; 712 713 /* 714 * Source routing with record. 715 * Find interface with current destination address. 716 * If none on this machine then drop if strictly routed, 717 * or do nothing if loosely routed. 718 * Record interface address and bring up next address 719 * component. If strictly routed make sure next 720 * address is on directly accessible net. 721 */ 722 case IPOPT_LSRR: 723 case IPOPT_SSRR: 724 if (ip_allowsrcrt == 0) { 725 type = ICMP_UNREACH; 726 code = ICMP_UNREACH_NET_PROHIB; 727 goto bad; 728 } 729 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 730 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 731 goto bad; 732 } 733 ipaddr.sin_addr = ip->ip_dst; 734 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 735 if (ia == 0) { 736 if (opt == IPOPT_SSRR) { 737 type = ICMP_UNREACH; 738 code = ICMP_UNREACH_SRCFAIL; 739 goto bad; 740 } 741 /* 742 * Loose routing, and not at next destination 743 * yet; nothing to do except forward. 744 */ 745 break; 746 } 747 off--; /* 0 origin */ 748 if (off > optlen - sizeof(struct in_addr)) { 749 /* 750 * End of source route. Should be for us. 751 */ 752 save_rte(cp, ip->ip_src); 753 break; 754 } 755 /* 756 * locate outgoing interface 757 */ 758 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 759 sizeof(ipaddr.sin_addr)); 760 if (opt == IPOPT_SSRR) { 761#define INA struct in_ifaddr * 762#define SA struct sockaddr * 763 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 764 } else 765 ia = ip_rtaddr(ipaddr.sin_addr); 766 if (ia == 0) { 767 type = ICMP_UNREACH; 768 code = ICMP_UNREACH_SRCFAIL; 769 goto bad; 770 } 771 ip->ip_dst = ipaddr.sin_addr; 772 bcopy((caddr_t)&ia->ia_addr.sin_addr, 773 (caddr_t)(cp + off), sizeof(struct in_addr)); 774 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 775 /* 776 * Let ip_intr's mcast routing check handle mcast pkts 777 */ 778 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 779 break; 780 781 case IPOPT_RR: 782 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 783 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 784 goto bad; 785 } 786 /* 787 * If no space remains, ignore. 788 */ 789 off--; /* 0 origin */ 790 if (off > optlen - sizeof(struct in_addr)) 791 break; 792 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 793 sizeof(ipaddr.sin_addr)); 794 /* 795 * locate outgoing interface; if we're the destination, 796 * use the incoming interface (should be same). 797 */ 798 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 799 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 800 type = ICMP_UNREACH; 801 code = ICMP_UNREACH_HOST; 802 goto bad; 803 } 804 bcopy((caddr_t)&ia->ia_addr.sin_addr, 805 (caddr_t)(cp + off), sizeof(struct in_addr)); 806 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 807 break; 808 809 case IPOPT_TS: 810 code = cp - (u_char *)ip; 811 ipt = (struct ip_timestamp *)cp; 812 if (ipt->ipt_len < 5) 813 goto bad; 814 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 815 if (++ipt->ipt_oflw == 0) 816 goto bad; 817 break; 818 } 819 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 820 switch (ipt->ipt_flg) { 821 822 case IPOPT_TS_TSONLY: 823 break; 824 825 case IPOPT_TS_TSANDADDR: 826 if (ipt->ipt_ptr + sizeof(n_time) + 827 sizeof(struct in_addr) > ipt->ipt_len) 828 goto bad; 829 ipaddr.sin_addr = dst; 830 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 831 m->m_pkthdr.rcvif); 832 if (ia == 0) 833 continue; 834 bcopy((caddr_t)&ia->ia_addr.sin_addr, 835 (caddr_t)sin, sizeof(struct in_addr)); 836 ipt->ipt_ptr += sizeof(struct in_addr); 837 break; 838 839 case IPOPT_TS_PRESPEC: 840 if (ipt->ipt_ptr + sizeof(n_time) + 841 sizeof(struct in_addr) > ipt->ipt_len) 842 goto bad; 843 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 844 sizeof(struct in_addr)); 845 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 846 continue; 847 ipt->ipt_ptr += sizeof(struct in_addr); 848 break; 849 850 default: 851 goto bad; 852 } 853 ntime = iptime(); 854 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 855 sizeof(n_time)); 856 ipt->ipt_ptr += sizeof(n_time); 857 } 858 } 859 if (forward) { 860 if (ip_forwsrcrt == 0) { 861 type = ICMP_UNREACH; 862 code = ICMP_UNREACH_SRCFAIL; 863 goto bad; 864 } 865 ip_forward(m, 1); 866 return (1); 867 } 868 return (0); 869bad: 870 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 871 icmp_error(m, type, code, 0, 0); 872 ipstat.ips_badoptions++; 873 return (1); 874} 875 876/* 877 * Given address of next destination (final or next hop), 878 * return internet address info of interface to be used to get there. 879 */ 880struct in_ifaddr * 881ip_rtaddr(dst) 882 struct in_addr dst; 883{ 884 register struct sockaddr_in *sin; 885 886 sin = satosin(&ipforward_rt.ro_dst); 887 888 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 889 if (ipforward_rt.ro_rt) { 890 RTFREE(ipforward_rt.ro_rt); 891 ipforward_rt.ro_rt = 0; 892 } 893 sin->sin_family = AF_INET; 894 sin->sin_len = sizeof(*sin); 895 sin->sin_addr = dst; 896 897 rtalloc(&ipforward_rt); 898 } 899 if (ipforward_rt.ro_rt == 0) 900 return ((struct in_ifaddr *)0); 901 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 902} 903 904/* 905 * Save incoming source route for use in replies, 906 * to be picked up later by ip_srcroute if the receiver is interested. 907 */ 908void 909save_rte(option, dst) 910 u_char *option; 911 struct in_addr dst; 912{ 913 unsigned olen; 914 915 olen = option[IPOPT_OLEN]; 916#ifdef DIAGNOSTIC 917 if (ipprintfs) 918 printf("save_rte: olen %d\n", olen); 919#endif 920 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 921 return; 922 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 923 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 924 ip_srcrt.dst = dst; 925} 926 927/* 928 * Retrieve incoming source route for use in replies, 929 * in the same form used by setsockopt. 930 * The first hop is placed before the options, will be removed later. 931 */ 932struct mbuf * 933ip_srcroute() 934{ 935 register struct in_addr *p, *q; 936 register struct mbuf *m; 937 938 if (ip_nhops == 0) 939 return ((struct mbuf *)0); 940 m = m_get(M_DONTWAIT, MT_SOOPTS); 941 if (m == 0) 942 return ((struct mbuf *)0); 943 944#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 945 946 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 947 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 948 OPTSIZ; 949#ifdef DIAGNOSTIC 950 if (ipprintfs) 951 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 952#endif 953 954 /* 955 * First save first hop for return route 956 */ 957 p = &ip_srcrt.route[ip_nhops - 1]; 958 *(mtod(m, struct in_addr *)) = *p--; 959#ifdef DIAGNOSTIC 960 if (ipprintfs) 961 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 962#endif 963 964 /* 965 * Copy option fields and padding (nop) to mbuf. 966 */ 967 ip_srcrt.nop = IPOPT_NOP; 968 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 969 bcopy((caddr_t)&ip_srcrt.nop, 970 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 971 q = (struct in_addr *)(mtod(m, caddr_t) + 972 sizeof(struct in_addr) + OPTSIZ); 973#undef OPTSIZ 974 /* 975 * Record return path as an IP source route, 976 * reversing the path (pointers are now aligned). 977 */ 978 while (p >= ip_srcrt.route) { 979#ifdef DIAGNOSTIC 980 if (ipprintfs) 981 printf(" %x", ntohl(q->s_addr)); 982#endif 983 *q++ = *p--; 984 } 985 /* 986 * Last hop goes to final destination. 987 */ 988 *q = ip_srcrt.dst; 989#ifdef DIAGNOSTIC 990 if (ipprintfs) 991 printf(" %x\n", ntohl(q->s_addr)); 992#endif 993 return (m); 994} 995 996/* 997 * Strip out IP options, at higher 998 * level protocol in the kernel. 999 * Second argument is buffer to which options 1000 * will be moved, and return value is their length. 1001 * XXX should be deleted; last arg currently ignored. 1002 */ 1003void 1004ip_stripoptions(m, mopt) 1005 register struct mbuf *m; 1006 struct mbuf *mopt; 1007{ 1008 register int i; 1009 struct ip *ip = mtod(m, struct ip *); 1010 register caddr_t opts; 1011 int olen; 1012 1013 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1014 opts = (caddr_t)(ip + 1); 1015 i = m->m_len - (sizeof (struct ip) + olen); 1016 bcopy(opts + olen, opts, (unsigned)i); 1017 m->m_len -= olen; 1018 if (m->m_flags & M_PKTHDR) 1019 m->m_pkthdr.len -= olen; 1020 ip->ip_hl = sizeof(struct ip) >> 2; 1021} 1022 1023int inetctlerrmap[PRC_NCMDS] = { 1024 0, 0, 0, 0, 1025 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1026 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1027 EMSGSIZE, EHOSTUNREACH, 0, 0, 1028 0, 0, 0, 0, 1029 ENOPROTOOPT 1030}; 1031 1032/* 1033 * Forward a packet. If some error occurs return the sender 1034 * an icmp packet. Note we can't always generate a meaningful 1035 * icmp message because icmp doesn't have a large enough repertoire 1036 * of codes and types. 1037 * 1038 * If not forwarding, just drop the packet. This could be confusing 1039 * if ipforwarding was zero but some routing protocol was advancing 1040 * us as a gateway to somewhere. However, we must let the routing 1041 * protocol deal with that. 1042 * 1043 * The srcrt parameter indicates whether the packet is being forwarded 1044 * via a source route. 1045 */ 1046void 1047ip_forward(m, srcrt) 1048 struct mbuf *m; 1049 int srcrt; 1050{ 1051 register struct ip *ip = mtod(m, struct ip *); 1052 register struct sockaddr_in *sin; 1053 register struct rtentry *rt; 1054 int error, type = 0, code = 0; 1055 struct mbuf *mcopy; 1056 n_long dest; 1057 struct ifnet *destifp; 1058 1059 dest = 0; 1060#ifdef DIAGNOSTIC 1061 if (ipprintfs) 1062 printf("forward: src %x dst %x ttl %x\n", 1063 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1064#endif 1065 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1066 ipstat.ips_cantforward++; 1067 m_freem(m); 1068 return; 1069 } 1070 HTONS(ip->ip_id); 1071 if (ip->ip_ttl <= IPTTLDEC) { 1072 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1073 return; 1074 } 1075 ip->ip_ttl -= IPTTLDEC; 1076 1077 sin = satosin(&ipforward_rt.ro_dst); 1078 if ((rt = ipforward_rt.ro_rt) == 0 || 1079 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1080 if (ipforward_rt.ro_rt) { 1081 RTFREE(ipforward_rt.ro_rt); 1082 ipforward_rt.ro_rt = 0; 1083 } 1084 sin->sin_family = AF_INET; 1085 sin->sin_len = sizeof(struct sockaddr_in); 1086 sin->sin_addr = ip->ip_dst; 1087 1088 rtalloc(&ipforward_rt); 1089 if (ipforward_rt.ro_rt == 0) { 1090 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1091 return; 1092 } 1093 rt = ipforward_rt.ro_rt; 1094 } 1095 1096 /* 1097 * Save at most 68 bytes of the packet in case 1098 * we need to generate an ICMP message to the src. 1099 */ 1100 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1101 1102 /* 1103 * If forwarding packet using same interface that it came in on, 1104 * perhaps should send a redirect to sender to shortcut a hop. 1105 * Only send redirect if source is sending directly to us, 1106 * and if packet was not source routed (or has any options). 1107 * Also, don't send redirect if forwarding using a default route 1108 * or a route modified by a redirect. 1109 */ 1110 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1111 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1112 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1113 ipsendredirects && !srcrt) { 1114 if (rt->rt_ifa && 1115 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1116 ifatoia(rt->rt_ifa)->ia_subnet) { 1117 if (rt->rt_flags & RTF_GATEWAY) 1118 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1119 else 1120 dest = ip->ip_dst.s_addr; 1121 /* Router requirements says to only send host redirects */ 1122 type = ICMP_REDIRECT; 1123 code = ICMP_REDIRECT_HOST; 1124#ifdef DIAGNOSTIC 1125 if (ipprintfs) 1126 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1127#endif 1128 } 1129 } 1130 1131 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1132 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1133 if (error) 1134 ipstat.ips_cantforward++; 1135 else { 1136 ipstat.ips_forward++; 1137 if (type) 1138 ipstat.ips_redirectsent++; 1139 else { 1140 if (mcopy) 1141 m_freem(mcopy); 1142 return; 1143 } 1144 } 1145 if (mcopy == NULL) 1146 return; 1147 destifp = NULL; 1148 1149 switch (error) { 1150 1151 case 0: /* forwarded, but need redirect */ 1152 /* type, code set above */ 1153 break; 1154 1155 case ENETUNREACH: /* shouldn't happen, checked above */ 1156 case EHOSTUNREACH: 1157 case ENETDOWN: 1158 case EHOSTDOWN: 1159 default: 1160 type = ICMP_UNREACH; 1161 code = ICMP_UNREACH_HOST; 1162 break; 1163 1164 case EMSGSIZE: 1165 type = ICMP_UNREACH; 1166 code = ICMP_UNREACH_NEEDFRAG; 1167 if (ipforward_rt.ro_rt) 1168 destifp = ipforward_rt.ro_rt->rt_ifp; 1169 ipstat.ips_cantfrag++; 1170 break; 1171 1172 case ENOBUFS: 1173 type = ICMP_SOURCEQUENCH; 1174 code = 0; 1175 break; 1176 } 1177 icmp_error(mcopy, type, code, dest, destifp); 1178} 1179 1180void 1181ip_savecontrol(inp, mp, ip, m) 1182 register struct inpcb *inp; 1183 register struct mbuf **mp; 1184 register struct ip *ip; 1185 register struct mbuf *m; 1186{ 1187 1188 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1189 struct timeval tv; 1190 1191 microtime(&tv); 1192 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1193 SCM_TIMESTAMP, SOL_SOCKET); 1194 if (*mp) 1195 mp = &(*mp)->m_next; 1196 } 1197 if (inp->inp_flags & INP_RECVDSTADDR) { 1198 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1199 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1200 if (*mp) 1201 mp = &(*mp)->m_next; 1202 } 1203#ifdef notyet 1204 /* 1205 * XXX 1206 * Moving these out of udp_input() made them even more broken 1207 * than they already were. 1208 * - fenner@parc.xerox.com 1209 */ 1210 /* options were tossed already */ 1211 if (inp->inp_flags & INP_RECVOPTS) { 1212 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1213 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1214 if (*mp) 1215 mp = &(*mp)->m_next; 1216 } 1217 /* ip_srcroute doesn't do what we want here, need to fix */ 1218 if (inp->inp_flags & INP_RECVRETOPTS) { 1219 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1220 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1221 if (*mp) 1222 mp = &(*mp)->m_next; 1223 } 1224#endif 1225 if (inp->inp_flags & INP_RECVIF) { 1226 struct sockaddr_dl sdl; 1227 1228 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1229 sdl.sdl_family = AF_LINK; 1230 sdl.sdl_index = m->m_pkthdr.rcvif ? 1231 m->m_pkthdr.rcvif->if_index : 0; 1232 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1233 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1234 IP_RECVIF, IPPROTO_IP); 1235 if (*mp) 1236 mp = &(*mp)->m_next; 1237 } 1238} 1239 1240int 1241ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1242 int *name; 1243 u_int namelen; 1244 void *oldp; 1245 size_t *oldlenp; 1246 void *newp; 1247 size_t newlen; 1248{ 1249 extern int subnetsarelocal; 1250 1251 int error, old; 1252 1253 /* All sysctl names at this level are terminal. */ 1254 if (namelen != 1) 1255 return (ENOTDIR); 1256 1257 switch (name[0]) { 1258 case IPCTL_FORWARDING: 1259 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1260 case IPCTL_SENDREDIRECTS: 1261 return (sysctl_int(oldp, oldlenp, newp, newlen, 1262 &ipsendredirects)); 1263 case IPCTL_DEFTTL: 1264 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1265#ifdef notyet 1266 case IPCTL_DEFMTU: 1267 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1268#endif 1269 case IPCTL_FORWSRCRT: 1270 /* Don't allow this to change in a secure environment. */ 1271 if (securelevel > 0) 1272 return (sysctl_rdint(oldp, oldlenp, newp, 1273 ip_forwsrcrt)); 1274 else 1275 return (sysctl_int(oldp, oldlenp, newp, newlen, 1276 &ip_forwsrcrt)); 1277 case IPCTL_DIRECTEDBCAST: 1278 return (sysctl_int(oldp, oldlenp, newp, newlen, 1279 &ip_directedbcast)); 1280 case IPCTL_ALLOWSRCRT: 1281 return (sysctl_int(oldp, oldlenp, newp, newlen, 1282 &ip_allowsrcrt)); 1283 case IPCTL_SUBNETSARELOCAL: 1284 return (sysctl_int(oldp, oldlenp, newp, newlen, 1285 &subnetsarelocal)); 1286 case IPCTL_MTUDISC: 1287 return (sysctl_int(oldp, oldlenp, newp, newlen, 1288 &ip_mtudisc)); 1289 case IPCTL_ANONPORTMIN: 1290 old = anonportmin; 1291 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1292 if (anonportmin >= anonportmax || anonportmin > 65535 1293#ifndef IPNOPRIVPORTS 1294 || anonportmin < IPPORT_RESERVED 1295#endif 1296 ) { 1297 anonportmin = old; 1298 return (EINVAL); 1299 } 1300 return (error); 1301 case IPCTL_ANONPORTMAX: 1302 old = anonportmax; 1303 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1304 if (anonportmin >= anonportmax || anonportmax > 65535 1305#ifndef IPNOPRIVPORTS 1306 || anonportmax < IPPORT_RESERVED 1307#endif 1308 ) { 1309 anonportmax = old; 1310 return (EINVAL); 1311 } 1312 return (error); 1313 default: 1314 return (EOPNOTSUPP); 1315 } 1316 /* NOTREACHED */ 1317} 1318