ip_input.c revision 1.44
1/* $NetBSD: ip_input.c,v 1.44 1997/01/11 05:21:10 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 <sys/param.h> 39#include <sys/systm.h> 40#include <sys/malloc.h> 41#include <sys/mbuf.h> 42#include <sys/domain.h> 43#include <sys/protosw.h> 44#include <sys/socket.h> 45#include <sys/socketvar.h> 46#include <sys/errno.h> 47#include <sys/time.h> 48#include <sys/kernel.h> 49#include <sys/proc.h> 50 51#include <vm/vm.h> 52#include <sys/sysctl.h> 53 54#include <net/if.h> 55#include <net/if_dl.h> 56#include <net/route.h> 57 58#include <netinet/in.h> 59#include <netinet/in_systm.h> 60#include <netinet/ip.h> 61#include <netinet/in_pcb.h> 62#include <netinet/in_var.h> 63#include <netinet/ip_var.h> 64#include <netinet/ip_icmp.h> 65 66#ifdef PFIL_HOOKS 67#include <net/pfil.h> 68#endif /* PFIL_HOOKS */ 69 70/* XXX should really put this in libkern.h */ 71#define offsetof(type, member) ((size_t)(&((type *)0)->member)) 72 73#ifndef IPFORWARDING 74#ifdef GATEWAY 75#define IPFORWARDING 1 /* forward IP packets not for us */ 76#else /* GATEWAY */ 77#define IPFORWARDING 0 /* don't forward IP packets not for us */ 78#endif /* GATEWAY */ 79#endif /* IPFORWARDING */ 80#ifndef IPSENDREDIRECTS 81#define IPSENDREDIRECTS 1 82#endif 83#ifndef IPFORWSRCRT 84#define IPFORWSRCRT 1 /* allow source-routed packets */ 85#endif 86/* 87 * Note: DIRECTED_BROADCAST is handled this way so that previous 88 * configuration using this option will Just Work. 89 */ 90#ifndef IPDIRECTEDBCAST 91#ifdef DIRECTED_BROADCAST 92#define IPDIRECTEDBCAST 1 93#else 94#define IPDIRECTEDBCAST 0 95#endif /* DIRECTED_BROADCAST */ 96#endif /* IPDIRECTEDBCAST */ 97int ipforwarding = IPFORWARDING; 98int ipsendredirects = IPSENDREDIRECTS; 99int ip_defttl = IPDEFTTL; 100int ip_forwsrcrt = IPFORWSRCRT; 101int ip_directedbcast = IPDIRECTEDBCAST; 102#ifdef DIAGNOSTIC 103int ipprintfs = 0; 104#endif 105 106extern struct domain inetdomain; 107extern struct protosw inetsw[]; 108u_char ip_protox[IPPROTO_MAX]; 109int ipqmaxlen = IFQ_MAXLEN; 110struct in_ifaddrhead in_ifaddr; 111struct ifqueue ipintrq; 112 113/* 114 * We need to save the IP options in case a protocol wants to respond 115 * to an incoming packet over the same route if the packet got here 116 * using IP source routing. This allows connection establishment and 117 * maintenance when the remote end is on a network that is not known 118 * to us. 119 */ 120int ip_nhops = 0; 121static struct ip_srcrt { 122 struct in_addr dst; /* final destination */ 123 char nop; /* one NOP to align */ 124 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 125 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 126} ip_srcrt; 127 128static void save_rte __P((u_char *, struct in_addr)); 129 130/* 131 * IP initialization: fill in IP protocol switch table. 132 * All protocols not implemented in kernel go to raw IP protocol handler. 133 */ 134void 135ip_init() 136{ 137 register struct protosw *pr; 138 register int i; 139 140 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 141 if (pr == 0) 142 panic("ip_init"); 143 for (i = 0; i < IPPROTO_MAX; i++) 144 ip_protox[i] = pr - inetsw; 145 for (pr = inetdomain.dom_protosw; 146 pr < inetdomain.dom_protoswNPROTOSW; pr++) 147 if (pr->pr_domain->dom_family == PF_INET && 148 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 149 ip_protox[pr->pr_protocol] = pr - inetsw; 150 LIST_INIT(&ipq); 151 ip_id = time.tv_sec & 0xffff; 152 ipintrq.ifq_maxlen = ipqmaxlen; 153 TAILQ_INIT(&in_ifaddr); 154} 155 156struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 157struct route ipforward_rt; 158 159/* 160 * Ip input routine. Checksum and byte swap header. If fragmented 161 * try to reassemble. Process options. Pass to next level. 162 */ 163void 164ipintr() 165{ 166 register struct ip *ip = NULL; 167 register struct mbuf *m; 168 register struct ipq *fp; 169 register struct in_ifaddr *ia; 170 struct ipqent *ipqe; 171 int hlen = 0, mff, len, s; 172#ifdef PFIL_HOOKS 173 struct packet_filter_hook *pfh; 174 struct mbuf *m0; 175 int rv; 176#endif /* PFIL_HOOKS */ 177 178next: 179 /* 180 * Get next datagram off input queue and get IP header 181 * in first mbuf. 182 */ 183 s = splimp(); 184 IF_DEQUEUE(&ipintrq, m); 185 splx(s); 186 if (m == 0) 187 return; 188#ifdef DIAGNOSTIC 189 if ((m->m_flags & M_PKTHDR) == 0) 190 panic("ipintr no HDR"); 191#endif 192 /* 193 * If no IP addresses have been set yet but the interfaces 194 * are receiving, can't do anything with incoming packets yet. 195 */ 196 if (in_ifaddr.tqh_first == 0) 197 goto bad; 198 ipstat.ips_total++; 199 if (m->m_len < sizeof (struct ip) && 200 (m = m_pullup(m, sizeof (struct ip))) == 0) { 201 ipstat.ips_toosmall++; 202 goto next; 203 } 204 ip = mtod(m, struct ip *); 205 if (ip->ip_v != IPVERSION) { 206 ipstat.ips_badvers++; 207 goto bad; 208 } 209 hlen = ip->ip_hl << 2; 210 if (hlen < sizeof(struct ip)) { /* minimum header length */ 211 ipstat.ips_badhlen++; 212 goto bad; 213 } 214 if (hlen > m->m_len) { 215 if ((m = m_pullup(m, hlen)) == 0) { 216 ipstat.ips_badhlen++; 217 goto next; 218 } 219 ip = mtod(m, struct ip *); 220 } 221 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 222 ipstat.ips_badsum++; 223 goto bad; 224 } 225 226 /* 227 * Convert fields to host representation. 228 */ 229 NTOHS(ip->ip_len); 230 NTOHS(ip->ip_id); 231 NTOHS(ip->ip_off); 232 len = ip->ip_len; 233 234 /* 235 * Check that the amount of data in the buffers 236 * is as at least much as the IP header would have us expect. 237 * Trim mbufs if longer than we expect. 238 * Drop packet if shorter than we expect. 239 */ 240 if (m->m_pkthdr.len < len) { 241 ipstat.ips_tooshort++; 242 goto bad; 243 } 244 if (m->m_pkthdr.len > len) { 245 if (m->m_len == m->m_pkthdr.len) { 246 m->m_len = len; 247 m->m_pkthdr.len = len; 248 } else 249 m_adj(m, len - m->m_pkthdr.len); 250 } 251 252#ifdef PFIL_HOOKS 253 /* 254 * Run through list of hooks for input packets. 255 */ 256 m0 = m; 257 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 258 if (pfh->pfil_func) { 259 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 260 ip = mtod(m = m0, struct ip *); 261 if (rv) 262 goto next; 263 } 264#endif /* PFIL_HOOKS */ 265 266 /* 267 * Process options and, if not destined for us, 268 * ship it on. ip_dooptions returns 1 when an 269 * error was detected (causing an icmp message 270 * to be sent and the original packet to be freed). 271 */ 272 ip_nhops = 0; /* for source routed packets */ 273 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 274 goto next; 275 276 /* 277 * Check our list of addresses, to see if the packet is for us. 278 */ 279 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 280 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 281 goto ours; 282 if (((ip_directedbcast == 0) || (ip_directedbcast && 283 ia->ia_ifp == m->m_pkthdr.rcvif)) && 284 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 285 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 286 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 287 /* 288 * Look for all-0's host part (old broadcast addr), 289 * either for subnet or net. 290 */ 291 ip->ip_dst.s_addr == ia->ia_subnet || 292 ip->ip_dst.s_addr == ia->ia_net) 293 goto ours; 294 } 295 } 296 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 297 struct in_multi *inm; 298#ifdef MROUTING 299 extern struct socket *ip_mrouter; 300 301 if (m->m_flags & M_EXT) { 302 if ((m = m_pullup(m, hlen)) == 0) { 303 ipstat.ips_toosmall++; 304 goto next; 305 } 306 ip = mtod(m, struct ip *); 307 } 308 309 if (ip_mrouter) { 310 /* 311 * If we are acting as a multicast router, all 312 * incoming multicast packets are passed to the 313 * kernel-level multicast forwarding function. 314 * The packet is returned (relatively) intact; if 315 * ip_mforward() returns a non-zero value, the packet 316 * must be discarded, else it may be accepted below. 317 * 318 * (The IP ident field is put in the same byte order 319 * as expected when ip_mforward() is called from 320 * ip_output().) 321 */ 322 ip->ip_id = htons(ip->ip_id); 323 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 324 ipstat.ips_cantforward++; 325 m_freem(m); 326 goto next; 327 } 328 ip->ip_id = ntohs(ip->ip_id); 329 330 /* 331 * The process-level routing demon needs to receive 332 * all multicast IGMP packets, whether or not this 333 * host belongs to their destination groups. 334 */ 335 if (ip->ip_p == IPPROTO_IGMP) 336 goto ours; 337 ipstat.ips_forward++; 338 } 339#endif 340 /* 341 * See if we belong to the destination multicast group on the 342 * arrival interface. 343 */ 344 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 345 if (inm == NULL) { 346 ipstat.ips_cantforward++; 347 m_freem(m); 348 goto next; 349 } 350 goto ours; 351 } 352 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 353 in_nullhost(ip->ip_dst)) 354 goto ours; 355 356 /* 357 * Not for us; forward if possible and desirable. 358 */ 359 if (ipforwarding == 0) { 360 ipstat.ips_cantforward++; 361 m_freem(m); 362 } else 363 ip_forward(m, 0); 364 goto next; 365 366ours: 367 /* 368 * If offset or IP_MF are set, must reassemble. 369 * Otherwise, nothing need be done. 370 * (We could look in the reassembly queue to see 371 * if the packet was previously fragmented, 372 * but it's not worth the time; just let them time out.) 373 */ 374 if (ip->ip_off & ~(IP_DF|IP_RF)) { 375 if (m->m_flags & M_EXT) { /* XXX */ 376 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 377 ipstat.ips_toosmall++; 378 goto next; 379 } 380 ip = mtod(m, struct ip *); 381 } 382 /* 383 * Look for queue of fragments 384 * of this datagram. 385 */ 386 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 387 if (ip->ip_id == fp->ipq_id && 388 in_hosteq(ip->ip_src, fp->ipq_src) && 389 in_hosteq(ip->ip_dst, fp->ipq_dst) && 390 ip->ip_p == fp->ipq_p) 391 goto found; 392 fp = 0; 393found: 394 395 /* 396 * Adjust ip_len to not reflect header, 397 * set ipqe_mff if more fragments are expected, 398 * convert offset of this to bytes. 399 */ 400 ip->ip_len -= hlen; 401 mff = (ip->ip_off & IP_MF) != 0; 402 if (mff) { 403 /* 404 * Make sure that fragments have a data length 405 * that's a non-zero multiple of 8 bytes. 406 */ 407 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 408 ipstat.ips_badfrags++; 409 goto bad; 410 } 411 } 412 ip->ip_off <<= 3; 413 414 /* 415 * If datagram marked as having more fragments 416 * or if this is not the first fragment, 417 * attempt reassembly; if it succeeds, proceed. 418 */ 419 if (mff || ip->ip_off) { 420 ipstat.ips_fragments++; 421 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 422 M_IPQ, M_NOWAIT); 423 if (ipqe == NULL) { 424 ipstat.ips_rcvmemdrop++; 425 goto bad; 426 } 427 ipqe->ipqe_mff = mff; 428 ipqe->ipqe_ip = ip; 429 ip = ip_reass(ipqe, fp); 430 if (ip == 0) 431 goto next; 432 ipstat.ips_reassembled++; 433 m = dtom(ip); 434 } else 435 if (fp) 436 ip_freef(fp); 437 } else 438 ip->ip_len -= hlen; 439 440 /* 441 * Switch out to protocol's input routine. 442 */ 443 ipstat.ips_delivered++; 444 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 445 goto next; 446bad: 447 m_freem(m); 448 goto next; 449} 450 451/* 452 * Take incoming datagram fragment and try to 453 * reassemble it into whole datagram. If a chain for 454 * reassembly of this datagram already exists, then it 455 * is given as fp; otherwise have to make a chain. 456 */ 457struct ip * 458ip_reass(ipqe, fp) 459 register struct ipqent *ipqe; 460 register struct ipq *fp; 461{ 462 register struct mbuf *m = dtom(ipqe->ipqe_ip); 463 register struct ipqent *nq, *p, *q; 464 struct ip *ip; 465 struct mbuf *t; 466 int hlen = ipqe->ipqe_ip->ip_hl << 2; 467 int i, next; 468 469 /* 470 * Presence of header sizes in mbufs 471 * would confuse code below. 472 */ 473 m->m_data += hlen; 474 m->m_len -= hlen; 475 476 /* 477 * If first fragment to arrive, create a reassembly queue. 478 */ 479 if (fp == 0) { 480 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 481 goto dropfrag; 482 fp = mtod(t, struct ipq *); 483 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 484 fp->ipq_ttl = IPFRAGTTL; 485 fp->ipq_p = ipqe->ipqe_ip->ip_p; 486 fp->ipq_id = ipqe->ipqe_ip->ip_id; 487 LIST_INIT(&fp->ipq_fragq); 488 fp->ipq_src = ipqe->ipqe_ip->ip_src; 489 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 490 p = NULL; 491 goto insert; 492 } 493 494 /* 495 * Find a segment which begins after this one does. 496 */ 497 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 498 p = q, q = q->ipqe_q.le_next) 499 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 500 break; 501 502 /* 503 * If there is a preceding segment, it may provide some of 504 * our data already. If so, drop the data from the incoming 505 * segment. If it provides all of our data, drop us. 506 */ 507 if (p != NULL) { 508 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 509 ipqe->ipqe_ip->ip_off; 510 if (i > 0) { 511 if (i >= ipqe->ipqe_ip->ip_len) 512 goto dropfrag; 513 m_adj(dtom(ipqe->ipqe_ip), i); 514 ipqe->ipqe_ip->ip_off += i; 515 ipqe->ipqe_ip->ip_len -= i; 516 } 517 } 518 519 /* 520 * While we overlap succeeding segments trim them or, 521 * if they are completely covered, dequeue them. 522 */ 523 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 524 q->ipqe_ip->ip_off; q = nq) { 525 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 526 q->ipqe_ip->ip_off; 527 if (i < q->ipqe_ip->ip_len) { 528 q->ipqe_ip->ip_len -= i; 529 q->ipqe_ip->ip_off += i; 530 m_adj(dtom(q->ipqe_ip), i); 531 break; 532 } 533 nq = q->ipqe_q.le_next; 534 m_freem(dtom(q->ipqe_ip)); 535 LIST_REMOVE(q, ipqe_q); 536 FREE(q, M_IPQ); 537 } 538 539insert: 540 /* 541 * Stick new segment in its place; 542 * check for complete reassembly. 543 */ 544 if (p == NULL) { 545 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 546 } else { 547 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 548 } 549 next = 0; 550 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 551 p = q, q = q->ipqe_q.le_next) { 552 if (q->ipqe_ip->ip_off != next) 553 return (0); 554 next += q->ipqe_ip->ip_len; 555 } 556 if (p->ipqe_mff) 557 return (0); 558 559 /* 560 * Reassembly is complete. Check for a bogus message size and 561 * concatenate fragments. 562 */ 563 q = fp->ipq_fragq.lh_first; 564 ip = q->ipqe_ip; 565 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 566 ipstat.ips_toolong++; 567 ip_freef(fp); 568 return (0); 569 } 570 m = dtom(q->ipqe_ip); 571 t = m->m_next; 572 m->m_next = 0; 573 m_cat(m, t); 574 nq = q->ipqe_q.le_next; 575 FREE(q, M_IPQ); 576 for (q = nq; q != NULL; q = nq) { 577 t = dtom(q->ipqe_ip); 578 nq = q->ipqe_q.le_next; 579 FREE(q, M_IPQ); 580 m_cat(m, t); 581 } 582 583 /* 584 * Create header for new ip packet by 585 * modifying header of first packet; 586 * dequeue and discard fragment reassembly header. 587 * Make header visible. 588 */ 589 ip->ip_len = next; 590 ip->ip_src = fp->ipq_src; 591 ip->ip_dst = fp->ipq_dst; 592 LIST_REMOVE(fp, ipq_q); 593 (void) m_free(dtom(fp)); 594 m->m_len += (ip->ip_hl << 2); 595 m->m_data -= (ip->ip_hl << 2); 596 /* some debugging cruft by sklower, below, will go away soon */ 597 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 598 register int plen = 0; 599 for (t = m; m; m = m->m_next) 600 plen += m->m_len; 601 t->m_pkthdr.len = plen; 602 } 603 return (ip); 604 605dropfrag: 606 ipstat.ips_fragdropped++; 607 m_freem(m); 608 FREE(ipqe, M_IPQ); 609 return (0); 610} 611 612/* 613 * Free a fragment reassembly header and all 614 * associated datagrams. 615 */ 616void 617ip_freef(fp) 618 struct ipq *fp; 619{ 620 register struct ipqent *q, *p; 621 622 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 623 p = q->ipqe_q.le_next; 624 m_freem(dtom(q->ipqe_ip)); 625 LIST_REMOVE(q, ipqe_q); 626 FREE(q, M_IPQ); 627 } 628 LIST_REMOVE(fp, ipq_q); 629 (void) m_free(dtom(fp)); 630} 631 632/* 633 * IP timer processing; 634 * if a timer expires on a reassembly 635 * queue, discard it. 636 */ 637void 638ip_slowtimo() 639{ 640 register struct ipq *fp, *nfp; 641 int s = splsoftnet(); 642 643 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 644 nfp = fp->ipq_q.le_next; 645 if (--fp->ipq_ttl == 0) { 646 ipstat.ips_fragtimeout++; 647 ip_freef(fp); 648 } 649 } 650 splx(s); 651} 652 653/* 654 * Drain off all datagram fragments. 655 */ 656void 657ip_drain() 658{ 659 660 while (ipq.lh_first != NULL) { 661 ipstat.ips_fragdropped++; 662 ip_freef(ipq.lh_first); 663 } 664} 665 666/* 667 * Do option processing on a datagram, 668 * possibly discarding it if bad options are encountered, 669 * or forwarding it if source-routed. 670 * Returns 1 if packet has been forwarded/freed, 671 * 0 if the packet should be processed further. 672 */ 673int 674ip_dooptions(m) 675 struct mbuf *m; 676{ 677 register struct ip *ip = mtod(m, struct ip *); 678 register u_char *cp; 679 register struct ip_timestamp *ipt; 680 register struct in_ifaddr *ia; 681 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 682 struct in_addr *sin, dst; 683 n_time ntime; 684 685 dst = ip->ip_dst; 686 cp = (u_char *)(ip + 1); 687 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 688 for (; cnt > 0; cnt -= optlen, cp += optlen) { 689 opt = cp[IPOPT_OPTVAL]; 690 if (opt == IPOPT_EOL) 691 break; 692 if (opt == IPOPT_NOP) 693 optlen = 1; 694 else { 695 optlen = cp[IPOPT_OLEN]; 696 if (optlen <= 0 || optlen > cnt) { 697 code = &cp[IPOPT_OLEN] - (u_char *)ip; 698 goto bad; 699 } 700 } 701 switch (opt) { 702 703 default: 704 break; 705 706 /* 707 * Source routing with record. 708 * Find interface with current destination address. 709 * If none on this machine then drop if strictly routed, 710 * or do nothing if loosely routed. 711 * Record interface address and bring up next address 712 * component. If strictly routed make sure next 713 * address is on directly accessible net. 714 */ 715 case IPOPT_LSRR: 716 case IPOPT_SSRR: 717 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 718 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 719 goto bad; 720 } 721 ipaddr.sin_addr = ip->ip_dst; 722 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 723 if (ia == 0) { 724 if (opt == IPOPT_SSRR) { 725 type = ICMP_UNREACH; 726 code = ICMP_UNREACH_SRCFAIL; 727 goto bad; 728 } 729 /* 730 * Loose routing, and not at next destination 731 * yet; nothing to do except forward. 732 */ 733 break; 734 } 735 off--; /* 0 origin */ 736 if (off > optlen - sizeof(struct in_addr)) { 737 /* 738 * End of source route. Should be for us. 739 */ 740 save_rte(cp, ip->ip_src); 741 break; 742 } 743 /* 744 * locate outgoing interface 745 */ 746 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 747 sizeof(ipaddr.sin_addr)); 748 if (opt == IPOPT_SSRR) { 749#define INA struct in_ifaddr * 750#define SA struct sockaddr * 751 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 752 } else 753 ia = ip_rtaddr(ipaddr.sin_addr); 754 if (ia == 0) { 755 type = ICMP_UNREACH; 756 code = ICMP_UNREACH_SRCFAIL; 757 goto bad; 758 } 759 ip->ip_dst = ipaddr.sin_addr; 760 bcopy((caddr_t)&ia->ia_addr.sin_addr, 761 (caddr_t)(cp + off), sizeof(struct in_addr)); 762 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 763 /* 764 * Let ip_intr's mcast routing check handle mcast pkts 765 */ 766 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 767 break; 768 769 case IPOPT_RR: 770 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 771 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 772 goto bad; 773 } 774 /* 775 * If no space remains, ignore. 776 */ 777 off--; /* 0 origin */ 778 if (off > optlen - sizeof(struct in_addr)) 779 break; 780 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 781 sizeof(ipaddr.sin_addr)); 782 /* 783 * locate outgoing interface; if we're the destination, 784 * use the incoming interface (should be same). 785 */ 786 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 787 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 788 type = ICMP_UNREACH; 789 code = ICMP_UNREACH_HOST; 790 goto bad; 791 } 792 bcopy((caddr_t)&ia->ia_addr.sin_addr, 793 (caddr_t)(cp + off), sizeof(struct in_addr)); 794 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 795 break; 796 797 case IPOPT_TS: 798 code = cp - (u_char *)ip; 799 ipt = (struct ip_timestamp *)cp; 800 if (ipt->ipt_len < 5) 801 goto bad; 802 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 803 if (++ipt->ipt_oflw == 0) 804 goto bad; 805 break; 806 } 807 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 808 switch (ipt->ipt_flg) { 809 810 case IPOPT_TS_TSONLY: 811 break; 812 813 case IPOPT_TS_TSANDADDR: 814 if (ipt->ipt_ptr + sizeof(n_time) + 815 sizeof(struct in_addr) > ipt->ipt_len) 816 goto bad; 817 ipaddr.sin_addr = dst; 818 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 819 m->m_pkthdr.rcvif); 820 if (ia == 0) 821 continue; 822 bcopy((caddr_t)&ia->ia_addr.sin_addr, 823 (caddr_t)sin, sizeof(struct in_addr)); 824 ipt->ipt_ptr += sizeof(struct in_addr); 825 break; 826 827 case IPOPT_TS_PRESPEC: 828 if (ipt->ipt_ptr + sizeof(n_time) + 829 sizeof(struct in_addr) > ipt->ipt_len) 830 goto bad; 831 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 832 sizeof(struct in_addr)); 833 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 834 continue; 835 ipt->ipt_ptr += sizeof(struct in_addr); 836 break; 837 838 default: 839 goto bad; 840 } 841 ntime = iptime(); 842 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 843 sizeof(n_time)); 844 ipt->ipt_ptr += sizeof(n_time); 845 } 846 } 847 if (forward) { 848 if (ip_forwsrcrt == 0) { 849 type = ICMP_UNREACH; 850 code = ICMP_UNREACH_SRCFAIL; 851 goto bad; 852 } 853 ip_forward(m, 1); 854 return (1); 855 } 856 return (0); 857bad: 858 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 859 icmp_error(m, type, code, 0, 0); 860 ipstat.ips_badoptions++; 861 return (1); 862} 863 864/* 865 * Given address of next destination (final or next hop), 866 * return internet address info of interface to be used to get there. 867 */ 868struct in_ifaddr * 869ip_rtaddr(dst) 870 struct in_addr dst; 871{ 872 register struct sockaddr_in *sin; 873 874 sin = satosin(&ipforward_rt.ro_dst); 875 876 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 877 if (ipforward_rt.ro_rt) { 878 RTFREE(ipforward_rt.ro_rt); 879 ipforward_rt.ro_rt = 0; 880 } 881 sin->sin_family = AF_INET; 882 sin->sin_len = sizeof(*sin); 883 sin->sin_addr = dst; 884 885 rtalloc(&ipforward_rt); 886 } 887 if (ipforward_rt.ro_rt == 0) 888 return ((struct in_ifaddr *)0); 889 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 890} 891 892/* 893 * Save incoming source route for use in replies, 894 * to be picked up later by ip_srcroute if the receiver is interested. 895 */ 896void 897save_rte(option, dst) 898 u_char *option; 899 struct in_addr dst; 900{ 901 unsigned olen; 902 903 olen = option[IPOPT_OLEN]; 904#ifdef DIAGNOSTIC 905 if (ipprintfs) 906 printf("save_rte: olen %d\n", olen); 907#endif 908 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 909 return; 910 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 911 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 912 ip_srcrt.dst = dst; 913} 914 915/* 916 * Retrieve incoming source route for use in replies, 917 * in the same form used by setsockopt. 918 * The first hop is placed before the options, will be removed later. 919 */ 920struct mbuf * 921ip_srcroute() 922{ 923 register struct in_addr *p, *q; 924 register struct mbuf *m; 925 926 if (ip_nhops == 0) 927 return ((struct mbuf *)0); 928 m = m_get(M_DONTWAIT, MT_SOOPTS); 929 if (m == 0) 930 return ((struct mbuf *)0); 931 932#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 933 934 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 935 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 936 OPTSIZ; 937#ifdef DIAGNOSTIC 938 if (ipprintfs) 939 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 940#endif 941 942 /* 943 * First save first hop for return route 944 */ 945 p = &ip_srcrt.route[ip_nhops - 1]; 946 *(mtod(m, struct in_addr *)) = *p--; 947#ifdef DIAGNOSTIC 948 if (ipprintfs) 949 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 950#endif 951 952 /* 953 * Copy option fields and padding (nop) to mbuf. 954 */ 955 ip_srcrt.nop = IPOPT_NOP; 956 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 957 bcopy((caddr_t)&ip_srcrt.nop, 958 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 959 q = (struct in_addr *)(mtod(m, caddr_t) + 960 sizeof(struct in_addr) + OPTSIZ); 961#undef OPTSIZ 962 /* 963 * Record return path as an IP source route, 964 * reversing the path (pointers are now aligned). 965 */ 966 while (p >= ip_srcrt.route) { 967#ifdef DIAGNOSTIC 968 if (ipprintfs) 969 printf(" %x", ntohl(q->s_addr)); 970#endif 971 *q++ = *p--; 972 } 973 /* 974 * Last hop goes to final destination. 975 */ 976 *q = ip_srcrt.dst; 977#ifdef DIAGNOSTIC 978 if (ipprintfs) 979 printf(" %x\n", ntohl(q->s_addr)); 980#endif 981 return (m); 982} 983 984/* 985 * Strip out IP options, at higher 986 * level protocol in the kernel. 987 * Second argument is buffer to which options 988 * will be moved, and return value is their length. 989 * XXX should be deleted; last arg currently ignored. 990 */ 991void 992ip_stripoptions(m, mopt) 993 register struct mbuf *m; 994 struct mbuf *mopt; 995{ 996 register int i; 997 struct ip *ip = mtod(m, struct ip *); 998 register caddr_t opts; 999 int olen; 1000 1001 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1002 opts = (caddr_t)(ip + 1); 1003 i = m->m_len - (sizeof (struct ip) + olen); 1004 bcopy(opts + olen, opts, (unsigned)i); 1005 m->m_len -= olen; 1006 if (m->m_flags & M_PKTHDR) 1007 m->m_pkthdr.len -= olen; 1008 ip->ip_hl = sizeof(struct ip) >> 2; 1009} 1010 1011int inetctlerrmap[PRC_NCMDS] = { 1012 0, 0, 0, 0, 1013 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1014 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1015 EMSGSIZE, EHOSTUNREACH, 0, 0, 1016 0, 0, 0, 0, 1017 ENOPROTOOPT 1018}; 1019 1020/* 1021 * Forward a packet. If some error occurs return the sender 1022 * an icmp packet. Note we can't always generate a meaningful 1023 * icmp message because icmp doesn't have a large enough repertoire 1024 * of codes and types. 1025 * 1026 * If not forwarding, just drop the packet. This could be confusing 1027 * if ipforwarding was zero but some routing protocol was advancing 1028 * us as a gateway to somewhere. However, we must let the routing 1029 * protocol deal with that. 1030 * 1031 * The srcrt parameter indicates whether the packet is being forwarded 1032 * via a source route. 1033 */ 1034void 1035ip_forward(m, srcrt) 1036 struct mbuf *m; 1037 int srcrt; 1038{ 1039 register struct ip *ip = mtod(m, struct ip *); 1040 register struct sockaddr_in *sin; 1041 register struct rtentry *rt; 1042 int error, type = 0, code = 0; 1043 struct mbuf *mcopy; 1044 n_long dest; 1045 struct ifnet *destifp; 1046 1047 dest = 0; 1048#ifdef DIAGNOSTIC 1049 if (ipprintfs) 1050 printf("forward: src %x dst %x ttl %x\n", 1051 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1052#endif 1053 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1054 ipstat.ips_cantforward++; 1055 m_freem(m); 1056 return; 1057 } 1058 HTONS(ip->ip_id); 1059 if (ip->ip_ttl <= IPTTLDEC) { 1060 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1061 return; 1062 } 1063 ip->ip_ttl -= IPTTLDEC; 1064 1065 sin = satosin(&ipforward_rt.ro_dst); 1066 if ((rt = ipforward_rt.ro_rt) == 0 || 1067 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1068 if (ipforward_rt.ro_rt) { 1069 RTFREE(ipforward_rt.ro_rt); 1070 ipforward_rt.ro_rt = 0; 1071 } 1072 sin->sin_family = AF_INET; 1073 sin->sin_len = sizeof(struct sockaddr_in); 1074 sin->sin_addr = ip->ip_dst; 1075 1076 rtalloc(&ipforward_rt); 1077 if (ipforward_rt.ro_rt == 0) { 1078 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1079 return; 1080 } 1081 rt = ipforward_rt.ro_rt; 1082 } 1083 1084 /* 1085 * Save at most 68 bytes of the packet in case 1086 * we need to generate an ICMP message to the src. 1087 */ 1088 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1089 1090 /* 1091 * If forwarding packet using same interface that it came in on, 1092 * perhaps should send a redirect to sender to shortcut a hop. 1093 * Only send redirect if source is sending directly to us, 1094 * and if packet was not source routed (or has any options). 1095 * Also, don't send redirect if forwarding using a default route 1096 * or a route modified by a redirect. 1097 */ 1098 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1099 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1100 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1101 ipsendredirects && !srcrt) { 1102 if (rt->rt_ifa && 1103 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1104 ifatoia(rt->rt_ifa)->ia_subnet) { 1105 if (rt->rt_flags & RTF_GATEWAY) 1106 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1107 else 1108 dest = ip->ip_dst.s_addr; 1109 /* Router requirements says to only send host redirects */ 1110 type = ICMP_REDIRECT; 1111 code = ICMP_REDIRECT_HOST; 1112#ifdef DIAGNOSTIC 1113 if (ipprintfs) 1114 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1115#endif 1116 } 1117 } 1118 1119 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1120 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1121 if (error) 1122 ipstat.ips_cantforward++; 1123 else { 1124 ipstat.ips_forward++; 1125 if (type) 1126 ipstat.ips_redirectsent++; 1127 else { 1128 if (mcopy) 1129 m_freem(mcopy); 1130 return; 1131 } 1132 } 1133 if (mcopy == NULL) 1134 return; 1135 destifp = NULL; 1136 1137 switch (error) { 1138 1139 case 0: /* forwarded, but need redirect */ 1140 /* type, code set above */ 1141 break; 1142 1143 case ENETUNREACH: /* shouldn't happen, checked above */ 1144 case EHOSTUNREACH: 1145 case ENETDOWN: 1146 case EHOSTDOWN: 1147 default: 1148 type = ICMP_UNREACH; 1149 code = ICMP_UNREACH_HOST; 1150 break; 1151 1152 case EMSGSIZE: 1153 type = ICMP_UNREACH; 1154 code = ICMP_UNREACH_NEEDFRAG; 1155 if (ipforward_rt.ro_rt) 1156 destifp = ipforward_rt.ro_rt->rt_ifp; 1157 ipstat.ips_cantfrag++; 1158 break; 1159 1160 case ENOBUFS: 1161 type = ICMP_SOURCEQUENCH; 1162 code = 0; 1163 break; 1164 } 1165 icmp_error(mcopy, type, code, dest, destifp); 1166} 1167 1168void 1169ip_savecontrol(inp, mp, ip, m) 1170 register struct inpcb *inp; 1171 register struct mbuf **mp; 1172 register struct ip *ip; 1173 register struct mbuf *m; 1174{ 1175 1176 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1177 struct timeval tv; 1178 1179 microtime(&tv); 1180 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1181 SCM_TIMESTAMP, SOL_SOCKET); 1182 if (*mp) 1183 mp = &(*mp)->m_next; 1184 } 1185 if (inp->inp_flags & INP_RECVDSTADDR) { 1186 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1187 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1188 if (*mp) 1189 mp = &(*mp)->m_next; 1190 } 1191#ifdef notyet 1192 /* 1193 * XXX 1194 * Moving these out of udp_input() made them even more broken 1195 * than they already were. 1196 * - fenner@parc.xerox.com 1197 */ 1198 /* options were tossed already */ 1199 if (inp->inp_flags & INP_RECVOPTS) { 1200 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1201 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1202 if (*mp) 1203 mp = &(*mp)->m_next; 1204 } 1205 /* ip_srcroute doesn't do what we want here, need to fix */ 1206 if (inp->inp_flags & INP_RECVRETOPTS) { 1207 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1208 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1209 if (*mp) 1210 mp = &(*mp)->m_next; 1211 } 1212#endif 1213 if (inp->inp_flags & INP_RECVIF) { 1214 struct sockaddr_dl sdl; 1215 1216 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1217 sdl.sdl_family = AF_LINK; 1218 sdl.sdl_index = m->m_pkthdr.rcvif ? 1219 m->m_pkthdr.rcvif->if_index : 0; 1220 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1221 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1222 IP_RECVIF, IPPROTO_IP); 1223 if (*mp) 1224 mp = &(*mp)->m_next; 1225 } 1226} 1227 1228int 1229ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1230 int *name; 1231 u_int namelen; 1232 void *oldp; 1233 size_t *oldlenp; 1234 void *newp; 1235 size_t newlen; 1236{ 1237 /* All sysctl names at this level are terminal. */ 1238 if (namelen != 1) 1239 return (ENOTDIR); 1240 1241 switch (name[0]) { 1242 case IPCTL_FORWARDING: 1243 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1244 case IPCTL_SENDREDIRECTS: 1245 return (sysctl_int(oldp, oldlenp, newp, newlen, 1246 &ipsendredirects)); 1247 case IPCTL_DEFTTL: 1248 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1249#ifdef notyet 1250 case IPCTL_DEFMTU: 1251 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1252#endif 1253 case IPCTL_FORWSRCRT: 1254 /* 1255 * Don't allow this to change in a secure environment. 1256 */ 1257 if (securelevel > 0) 1258 return (EPERM); 1259 return (sysctl_int(oldp, oldlenp, newp, newlen, 1260 &ip_forwsrcrt)); 1261 case IPCTL_DIRECTEDBCAST: 1262 return (sysctl_int(oldp, oldlenp, newp, newlen, 1263 &ip_directedbcast)); 1264 default: 1265 return (EOPNOTSUPP); 1266 } 1267 /* NOTREACHED */ 1268} 1269