ip_input.c revision 1.55
1/* $NetBSD: ip_input.c,v 1.55 1998/01/12 03:02:51 scottr 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/* XXX should really put this in libkern.h */ 70#define offsetof(type, member) ((size_t)(&((type *)0)->member)) 71 72#ifndef IPFORWARDING 73#ifdef GATEWAY 74#define IPFORWARDING 1 /* forward IP packets not for us */ 75#else /* GATEWAY */ 76#define IPFORWARDING 0 /* don't forward IP packets not for us */ 77#endif /* GATEWAY */ 78#endif /* IPFORWARDING */ 79#ifndef IPSENDREDIRECTS 80#define IPSENDREDIRECTS 1 81#endif 82#ifndef IPFORWSRCRT 83#define IPFORWSRCRT 1 /* forward source-routed packets */ 84#endif 85#ifndef IPALLOWSRCRT 86#define IPALLOWSRCRT 1 /* allow source-routed packets */ 87#endif 88#ifndef IPMTUDISC 89#define IPMTUDISC 0 90#endif 91 92/* 93 * Note: DIRECTED_BROADCAST is handled this way so that previous 94 * configuration using this option will Just Work. 95 */ 96#ifndef IPDIRECTEDBCAST 97#ifdef DIRECTED_BROADCAST 98#define IPDIRECTEDBCAST 1 99#else 100#define IPDIRECTEDBCAST 0 101#endif /* DIRECTED_BROADCAST */ 102#endif /* IPDIRECTEDBCAST */ 103int ipforwarding = IPFORWARDING; 104int ipsendredirects = IPSENDREDIRECTS; 105int ip_defttl = IPDEFTTL; 106int ip_forwsrcrt = IPFORWSRCRT; 107int ip_directedbcast = IPDIRECTEDBCAST; 108int ip_allowsrcrt = IPALLOWSRCRT; 109int ip_mtudisc = IPMTUDISC; 110#ifdef DIAGNOSTIC 111int ipprintfs = 0; 112#endif 113 114extern struct domain inetdomain; 115extern struct protosw inetsw[]; 116u_char ip_protox[IPPROTO_MAX]; 117int ipqmaxlen = IFQ_MAXLEN; 118struct in_ifaddrhead in_ifaddr; 119struct ifqueue ipintrq; 120 121/* 122 * We need to save the IP options in case a protocol wants to respond 123 * to an incoming packet over the same route if the packet got here 124 * using IP source routing. This allows connection establishment and 125 * maintenance when the remote end is on a network that is not known 126 * to us. 127 */ 128int ip_nhops = 0; 129static struct ip_srcrt { 130 struct in_addr dst; /* final destination */ 131 char nop; /* one NOP to align */ 132 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 133 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 134} ip_srcrt; 135 136static void save_rte __P((u_char *, struct in_addr)); 137 138/* 139 * IP initialization: fill in IP protocol switch table. 140 * All protocols not implemented in kernel go to raw IP protocol handler. 141 */ 142void 143ip_init() 144{ 145 register struct protosw *pr; 146 register int i; 147 148 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 149 if (pr == 0) 150 panic("ip_init"); 151 for (i = 0; i < IPPROTO_MAX; i++) 152 ip_protox[i] = pr - inetsw; 153 for (pr = inetdomain.dom_protosw; 154 pr < inetdomain.dom_protoswNPROTOSW; pr++) 155 if (pr->pr_domain->dom_family == PF_INET && 156 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 157 ip_protox[pr->pr_protocol] = pr - inetsw; 158 LIST_INIT(&ipq); 159 ip_id = time.tv_sec & 0xffff; 160 ipintrq.ifq_maxlen = ipqmaxlen; 161 TAILQ_INIT(&in_ifaddr); 162} 163 164struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 165struct route ipforward_rt; 166 167/* 168 * Ip input routine. Checksum and byte swap header. If fragmented 169 * try to reassemble. Process options. Pass to next level. 170 */ 171void 172ipintr() 173{ 174 register struct ip *ip = NULL; 175 register struct mbuf *m; 176 register struct ipq *fp; 177 register struct in_ifaddr *ia; 178 struct ipqent *ipqe; 179 int hlen = 0, mff, len, s; 180#ifdef PFIL_HOOKS 181 struct packet_filter_hook *pfh; 182 struct mbuf *m0; 183 int rv; 184#endif /* PFIL_HOOKS */ 185 186next: 187 /* 188 * Get next datagram off input queue and get IP header 189 * in first mbuf. 190 */ 191 s = splimp(); 192 IF_DEQUEUE(&ipintrq, m); 193 splx(s); 194 if (m == 0) 195 return; 196#ifdef DIAGNOSTIC 197 if ((m->m_flags & M_PKTHDR) == 0) 198 panic("ipintr no HDR"); 199#endif 200 /* 201 * If no IP addresses have been set yet but the interfaces 202 * are receiving, can't do anything with incoming packets yet. 203 */ 204 if (in_ifaddr.tqh_first == 0) 205 goto bad; 206 ipstat.ips_total++; 207 if (m->m_len < sizeof (struct ip) && 208 (m = m_pullup(m, sizeof (struct ip))) == 0) { 209 ipstat.ips_toosmall++; 210 goto next; 211 } 212 ip = mtod(m, struct ip *); 213 if (ip->ip_v != IPVERSION) { 214 ipstat.ips_badvers++; 215 goto bad; 216 } 217 hlen = ip->ip_hl << 2; 218 if (hlen < sizeof(struct ip)) { /* minimum header length */ 219 ipstat.ips_badhlen++; 220 goto bad; 221 } 222 if (hlen > m->m_len) { 223 if ((m = m_pullup(m, hlen)) == 0) { 224 ipstat.ips_badhlen++; 225 goto next; 226 } 227 ip = mtod(m, struct ip *); 228 } 229 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 230 ipstat.ips_badsum++; 231 goto bad; 232 } 233 234 /* 235 * Convert fields to host representation. 236 */ 237 NTOHS(ip->ip_len); 238 NTOHS(ip->ip_id); 239 NTOHS(ip->ip_off); 240 len = ip->ip_len; 241 242 /* 243 * Check that the amount of data in the buffers 244 * is as at least much as the IP header would have us expect. 245 * Trim mbufs if longer than we expect. 246 * Drop packet if shorter than we expect. 247 */ 248 if (m->m_pkthdr.len < len) { 249 ipstat.ips_tooshort++; 250 goto bad; 251 } 252 if (m->m_pkthdr.len > len) { 253 if (m->m_len == m->m_pkthdr.len) { 254 m->m_len = len; 255 m->m_pkthdr.len = len; 256 } else 257 m_adj(m, len - m->m_pkthdr.len); 258 } 259 260#ifdef PFIL_HOOKS 261 /* 262 * Run through list of hooks for input packets. 263 */ 264 m0 = m; 265 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 266 if (pfh->pfil_func) { 267 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 268 if (rv) 269 goto next; 270 ip = mtod(m = m0, struct ip *); 271 } 272#endif /* PFIL_HOOKS */ 273 274 /* 275 * Process options and, if not destined for us, 276 * ship it on. ip_dooptions returns 1 when an 277 * error was detected (causing an icmp message 278 * to be sent and the original packet to be freed). 279 */ 280 ip_nhops = 0; /* for source routed packets */ 281 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 282 goto next; 283 284 /* 285 * Check our list of addresses, to see if the packet is for us. 286 */ 287 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 288 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 289 goto ours; 290 if (((ip_directedbcast == 0) || (ip_directedbcast && 291 ia->ia_ifp == m->m_pkthdr.rcvif)) && 292 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 293 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 294 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 295 /* 296 * Look for all-0's host part (old broadcast addr), 297 * either for subnet or net. 298 */ 299 ip->ip_dst.s_addr == ia->ia_subnet || 300 ip->ip_dst.s_addr == ia->ia_net) 301 goto ours; 302 } 303 /* 304 * An interface with IP address zero accepts 305 * all packets that arrive on that interface. 306 */ 307 if ((ia->ia_ifp == m->m_pkthdr.rcvif) && 308 in_nullhost(ia->ia_addr.sin_addr)) 309 goto ours; 310 } 311 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 312 struct in_multi *inm; 313#ifdef MROUTING 314 extern struct socket *ip_mrouter; 315 316 if (m->m_flags & M_EXT) { 317 if ((m = m_pullup(m, hlen)) == 0) { 318 ipstat.ips_toosmall++; 319 goto next; 320 } 321 ip = mtod(m, struct ip *); 322 } 323 324 if (ip_mrouter) { 325 /* 326 * If we are acting as a multicast router, all 327 * incoming multicast packets are passed to the 328 * kernel-level multicast forwarding function. 329 * The packet is returned (relatively) intact; if 330 * ip_mforward() returns a non-zero value, the packet 331 * must be discarded, else it may be accepted below. 332 * 333 * (The IP ident field is put in the same byte order 334 * as expected when ip_mforward() is called from 335 * ip_output().) 336 */ 337 ip->ip_id = htons(ip->ip_id); 338 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 339 ipstat.ips_cantforward++; 340 m_freem(m); 341 goto next; 342 } 343 ip->ip_id = ntohs(ip->ip_id); 344 345 /* 346 * The process-level routing demon needs to receive 347 * all multicast IGMP packets, whether or not this 348 * host belongs to their destination groups. 349 */ 350 if (ip->ip_p == IPPROTO_IGMP) 351 goto ours; 352 ipstat.ips_forward++; 353 } 354#endif 355 /* 356 * See if we belong to the destination multicast group on the 357 * arrival interface. 358 */ 359 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 360 if (inm == NULL) { 361 ipstat.ips_cantforward++; 362 m_freem(m); 363 goto next; 364 } 365 goto ours; 366 } 367 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 368 in_nullhost(ip->ip_dst)) 369 goto ours; 370 371 /* 372 * Not for us; forward if possible and desirable. 373 */ 374 if (ipforwarding == 0) { 375 ipstat.ips_cantforward++; 376 m_freem(m); 377 } else 378 ip_forward(m, 0); 379 goto next; 380 381ours: 382 /* 383 * If offset or IP_MF are set, must reassemble. 384 * Otherwise, nothing need be done. 385 * (We could look in the reassembly queue to see 386 * if the packet was previously fragmented, 387 * but it's not worth the time; just let them time out.) 388 */ 389 if (ip->ip_off & ~(IP_DF|IP_RF)) { 390 /* 391 * Look for queue of fragments 392 * of this datagram. 393 */ 394 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 395 if (ip->ip_id == fp->ipq_id && 396 in_hosteq(ip->ip_src, fp->ipq_src) && 397 in_hosteq(ip->ip_dst, fp->ipq_dst) && 398 ip->ip_p == fp->ipq_p) 399 goto found; 400 fp = 0; 401found: 402 403 /* 404 * Adjust ip_len to not reflect header, 405 * set ipqe_mff if more fragments are expected, 406 * convert offset of this to bytes. 407 */ 408 ip->ip_len -= hlen; 409 mff = (ip->ip_off & IP_MF) != 0; 410 if (mff) { 411 /* 412 * Make sure that fragments have a data length 413 * that's a non-zero multiple of 8 bytes. 414 */ 415 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 416 ipstat.ips_badfrags++; 417 goto bad; 418 } 419 } 420 ip->ip_off <<= 3; 421 422 /* 423 * If datagram marked as having more fragments 424 * or if this is not the first fragment, 425 * attempt reassembly; if it succeeds, proceed. 426 */ 427 if (mff || ip->ip_off) { 428 ipstat.ips_fragments++; 429 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 430 M_IPQ, M_NOWAIT); 431 if (ipqe == NULL) { 432 ipstat.ips_rcvmemdrop++; 433 goto bad; 434 } 435 ipqe->ipqe_mff = mff; 436 ipqe->ipqe_m = m; 437 ipqe->ipqe_ip = ip; 438 m = ip_reass(ipqe, fp); 439 if (m == 0) 440 goto next; 441 ipstat.ips_reassembled++; 442 ip = mtod(m, struct ip *); 443 } else 444 if (fp) 445 ip_freef(fp); 446 } else 447 ip->ip_len -= hlen; 448 449 /* 450 * Switch out to protocol's input routine. 451 */ 452 ipstat.ips_delivered++; 453 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 454 goto next; 455bad: 456 m_freem(m); 457 goto next; 458} 459 460/* 461 * Take incoming datagram fragment and try to 462 * reassemble it into whole datagram. If a chain for 463 * reassembly of this datagram already exists, then it 464 * is given as fp; otherwise have to make a chain. 465 */ 466struct mbuf * 467ip_reass(ipqe, fp) 468 register struct ipqent *ipqe; 469 register struct ipq *fp; 470{ 471 register struct mbuf *m = ipqe->ipqe_m; 472 register struct ipqent *nq, *p, *q; 473 struct ip *ip; 474 struct mbuf *t; 475 int hlen = ipqe->ipqe_ip->ip_hl << 2; 476 int i, next; 477 478 /* 479 * Presence of header sizes in mbufs 480 * would confuse code below. 481 */ 482 m->m_data += hlen; 483 m->m_len -= hlen; 484 485 /* 486 * If first fragment to arrive, create a reassembly queue. 487 */ 488 if (fp == 0) { 489 MALLOC(fp, struct ipq *, sizeof (struct ipq), 490 M_FTABLE, M_NOWAIT); 491 if (fp == NULL) 492 goto dropfrag; 493 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 494 fp->ipq_ttl = IPFRAGTTL; 495 fp->ipq_p = ipqe->ipqe_ip->ip_p; 496 fp->ipq_id = ipqe->ipqe_ip->ip_id; 497 LIST_INIT(&fp->ipq_fragq); 498 fp->ipq_src = ipqe->ipqe_ip->ip_src; 499 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 500 p = NULL; 501 goto insert; 502 } 503 504 /* 505 * Find a segment which begins after this one does. 506 */ 507 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 508 p = q, q = q->ipqe_q.le_next) 509 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 510 break; 511 512 /* 513 * If there is a preceding segment, it may provide some of 514 * our data already. If so, drop the data from the incoming 515 * segment. If it provides all of our data, drop us. 516 */ 517 if (p != NULL) { 518 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 519 ipqe->ipqe_ip->ip_off; 520 if (i > 0) { 521 if (i >= ipqe->ipqe_ip->ip_len) 522 goto dropfrag; 523 m_adj(ipqe->ipqe_m, i); 524 ipqe->ipqe_ip->ip_off += i; 525 ipqe->ipqe_ip->ip_len -= i; 526 } 527 } 528 529 /* 530 * While we overlap succeeding segments trim them or, 531 * if they are completely covered, dequeue them. 532 */ 533 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 534 q->ipqe_ip->ip_off; q = nq) { 535 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 536 q->ipqe_ip->ip_off; 537 if (i < q->ipqe_ip->ip_len) { 538 q->ipqe_ip->ip_len -= i; 539 q->ipqe_ip->ip_off += i; 540 m_adj(q->ipqe_m, i); 541 break; 542 } 543 nq = q->ipqe_q.le_next; 544 m_freem(q->ipqe_m); 545 LIST_REMOVE(q, ipqe_q); 546 FREE(q, M_IPQ); 547 } 548 549insert: 550 /* 551 * Stick new segment in its place; 552 * check for complete reassembly. 553 */ 554 if (p == NULL) { 555 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 556 } else { 557 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 558 } 559 next = 0; 560 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 561 p = q, q = q->ipqe_q.le_next) { 562 if (q->ipqe_ip->ip_off != next) 563 return (0); 564 next += q->ipqe_ip->ip_len; 565 } 566 if (p->ipqe_mff) 567 return (0); 568 569 /* 570 * Reassembly is complete. Check for a bogus message size and 571 * concatenate fragments. 572 */ 573 q = fp->ipq_fragq.lh_first; 574 ip = q->ipqe_ip; 575 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 576 ipstat.ips_toolong++; 577 ip_freef(fp); 578 return (0); 579 } 580 m = q->ipqe_m; 581 t = m->m_next; 582 m->m_next = 0; 583 m_cat(m, t); 584 nq = q->ipqe_q.le_next; 585 FREE(q, M_IPQ); 586 for (q = nq; q != NULL; q = nq) { 587 t = q->ipqe_m; 588 nq = q->ipqe_q.le_next; 589 FREE(q, M_IPQ); 590 m_cat(m, t); 591 } 592 593 /* 594 * Create header for new ip packet by 595 * modifying header of first packet; 596 * dequeue and discard fragment reassembly header. 597 * Make header visible. 598 */ 599 ip->ip_len = next; 600 ip->ip_src = fp->ipq_src; 601 ip->ip_dst = fp->ipq_dst; 602 LIST_REMOVE(fp, ipq_q); 603 FREE(fp, M_FTABLE); 604 m->m_len += (ip->ip_hl << 2); 605 m->m_data -= (ip->ip_hl << 2); 606 /* some debugging cruft by sklower, below, will go away soon */ 607 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 608 register int plen = 0; 609 for (t = m; t; t = t->m_next) 610 plen += t->m_len; 611 m->m_pkthdr.len = plen; 612 } 613 return (m); 614 615dropfrag: 616 ipstat.ips_fragdropped++; 617 m_freem(m); 618 FREE(ipqe, M_IPQ); 619 return (0); 620} 621 622/* 623 * Free a fragment reassembly header and all 624 * associated datagrams. 625 */ 626void 627ip_freef(fp) 628 struct ipq *fp; 629{ 630 register struct ipqent *q, *p; 631 632 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 633 p = q->ipqe_q.le_next; 634 m_freem(q->ipqe_m); 635 LIST_REMOVE(q, ipqe_q); 636 FREE(q, M_IPQ); 637 } 638 LIST_REMOVE(fp, ipq_q); 639 FREE(fp, M_FTABLE); 640} 641 642/* 643 * IP timer processing; 644 * if a timer expires on a reassembly 645 * queue, discard it. 646 */ 647void 648ip_slowtimo() 649{ 650 register struct ipq *fp, *nfp; 651 int s = splsoftnet(); 652 653 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 654 nfp = fp->ipq_q.le_next; 655 if (--fp->ipq_ttl == 0) { 656 ipstat.ips_fragtimeout++; 657 ip_freef(fp); 658 } 659 } 660 splx(s); 661} 662 663/* 664 * Drain off all datagram fragments. 665 */ 666void 667ip_drain() 668{ 669 670 while (ipq.lh_first != NULL) { 671 ipstat.ips_fragdropped++; 672 ip_freef(ipq.lh_first); 673 } 674} 675 676/* 677 * Do option processing on a datagram, 678 * possibly discarding it if bad options are encountered, 679 * or forwarding it if source-routed. 680 * Returns 1 if packet has been forwarded/freed, 681 * 0 if the packet should be processed further. 682 */ 683int 684ip_dooptions(m) 685 struct mbuf *m; 686{ 687 register struct ip *ip = mtod(m, struct ip *); 688 register u_char *cp; 689 register struct ip_timestamp *ipt; 690 register struct in_ifaddr *ia; 691 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 692 struct in_addr *sin, dst; 693 n_time ntime; 694 695 dst = ip->ip_dst; 696 cp = (u_char *)(ip + 1); 697 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 698 for (; cnt > 0; cnt -= optlen, cp += optlen) { 699 opt = cp[IPOPT_OPTVAL]; 700 if (opt == IPOPT_EOL) 701 break; 702 if (opt == IPOPT_NOP) 703 optlen = 1; 704 else { 705 optlen = cp[IPOPT_OLEN]; 706 if (optlen <= 0 || optlen > cnt) { 707 code = &cp[IPOPT_OLEN] - (u_char *)ip; 708 goto bad; 709 } 710 } 711 switch (opt) { 712 713 default: 714 break; 715 716 /* 717 * Source routing with record. 718 * Find interface with current destination address. 719 * If none on this machine then drop if strictly routed, 720 * or do nothing if loosely routed. 721 * Record interface address and bring up next address 722 * component. If strictly routed make sure next 723 * address is on directly accessible net. 724 */ 725 case IPOPT_LSRR: 726 case IPOPT_SSRR: 727 if (ip_allowsrcrt == 0) { 728 type = ICMP_UNREACH; 729 code = ICMP_UNREACH_NET_PROHIB; 730 goto bad; 731 } 732 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 733 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 734 goto bad; 735 } 736 ipaddr.sin_addr = ip->ip_dst; 737 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 738 if (ia == 0) { 739 if (opt == IPOPT_SSRR) { 740 type = ICMP_UNREACH; 741 code = ICMP_UNREACH_SRCFAIL; 742 goto bad; 743 } 744 /* 745 * Loose routing, and not at next destination 746 * yet; nothing to do except forward. 747 */ 748 break; 749 } 750 off--; /* 0 origin */ 751 if (off > optlen - sizeof(struct in_addr)) { 752 /* 753 * End of source route. Should be for us. 754 */ 755 save_rte(cp, ip->ip_src); 756 break; 757 } 758 /* 759 * locate outgoing interface 760 */ 761 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 762 sizeof(ipaddr.sin_addr)); 763 if (opt == IPOPT_SSRR) { 764#define INA struct in_ifaddr * 765#define SA struct sockaddr * 766 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 767 } else 768 ia = ip_rtaddr(ipaddr.sin_addr); 769 if (ia == 0) { 770 type = ICMP_UNREACH; 771 code = ICMP_UNREACH_SRCFAIL; 772 goto bad; 773 } 774 ip->ip_dst = ipaddr.sin_addr; 775 bcopy((caddr_t)&ia->ia_addr.sin_addr, 776 (caddr_t)(cp + off), sizeof(struct in_addr)); 777 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 778 /* 779 * Let ip_intr's mcast routing check handle mcast pkts 780 */ 781 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 782 break; 783 784 case IPOPT_RR: 785 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 786 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 787 goto bad; 788 } 789 /* 790 * If no space remains, ignore. 791 */ 792 off--; /* 0 origin */ 793 if (off > optlen - sizeof(struct in_addr)) 794 break; 795 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 796 sizeof(ipaddr.sin_addr)); 797 /* 798 * locate outgoing interface; if we're the destination, 799 * use the incoming interface (should be same). 800 */ 801 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 802 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 803 type = ICMP_UNREACH; 804 code = ICMP_UNREACH_HOST; 805 goto bad; 806 } 807 bcopy((caddr_t)&ia->ia_addr.sin_addr, 808 (caddr_t)(cp + off), sizeof(struct in_addr)); 809 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 810 break; 811 812 case IPOPT_TS: 813 code = cp - (u_char *)ip; 814 ipt = (struct ip_timestamp *)cp; 815 if (ipt->ipt_len < 5) 816 goto bad; 817 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 818 if (++ipt->ipt_oflw == 0) 819 goto bad; 820 break; 821 } 822 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 823 switch (ipt->ipt_flg) { 824 825 case IPOPT_TS_TSONLY: 826 break; 827 828 case IPOPT_TS_TSANDADDR: 829 if (ipt->ipt_ptr + sizeof(n_time) + 830 sizeof(struct in_addr) > ipt->ipt_len) 831 goto bad; 832 ipaddr.sin_addr = dst; 833 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 834 m->m_pkthdr.rcvif); 835 if (ia == 0) 836 continue; 837 bcopy((caddr_t)&ia->ia_addr.sin_addr, 838 (caddr_t)sin, sizeof(struct in_addr)); 839 ipt->ipt_ptr += sizeof(struct in_addr); 840 break; 841 842 case IPOPT_TS_PRESPEC: 843 if (ipt->ipt_ptr + sizeof(n_time) + 844 sizeof(struct in_addr) > ipt->ipt_len) 845 goto bad; 846 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 847 sizeof(struct in_addr)); 848 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 849 continue; 850 ipt->ipt_ptr += sizeof(struct in_addr); 851 break; 852 853 default: 854 goto bad; 855 } 856 ntime = iptime(); 857 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 858 sizeof(n_time)); 859 ipt->ipt_ptr += sizeof(n_time); 860 } 861 } 862 if (forward) { 863 if (ip_forwsrcrt == 0) { 864 type = ICMP_UNREACH; 865 code = ICMP_UNREACH_SRCFAIL; 866 goto bad; 867 } 868 ip_forward(m, 1); 869 return (1); 870 } 871 return (0); 872bad: 873 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 874 icmp_error(m, type, code, 0, 0); 875 ipstat.ips_badoptions++; 876 return (1); 877} 878 879/* 880 * Given address of next destination (final or next hop), 881 * return internet address info of interface to be used to get there. 882 */ 883struct in_ifaddr * 884ip_rtaddr(dst) 885 struct in_addr dst; 886{ 887 register struct sockaddr_in *sin; 888 889 sin = satosin(&ipforward_rt.ro_dst); 890 891 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 892 if (ipforward_rt.ro_rt) { 893 RTFREE(ipforward_rt.ro_rt); 894 ipforward_rt.ro_rt = 0; 895 } 896 sin->sin_family = AF_INET; 897 sin->sin_len = sizeof(*sin); 898 sin->sin_addr = dst; 899 900 rtalloc(&ipforward_rt); 901 } 902 if (ipforward_rt.ro_rt == 0) 903 return ((struct in_ifaddr *)0); 904 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 905} 906 907/* 908 * Save incoming source route for use in replies, 909 * to be picked up later by ip_srcroute if the receiver is interested. 910 */ 911void 912save_rte(option, dst) 913 u_char *option; 914 struct in_addr dst; 915{ 916 unsigned olen; 917 918 olen = option[IPOPT_OLEN]; 919#ifdef DIAGNOSTIC 920 if (ipprintfs) 921 printf("save_rte: olen %d\n", olen); 922#endif 923 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 924 return; 925 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 926 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 927 ip_srcrt.dst = dst; 928} 929 930/* 931 * Retrieve incoming source route for use in replies, 932 * in the same form used by setsockopt. 933 * The first hop is placed before the options, will be removed later. 934 */ 935struct mbuf * 936ip_srcroute() 937{ 938 register struct in_addr *p, *q; 939 register struct mbuf *m; 940 941 if (ip_nhops == 0) 942 return ((struct mbuf *)0); 943 m = m_get(M_DONTWAIT, MT_SOOPTS); 944 if (m == 0) 945 return ((struct mbuf *)0); 946 947#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 948 949 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 950 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 951 OPTSIZ; 952#ifdef DIAGNOSTIC 953 if (ipprintfs) 954 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 955#endif 956 957 /* 958 * First save first hop for return route 959 */ 960 p = &ip_srcrt.route[ip_nhops - 1]; 961 *(mtod(m, struct in_addr *)) = *p--; 962#ifdef DIAGNOSTIC 963 if (ipprintfs) 964 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 965#endif 966 967 /* 968 * Copy option fields and padding (nop) to mbuf. 969 */ 970 ip_srcrt.nop = IPOPT_NOP; 971 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 972 bcopy((caddr_t)&ip_srcrt.nop, 973 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 974 q = (struct in_addr *)(mtod(m, caddr_t) + 975 sizeof(struct in_addr) + OPTSIZ); 976#undef OPTSIZ 977 /* 978 * Record return path as an IP source route, 979 * reversing the path (pointers are now aligned). 980 */ 981 while (p >= ip_srcrt.route) { 982#ifdef DIAGNOSTIC 983 if (ipprintfs) 984 printf(" %x", ntohl(q->s_addr)); 985#endif 986 *q++ = *p--; 987 } 988 /* 989 * Last hop goes to final destination. 990 */ 991 *q = ip_srcrt.dst; 992#ifdef DIAGNOSTIC 993 if (ipprintfs) 994 printf(" %x\n", ntohl(q->s_addr)); 995#endif 996 return (m); 997} 998 999/* 1000 * Strip out IP options, at higher 1001 * level protocol in the kernel. 1002 * Second argument is buffer to which options 1003 * will be moved, and return value is their length. 1004 * XXX should be deleted; last arg currently ignored. 1005 */ 1006void 1007ip_stripoptions(m, mopt) 1008 register struct mbuf *m; 1009 struct mbuf *mopt; 1010{ 1011 register int i; 1012 struct ip *ip = mtod(m, struct ip *); 1013 register caddr_t opts; 1014 int olen; 1015 1016 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1017 opts = (caddr_t)(ip + 1); 1018 i = m->m_len - (sizeof (struct ip) + olen); 1019 bcopy(opts + olen, opts, (unsigned)i); 1020 m->m_len -= olen; 1021 if (m->m_flags & M_PKTHDR) 1022 m->m_pkthdr.len -= olen; 1023 ip->ip_hl = sizeof(struct ip) >> 2; 1024} 1025 1026int inetctlerrmap[PRC_NCMDS] = { 1027 0, 0, 0, 0, 1028 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1029 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1030 EMSGSIZE, EHOSTUNREACH, 0, 0, 1031 0, 0, 0, 0, 1032 ENOPROTOOPT 1033}; 1034 1035/* 1036 * Forward a packet. If some error occurs return the sender 1037 * an icmp packet. Note we can't always generate a meaningful 1038 * icmp message because icmp doesn't have a large enough repertoire 1039 * of codes and types. 1040 * 1041 * If not forwarding, just drop the packet. This could be confusing 1042 * if ipforwarding was zero but some routing protocol was advancing 1043 * us as a gateway to somewhere. However, we must let the routing 1044 * protocol deal with that. 1045 * 1046 * The srcrt parameter indicates whether the packet is being forwarded 1047 * via a source route. 1048 */ 1049void 1050ip_forward(m, srcrt) 1051 struct mbuf *m; 1052 int srcrt; 1053{ 1054 register struct ip *ip = mtod(m, struct ip *); 1055 register struct sockaddr_in *sin; 1056 register struct rtentry *rt; 1057 int error, type = 0, code = 0; 1058 struct mbuf *mcopy; 1059 n_long dest; 1060 struct ifnet *destifp; 1061 1062 dest = 0; 1063#ifdef DIAGNOSTIC 1064 if (ipprintfs) 1065 printf("forward: src %x dst %x ttl %x\n", 1066 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1067#endif 1068 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1069 ipstat.ips_cantforward++; 1070 m_freem(m); 1071 return; 1072 } 1073 HTONS(ip->ip_id); 1074 if (ip->ip_ttl <= IPTTLDEC) { 1075 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1076 return; 1077 } 1078 ip->ip_ttl -= IPTTLDEC; 1079 1080 sin = satosin(&ipforward_rt.ro_dst); 1081 if ((rt = ipforward_rt.ro_rt) == 0 || 1082 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1083 if (ipforward_rt.ro_rt) { 1084 RTFREE(ipforward_rt.ro_rt); 1085 ipforward_rt.ro_rt = 0; 1086 } 1087 sin->sin_family = AF_INET; 1088 sin->sin_len = sizeof(struct sockaddr_in); 1089 sin->sin_addr = ip->ip_dst; 1090 1091 rtalloc(&ipforward_rt); 1092 if (ipforward_rt.ro_rt == 0) { 1093 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1094 return; 1095 } 1096 rt = ipforward_rt.ro_rt; 1097 } 1098 1099 /* 1100 * Save at most 68 bytes of the packet in case 1101 * we need to generate an ICMP message to the src. 1102 */ 1103 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1104 1105 /* 1106 * If forwarding packet using same interface that it came in on, 1107 * perhaps should send a redirect to sender to shortcut a hop. 1108 * Only send redirect if source is sending directly to us, 1109 * and if packet was not source routed (or has any options). 1110 * Also, don't send redirect if forwarding using a default route 1111 * or a route modified by a redirect. 1112 */ 1113 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1114 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1115 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1116 ipsendredirects && !srcrt) { 1117 if (rt->rt_ifa && 1118 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1119 ifatoia(rt->rt_ifa)->ia_subnet) { 1120 if (rt->rt_flags & RTF_GATEWAY) 1121 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1122 else 1123 dest = ip->ip_dst.s_addr; 1124 /* Router requirements says to only send host redirects */ 1125 type = ICMP_REDIRECT; 1126 code = ICMP_REDIRECT_HOST; 1127#ifdef DIAGNOSTIC 1128 if (ipprintfs) 1129 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1130#endif 1131 } 1132 } 1133 1134 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1135 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1136 if (error) 1137 ipstat.ips_cantforward++; 1138 else { 1139 ipstat.ips_forward++; 1140 if (type) 1141 ipstat.ips_redirectsent++; 1142 else { 1143 if (mcopy) 1144 m_freem(mcopy); 1145 return; 1146 } 1147 } 1148 if (mcopy == NULL) 1149 return; 1150 destifp = NULL; 1151 1152 switch (error) { 1153 1154 case 0: /* forwarded, but need redirect */ 1155 /* type, code set above */ 1156 break; 1157 1158 case ENETUNREACH: /* shouldn't happen, checked above */ 1159 case EHOSTUNREACH: 1160 case ENETDOWN: 1161 case EHOSTDOWN: 1162 default: 1163 type = ICMP_UNREACH; 1164 code = ICMP_UNREACH_HOST; 1165 break; 1166 1167 case EMSGSIZE: 1168 type = ICMP_UNREACH; 1169 code = ICMP_UNREACH_NEEDFRAG; 1170 if (ipforward_rt.ro_rt) 1171 destifp = ipforward_rt.ro_rt->rt_ifp; 1172 ipstat.ips_cantfrag++; 1173 break; 1174 1175 case ENOBUFS: 1176 type = ICMP_SOURCEQUENCH; 1177 code = 0; 1178 break; 1179 } 1180 icmp_error(mcopy, type, code, dest, destifp); 1181} 1182 1183void 1184ip_savecontrol(inp, mp, ip, m) 1185 register struct inpcb *inp; 1186 register struct mbuf **mp; 1187 register struct ip *ip; 1188 register struct mbuf *m; 1189{ 1190 1191 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1192 struct timeval tv; 1193 1194 microtime(&tv); 1195 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1196 SCM_TIMESTAMP, SOL_SOCKET); 1197 if (*mp) 1198 mp = &(*mp)->m_next; 1199 } 1200 if (inp->inp_flags & INP_RECVDSTADDR) { 1201 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1202 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1203 if (*mp) 1204 mp = &(*mp)->m_next; 1205 } 1206#ifdef notyet 1207 /* 1208 * XXX 1209 * Moving these out of udp_input() made them even more broken 1210 * than they already were. 1211 * - fenner@parc.xerox.com 1212 */ 1213 /* options were tossed already */ 1214 if (inp->inp_flags & INP_RECVOPTS) { 1215 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1216 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1217 if (*mp) 1218 mp = &(*mp)->m_next; 1219 } 1220 /* ip_srcroute doesn't do what we want here, need to fix */ 1221 if (inp->inp_flags & INP_RECVRETOPTS) { 1222 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1223 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1224 if (*mp) 1225 mp = &(*mp)->m_next; 1226 } 1227#endif 1228 if (inp->inp_flags & INP_RECVIF) { 1229 struct sockaddr_dl sdl; 1230 1231 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1232 sdl.sdl_family = AF_LINK; 1233 sdl.sdl_index = m->m_pkthdr.rcvif ? 1234 m->m_pkthdr.rcvif->if_index : 0; 1235 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1236 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1237 IP_RECVIF, IPPROTO_IP); 1238 if (*mp) 1239 mp = &(*mp)->m_next; 1240 } 1241} 1242 1243int 1244ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1245 int *name; 1246 u_int namelen; 1247 void *oldp; 1248 size_t *oldlenp; 1249 void *newp; 1250 size_t newlen; 1251{ 1252 extern int subnetsarelocal; 1253 1254 int error, old; 1255 1256 /* All sysctl names at this level are terminal. */ 1257 if (namelen != 1) 1258 return (ENOTDIR); 1259 1260 switch (name[0]) { 1261 case IPCTL_FORWARDING: 1262 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1263 case IPCTL_SENDREDIRECTS: 1264 return (sysctl_int(oldp, oldlenp, newp, newlen, 1265 &ipsendredirects)); 1266 case IPCTL_DEFTTL: 1267 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1268#ifdef notyet 1269 case IPCTL_DEFMTU: 1270 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1271#endif 1272 case IPCTL_FORWSRCRT: 1273 /* Don't allow this to change in a secure environment. */ 1274 if (securelevel > 0) 1275 return (sysctl_rdint(oldp, oldlenp, newp, 1276 ip_forwsrcrt)); 1277 else 1278 return (sysctl_int(oldp, oldlenp, newp, newlen, 1279 &ip_forwsrcrt)); 1280 case IPCTL_DIRECTEDBCAST: 1281 return (sysctl_int(oldp, oldlenp, newp, newlen, 1282 &ip_directedbcast)); 1283 case IPCTL_ALLOWSRCRT: 1284 return (sysctl_int(oldp, oldlenp, newp, newlen, 1285 &ip_allowsrcrt)); 1286 case IPCTL_SUBNETSARELOCAL: 1287 return (sysctl_int(oldp, oldlenp, newp, newlen, 1288 &subnetsarelocal)); 1289 case IPCTL_MTUDISC: 1290 return (sysctl_int(oldp, oldlenp, newp, newlen, 1291 &ip_mtudisc)); 1292 case IPCTL_ANONPORTMIN: 1293 old = anonportmin; 1294 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1295 if (anonportmin >= anonportmax || anonportmin > 65535 1296#ifndef IPNOPRIVPORTS 1297 || anonportmin < IPPORT_RESERVED 1298#endif 1299 ) { 1300 anonportmin = old; 1301 return (EINVAL); 1302 } 1303 return (error); 1304 case IPCTL_ANONPORTMAX: 1305 old = anonportmax; 1306 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1307 if (anonportmin >= anonportmax || anonportmax > 65535 1308#ifndef IPNOPRIVPORTS 1309 || anonportmax < IPPORT_RESERVED 1310#endif 1311 ) { 1312 anonportmax = old; 1313 return (EINVAL); 1314 } 1315 return (error); 1316 default: 1317 return (EOPNOTSUPP); 1318 } 1319 /* NOTREACHED */ 1320} 1321