1 /* $NetBSD: ip_input.c,v 1.54 1998/01/05 09:52:04 lukem 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 #include <net/pfil.h> 58 59 #include <netinet/in.h> 60 #include <netinet/in_systm.h> 61 #include <netinet/ip.h> 62 #include <netinet/in_pcb.h> 63 #include <netinet/in_var.h> 64 #include <netinet/ip_var.h> 65 #include <netinet/ip_icmp.h> 66 67 /* XXX should really put this in libkern.h */ 68 #define offsetof(type, member) ((size_t)(&((type *)0)->member)) 69 70 #ifndef IPFORWARDING 71 #ifdef GATEWAY 72 #define IPFORWARDING 1 /* forward IP packets not for us */ 73 #else /* GATEWAY */ 74 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 75 #endif /* GATEWAY */ 76 #endif /* IPFORWARDING */ 77 #ifndef IPSENDREDIRECTS 78 #define IPSENDREDIRECTS 1 79 #endif 80 #ifndef IPFORWSRCRT 81 #define IPFORWSRCRT 1 /* forward source-routed packets */ 82 #endif 83 #ifndef IPALLOWSRCRT 84 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 85 #endif 86 #ifndef IPMTUDISC 87 #define IPMTUDISC 0 88 #endif 89 90 /* 91 * Note: DIRECTED_BROADCAST is handled this way so that previous 92 * configuration using this option will Just Work. 93 */ 94 #ifndef IPDIRECTEDBCAST 95 #ifdef DIRECTED_BROADCAST 96 #define IPDIRECTEDBCAST 1 97 #else 98 #define IPDIRECTEDBCAST 0 99 #endif /* DIRECTED_BROADCAST */ 100 #endif /* IPDIRECTEDBCAST */ 101 int ipforwarding = IPFORWARDING; 102 int ipsendredirects = IPSENDREDIRECTS; 103 int ip_defttl = IPDEFTTL; 104 int ip_forwsrcrt = IPFORWSRCRT; 105 int ip_directedbcast = IPDIRECTEDBCAST; 106 int ip_allowsrcrt = IPALLOWSRCRT; 107 int ip_mtudisc = IPMTUDISC; 108 #ifdef DIAGNOSTIC 109 int ipprintfs = 0; 110 #endif 111 112 extern struct domain inetdomain; 113 extern struct protosw inetsw[]; 114 u_char ip_protox[IPPROTO_MAX]; 115 int ipqmaxlen = IFQ_MAXLEN; 116 struct in_ifaddrhead in_ifaddr; 117 struct ifqueue ipintrq; 118 119 /* 120 * We need to save the IP options in case a protocol wants to respond 121 * to an incoming packet over the same route if the packet got here 122 * using IP source routing. This allows connection establishment and 123 * maintenance when the remote end is on a network that is not known 124 * to us. 125 */ 126 int ip_nhops = 0; 127 static struct ip_srcrt { 128 struct in_addr dst; /* final destination */ 129 char nop; /* one NOP to align */ 130 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 131 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 132 } ip_srcrt; 133 134 static void save_rte __P((u_char *, struct in_addr)); 135 136 /* 137 * IP initialization: fill in IP protocol switch table. 138 * All protocols not implemented in kernel go to raw IP protocol handler. 139 */ 140 void 141 ip_init() 142 { 143 register struct protosw *pr; 144 register int i; 145 146 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 147 if (pr == 0) 148 panic("ip_init"); 149 for (i = 0; i < IPPROTO_MAX; i++) 150 ip_protox[i] = pr - inetsw; 151 for (pr = inetdomain.dom_protosw; 152 pr < inetdomain.dom_protoswNPROTOSW; pr++) 153 if (pr->pr_domain->dom_family == PF_INET && 154 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 155 ip_protox[pr->pr_protocol] = pr - inetsw; 156 LIST_INIT(&ipq); 157 ip_id = time.tv_sec & 0xffff; 158 ipintrq.ifq_maxlen = ipqmaxlen; 159 TAILQ_INIT(&in_ifaddr); 160 } 161 162 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 163 struct route ipforward_rt; 164 165 /* 166 * Ip input routine. Checksum and byte swap header. If fragmented 167 * try to reassemble. Process options. Pass to next level. 168 */ 169 void 170 ipintr() 171 { 172 register struct ip *ip = NULL; 173 register struct mbuf *m; 174 register struct ipq *fp; 175 register struct in_ifaddr *ia; 176 struct ipqent *ipqe; 177 int hlen = 0, mff, len, s; 178 #ifdef PFIL_HOOKS 179 struct packet_filter_hook *pfh; 180 struct mbuf *m0; 181 int rv; 182 #endif /* PFIL_HOOKS */ 183 184 next: 185 /* 186 * Get next datagram off input queue and get IP header 187 * in first mbuf. 188 */ 189 s = splimp(); 190 IF_DEQUEUE(&ipintrq, m); 191 splx(s); 192 if (m == 0) 193 return; 194 #ifdef DIAGNOSTIC 195 if ((m->m_flags & M_PKTHDR) == 0) 196 panic("ipintr no HDR"); 197 #endif 198 /* 199 * If no IP addresses have been set yet but the interfaces 200 * are receiving, can't do anything with incoming packets yet. 201 */ 202 if (in_ifaddr.tqh_first == 0) 203 goto bad; 204 ipstat.ips_total++; 205 if (m->m_len < sizeof (struct ip) && 206 (m = m_pullup(m, sizeof (struct ip))) == 0) { 207 ipstat.ips_toosmall++; 208 goto next; 209 } 210 ip = mtod(m, struct ip *); 211 if (ip->ip_v != IPVERSION) { 212 ipstat.ips_badvers++; 213 goto bad; 214 } 215 hlen = ip->ip_hl << 2; 216 if (hlen < sizeof(struct ip)) { /* minimum header length */ 217 ipstat.ips_badhlen++; 218 goto bad; 219 } 220 if (hlen > m->m_len) { 221 if ((m = m_pullup(m, hlen)) == 0) { 222 ipstat.ips_badhlen++; 223 goto next; 224 } 225 ip = mtod(m, struct ip *); 226 } 227 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 228 ipstat.ips_badsum++; 229 goto bad; 230 } 231 232 /* 233 * Convert fields to host representation. 234 */ 235 NTOHS(ip->ip_len); 236 NTOHS(ip->ip_id); 237 NTOHS(ip->ip_off); 238 len = ip->ip_len; 239 240 /* 241 * Check that the amount of data in the buffers 242 * is as at least much as the IP header would have us expect. 243 * Trim mbufs if longer than we expect. 244 * Drop packet if shorter than we expect. 245 */ 246 if (m->m_pkthdr.len < len) { 247 ipstat.ips_tooshort++; 248 goto bad; 249 } 250 if (m->m_pkthdr.len > len) { 251 if (m->m_len == m->m_pkthdr.len) { 252 m->m_len = len; 253 m->m_pkthdr.len = len; 254 } else 255 m_adj(m, len - m->m_pkthdr.len); 256 } 257 258 #ifdef PFIL_HOOKS 259 /* 260 * Run through list of hooks for input packets. 261 */ 262 m0 = m; 263 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 264 if (pfh->pfil_func) { 265 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 266 if (rv) 267 goto next; 268 ip = mtod(m = m0, struct ip *); 269 } 270 #endif /* PFIL_HOOKS */ 271 272 /* 273 * Process options and, if not destined for us, 274 * ship it on. ip_dooptions returns 1 when an 275 * error was detected (causing an icmp message 276 * to be sent and the original packet to be freed). 277 */ 278 ip_nhops = 0; /* for source routed packets */ 279 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 280 goto next; 281 282 /* 283 * Check our list of addresses, to see if the packet is for us. 284 */ 285 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 286 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 287 goto ours; 288 if (((ip_directedbcast == 0) || (ip_directedbcast && 289 ia->ia_ifp == m->m_pkthdr.rcvif)) && 290 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 291 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 292 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 293 /* 294 * Look for all-0's host part (old broadcast addr), 295 * either for subnet or net. 296 */ 297 ip->ip_dst.s_addr == ia->ia_subnet || 298 ip->ip_dst.s_addr == ia->ia_net) 299 goto ours; 300 } 301 /* 302 * An interface with IP address zero accepts 303 * all packets that arrive on that interface. 304 */ 305 if ((ia->ia_ifp == m->m_pkthdr.rcvif) && 306 in_nullhost(ia->ia_addr.sin_addr)) 307 goto ours; 308 } 309 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 310 struct in_multi *inm; 311 #ifdef MROUTING 312 extern struct socket *ip_mrouter; 313 314 if (m->m_flags & M_EXT) { 315 if ((m = m_pullup(m, hlen)) == 0) { 316 ipstat.ips_toosmall++; 317 goto next; 318 } 319 ip = mtod(m, struct ip *); 320 } 321 322 if (ip_mrouter) { 323 /* 324 * If we are acting as a multicast router, all 325 * incoming multicast packets are passed to the 326 * kernel-level multicast forwarding function. 327 * The packet is returned (relatively) intact; if 328 * ip_mforward() returns a non-zero value, the packet 329 * must be discarded, else it may be accepted below. 330 * 331 * (The IP ident field is put in the same byte order 332 * as expected when ip_mforward() is called from 333 * ip_output().) 334 */ 335 ip->ip_id = htons(ip->ip_id); 336 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 337 ipstat.ips_cantforward++; 338 m_freem(m); 339 goto next; 340 } 341 ip->ip_id = ntohs(ip->ip_id); 342 343 /* 344 * The process-level routing demon needs to receive 345 * all multicast IGMP packets, whether or not this 346 * host belongs to their destination groups. 347 */ 348 if (ip->ip_p == IPPROTO_IGMP) 349 goto ours; 350 ipstat.ips_forward++; 351 } 352 #endif 353 /* 354 * See if we belong to the destination multicast group on the 355 * arrival interface. 356 */ 357 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 358 if (inm == NULL) { 359 ipstat.ips_cantforward++; 360 m_freem(m); 361 goto next; 362 } 363 goto ours; 364 } 365 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 366 in_nullhost(ip->ip_dst)) 367 goto ours; 368 369 /* 370 * Not for us; forward if possible and desirable. 371 */ 372 if (ipforwarding == 0) { 373 ipstat.ips_cantforward++; 374 m_freem(m); 375 } else 376 ip_forward(m, 0); 377 goto next; 378 379 ours: 380 /* 381 * If offset or IP_MF are set, must reassemble. 382 * Otherwise, nothing need be done. 383 * (We could look in the reassembly queue to see 384 * if the packet was previously fragmented, 385 * but it's not worth the time; just let them time out.) 386 */ 387 if (ip->ip_off & ~(IP_DF|IP_RF)) { 388 /* 389 * Look for queue of fragments 390 * of this datagram. 391 */ 392 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 393 if (ip->ip_id == fp->ipq_id && 394 in_hosteq(ip->ip_src, fp->ipq_src) && 395 in_hosteq(ip->ip_dst, fp->ipq_dst) && 396 ip->ip_p == fp->ipq_p) 397 goto found; 398 fp = 0; 399 found: 400 401 /* 402 * Adjust ip_len to not reflect header, 403 * set ipqe_mff if more fragments are expected, 404 * convert offset of this to bytes. 405 */ 406 ip->ip_len -= hlen; 407 mff = (ip->ip_off & IP_MF) != 0; 408 if (mff) { 409 /* 410 * Make sure that fragments have a data length 411 * that's a non-zero multiple of 8 bytes. 412 */ 413 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 414 ipstat.ips_badfrags++; 415 goto bad; 416 } 417 } 418 ip->ip_off <<= 3; 419 420 /* 421 * If datagram marked as having more fragments 422 * or if this is not the first fragment, 423 * attempt reassembly; if it succeeds, proceed. 424 */ 425 if (mff || ip->ip_off) { 426 ipstat.ips_fragments++; 427 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 428 M_IPQ, M_NOWAIT); 429 if (ipqe == NULL) { 430 ipstat.ips_rcvmemdrop++; 431 goto bad; 432 } 433 ipqe->ipqe_mff = mff; 434 ipqe->ipqe_m = m; 435 ipqe->ipqe_ip = ip; 436 m = ip_reass(ipqe, fp); 437 if (m == 0) 438 goto next; 439 ipstat.ips_reassembled++; 440 ip = mtod(m, struct ip *); 441 } else 442 if (fp) 443 ip_freef(fp); 444 } else 445 ip->ip_len -= hlen; 446 447 /* 448 * Switch out to protocol's input routine. 449 */ 450 ipstat.ips_delivered++; 451 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 452 goto next; 453 bad: 454 m_freem(m); 455 goto next; 456 } 457 458 /* 459 * Take incoming datagram fragment and try to 460 * reassemble it into whole datagram. If a chain for 461 * reassembly of this datagram already exists, then it 462 * is given as fp; otherwise have to make a chain. 463 */ 464 struct mbuf * 465 ip_reass(ipqe, fp) 466 register struct ipqent *ipqe; 467 register struct ipq *fp; 468 { 469 register struct mbuf *m = ipqe->ipqe_m; 470 register struct ipqent *nq, *p, *q; 471 struct ip *ip; 472 struct mbuf *t; 473 int hlen = ipqe->ipqe_ip->ip_hl << 2; 474 int i, next; 475 476 /* 477 * Presence of header sizes in mbufs 478 * would confuse code below. 479 */ 480 m->m_data += hlen; 481 m->m_len -= hlen; 482 483 /* 484 * If first fragment to arrive, create a reassembly queue. 485 */ 486 if (fp == 0) { 487 MALLOC(fp, struct ipq *, sizeof (struct ipq), 488 M_FTABLE, M_NOWAIT); 489 if (fp == NULL) 490 goto dropfrag; 491 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 492 fp->ipq_ttl = IPFRAGTTL; 493 fp->ipq_p = ipqe->ipqe_ip->ip_p; 494 fp->ipq_id = ipqe->ipqe_ip->ip_id; 495 LIST_INIT(&fp->ipq_fragq); 496 fp->ipq_src = ipqe->ipqe_ip->ip_src; 497 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 498 p = NULL; 499 goto insert; 500 } 501 502 /* 503 * Find a segment which begins after this one does. 504 */ 505 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 506 p = q, q = q->ipqe_q.le_next) 507 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 508 break; 509 510 /* 511 * If there is a preceding segment, it may provide some of 512 * our data already. If so, drop the data from the incoming 513 * segment. If it provides all of our data, drop us. 514 */ 515 if (p != NULL) { 516 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 517 ipqe->ipqe_ip->ip_off; 518 if (i > 0) { 519 if (i >= ipqe->ipqe_ip->ip_len) 520 goto dropfrag; 521 m_adj(ipqe->ipqe_m, i); 522 ipqe->ipqe_ip->ip_off += i; 523 ipqe->ipqe_ip->ip_len -= i; 524 } 525 } 526 527 /* 528 * While we overlap succeeding segments trim them or, 529 * if they are completely covered, dequeue them. 530 */ 531 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 532 q->ipqe_ip->ip_off; q = nq) { 533 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 534 q->ipqe_ip->ip_off; 535 if (i < q->ipqe_ip->ip_len) { 536 q->ipqe_ip->ip_len -= i; 537 q->ipqe_ip->ip_off += i; 538 m_adj(q->ipqe_m, i); 539 break; 540 } 541 nq = q->ipqe_q.le_next; 542 m_freem(q->ipqe_m); 543 LIST_REMOVE(q, ipqe_q); 544 FREE(q, M_IPQ); 545 } 546 547 insert: 548 /* 549 * Stick new segment in its place; 550 * check for complete reassembly. 551 */ 552 if (p == NULL) { 553 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 554 } else { 555 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 556 } 557 next = 0; 558 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 559 p = q, q = q->ipqe_q.le_next) { 560 if (q->ipqe_ip->ip_off != next) 561 return (0); 562 next += q->ipqe_ip->ip_len; 563 } 564 if (p->ipqe_mff) 565 return (0); 566 567 /* 568 * Reassembly is complete. Check for a bogus message size and 569 * concatenate fragments. 570 */ 571 q = fp->ipq_fragq.lh_first; 572 ip = q->ipqe_ip; 573 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 574 ipstat.ips_toolong++; 575 ip_freef(fp); 576 return (0); 577 } 578 m = q->ipqe_m; 579 t = m->m_next; 580 m->m_next = 0; 581 m_cat(m, t); 582 nq = q->ipqe_q.le_next; 583 FREE(q, M_IPQ); 584 for (q = nq; q != NULL; q = nq) { 585 t = q->ipqe_m; 586 nq = q->ipqe_q.le_next; 587 FREE(q, M_IPQ); 588 m_cat(m, t); 589 } 590 591 /* 592 * Create header for new ip packet by 593 * modifying header of first packet; 594 * dequeue and discard fragment reassembly header. 595 * Make header visible. 596 */ 597 ip->ip_len = next; 598 ip->ip_src = fp->ipq_src; 599 ip->ip_dst = fp->ipq_dst; 600 LIST_REMOVE(fp, ipq_q); 601 FREE(fp, M_FTABLE); 602 m->m_len += (ip->ip_hl << 2); 603 m->m_data -= (ip->ip_hl << 2); 604 /* some debugging cruft by sklower, below, will go away soon */ 605 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 606 register int plen = 0; 607 for (t = m; t; t = t->m_next) 608 plen += t->m_len; 609 m->m_pkthdr.len = plen; 610 } 611 return (m); 612 613 dropfrag: 614 ipstat.ips_fragdropped++; 615 m_freem(m); 616 FREE(ipqe, M_IPQ); 617 return (0); 618 } 619 620 /* 621 * Free a fragment reassembly header and all 622 * associated datagrams. 623 */ 624 void 625 ip_freef(fp) 626 struct ipq *fp; 627 { 628 register struct ipqent *q, *p; 629 630 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 631 p = q->ipqe_q.le_next; 632 m_freem(q->ipqe_m); 633 LIST_REMOVE(q, ipqe_q); 634 FREE(q, M_IPQ); 635 } 636 LIST_REMOVE(fp, ipq_q); 637 FREE(fp, M_FTABLE); 638 } 639 640 /* 641 * IP timer processing; 642 * if a timer expires on a reassembly 643 * queue, discard it. 644 */ 645 void 646 ip_slowtimo() 647 { 648 register struct ipq *fp, *nfp; 649 int s = splsoftnet(); 650 651 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 652 nfp = fp->ipq_q.le_next; 653 if (--fp->ipq_ttl == 0) { 654 ipstat.ips_fragtimeout++; 655 ip_freef(fp); 656 } 657 } 658 splx(s); 659 } 660 661 /* 662 * Drain off all datagram fragments. 663 */ 664 void 665 ip_drain() 666 { 667 668 while (ipq.lh_first != NULL) { 669 ipstat.ips_fragdropped++; 670 ip_freef(ipq.lh_first); 671 } 672 } 673 674 /* 675 * Do option processing on a datagram, 676 * possibly discarding it if bad options are encountered, 677 * or forwarding it if source-routed. 678 * Returns 1 if packet has been forwarded/freed, 679 * 0 if the packet should be processed further. 680 */ 681 int 682 ip_dooptions(m) 683 struct mbuf *m; 684 { 685 register struct ip *ip = mtod(m, struct ip *); 686 register u_char *cp; 687 register struct ip_timestamp *ipt; 688 register struct in_ifaddr *ia; 689 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 690 struct in_addr *sin, dst; 691 n_time ntime; 692 693 dst = ip->ip_dst; 694 cp = (u_char *)(ip + 1); 695 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 696 for (; cnt > 0; cnt -= optlen, cp += optlen) { 697 opt = cp[IPOPT_OPTVAL]; 698 if (opt == IPOPT_EOL) 699 break; 700 if (opt == IPOPT_NOP) 701 optlen = 1; 702 else { 703 optlen = cp[IPOPT_OLEN]; 704 if (optlen <= 0 || optlen > cnt) { 705 code = &cp[IPOPT_OLEN] - (u_char *)ip; 706 goto bad; 707 } 708 } 709 switch (opt) { 710 711 default: 712 break; 713 714 /* 715 * Source routing with record. 716 * Find interface with current destination address. 717 * If none on this machine then drop if strictly routed, 718 * or do nothing if loosely routed. 719 * Record interface address and bring up next address 720 * component. If strictly routed make sure next 721 * address is on directly accessible net. 722 */ 723 case IPOPT_LSRR: 724 case IPOPT_SSRR: 725 if (ip_allowsrcrt == 0) { 726 type = ICMP_UNREACH; 727 code = ICMP_UNREACH_NET_PROHIB; 728 goto bad; 729 } 730 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 731 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 732 goto bad; 733 } 734 ipaddr.sin_addr = ip->ip_dst; 735 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 736 if (ia == 0) { 737 if (opt == IPOPT_SSRR) { 738 type = ICMP_UNREACH; 739 code = ICMP_UNREACH_SRCFAIL; 740 goto bad; 741 } 742 /* 743 * Loose routing, and not at next destination 744 * yet; nothing to do except forward. 745 */ 746 break; 747 } 748 off--; /* 0 origin */ 749 if (off > optlen - sizeof(struct in_addr)) { 750 /* 751 * End of source route. Should be for us. 752 */ 753 save_rte(cp, ip->ip_src); 754 break; 755 } 756 /* 757 * locate outgoing interface 758 */ 759 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 760 sizeof(ipaddr.sin_addr)); 761 if (opt == IPOPT_SSRR) { 762 #define INA struct in_ifaddr * 763 #define SA struct sockaddr * 764 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 765 } else 766 ia = ip_rtaddr(ipaddr.sin_addr); 767 if (ia == 0) { 768 type = ICMP_UNREACH; 769 code = ICMP_UNREACH_SRCFAIL; 770 goto bad; 771 } 772 ip->ip_dst = ipaddr.sin_addr; 773 bcopy((caddr_t)&ia->ia_addr.sin_addr, 774 (caddr_t)(cp + off), sizeof(struct in_addr)); 775 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 776 /* 777 * Let ip_intr's mcast routing check handle mcast pkts 778 */ 779 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 780 break; 781 782 case IPOPT_RR: 783 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 784 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 785 goto bad; 786 } 787 /* 788 * If no space remains, ignore. 789 */ 790 off--; /* 0 origin */ 791 if (off > optlen - sizeof(struct in_addr)) 792 break; 793 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 794 sizeof(ipaddr.sin_addr)); 795 /* 796 * locate outgoing interface; if we're the destination, 797 * use the incoming interface (should be same). 798 */ 799 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 800 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 801 type = ICMP_UNREACH; 802 code = ICMP_UNREACH_HOST; 803 goto bad; 804 } 805 bcopy((caddr_t)&ia->ia_addr.sin_addr, 806 (caddr_t)(cp + off), sizeof(struct in_addr)); 807 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 808 break; 809 810 case IPOPT_TS: 811 code = cp - (u_char *)ip; 812 ipt = (struct ip_timestamp *)cp; 813 if (ipt->ipt_len < 5) 814 goto bad; 815 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 816 if (++ipt->ipt_oflw == 0) 817 goto bad; 818 break; 819 } 820 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 821 switch (ipt->ipt_flg) { 822 823 case IPOPT_TS_TSONLY: 824 break; 825 826 case IPOPT_TS_TSANDADDR: 827 if (ipt->ipt_ptr + sizeof(n_time) + 828 sizeof(struct in_addr) > ipt->ipt_len) 829 goto bad; 830 ipaddr.sin_addr = dst; 831 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 832 m->m_pkthdr.rcvif); 833 if (ia == 0) 834 continue; 835 bcopy((caddr_t)&ia->ia_addr.sin_addr, 836 (caddr_t)sin, sizeof(struct in_addr)); 837 ipt->ipt_ptr += sizeof(struct in_addr); 838 break; 839 840 case IPOPT_TS_PRESPEC: 841 if (ipt->ipt_ptr + sizeof(n_time) + 842 sizeof(struct in_addr) > ipt->ipt_len) 843 goto bad; 844 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 845 sizeof(struct in_addr)); 846 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 847 continue; 848 ipt->ipt_ptr += sizeof(struct in_addr); 849 break; 850 851 default: 852 goto bad; 853 } 854 ntime = iptime(); 855 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 856 sizeof(n_time)); 857 ipt->ipt_ptr += sizeof(n_time); 858 } 859 } 860 if (forward) { 861 if (ip_forwsrcrt == 0) { 862 type = ICMP_UNREACH; 863 code = ICMP_UNREACH_SRCFAIL; 864 goto bad; 865 } 866 ip_forward(m, 1); 867 return (1); 868 } 869 return (0); 870 bad: 871 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 872 icmp_error(m, type, code, 0, 0); 873 ipstat.ips_badoptions++; 874 return (1); 875 } 876 877 /* 878 * Given address of next destination (final or next hop), 879 * return internet address info of interface to be used to get there. 880 */ 881 struct in_ifaddr * 882 ip_rtaddr(dst) 883 struct in_addr dst; 884 { 885 register struct sockaddr_in *sin; 886 887 sin = satosin(&ipforward_rt.ro_dst); 888 889 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 890 if (ipforward_rt.ro_rt) { 891 RTFREE(ipforward_rt.ro_rt); 892 ipforward_rt.ro_rt = 0; 893 } 894 sin->sin_family = AF_INET; 895 sin->sin_len = sizeof(*sin); 896 sin->sin_addr = dst; 897 898 rtalloc(&ipforward_rt); 899 } 900 if (ipforward_rt.ro_rt == 0) 901 return ((struct in_ifaddr *)0); 902 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 903 } 904 905 /* 906 * Save incoming source route for use in replies, 907 * to be picked up later by ip_srcroute if the receiver is interested. 908 */ 909 void 910 save_rte(option, dst) 911 u_char *option; 912 struct in_addr dst; 913 { 914 unsigned olen; 915 916 olen = option[IPOPT_OLEN]; 917 #ifdef DIAGNOSTIC 918 if (ipprintfs) 919 printf("save_rte: olen %d\n", olen); 920 #endif 921 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 922 return; 923 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 924 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 925 ip_srcrt.dst = dst; 926 } 927 928 /* 929 * Retrieve incoming source route for use in replies, 930 * in the same form used by setsockopt. 931 * The first hop is placed before the options, will be removed later. 932 */ 933 struct mbuf * 934 ip_srcroute() 935 { 936 register struct in_addr *p, *q; 937 register struct mbuf *m; 938 939 if (ip_nhops == 0) 940 return ((struct mbuf *)0); 941 m = m_get(M_DONTWAIT, MT_SOOPTS); 942 if (m == 0) 943 return ((struct mbuf *)0); 944 945 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 946 947 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 948 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 949 OPTSIZ; 950 #ifdef DIAGNOSTIC 951 if (ipprintfs) 952 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 953 #endif 954 955 /* 956 * First save first hop for return route 957 */ 958 p = &ip_srcrt.route[ip_nhops - 1]; 959 *(mtod(m, struct in_addr *)) = *p--; 960 #ifdef DIAGNOSTIC 961 if (ipprintfs) 962 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 963 #endif 964 965 /* 966 * Copy option fields and padding (nop) to mbuf. 967 */ 968 ip_srcrt.nop = IPOPT_NOP; 969 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 970 bcopy((caddr_t)&ip_srcrt.nop, 971 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 972 q = (struct in_addr *)(mtod(m, caddr_t) + 973 sizeof(struct in_addr) + OPTSIZ); 974 #undef OPTSIZ 975 /* 976 * Record return path as an IP source route, 977 * reversing the path (pointers are now aligned). 978 */ 979 while (p >= ip_srcrt.route) { 980 #ifdef DIAGNOSTIC 981 if (ipprintfs) 982 printf(" %x", ntohl(q->s_addr)); 983 #endif 984 *q++ = *p--; 985 } 986 /* 987 * Last hop goes to final destination. 988 */ 989 *q = ip_srcrt.dst; 990 #ifdef DIAGNOSTIC 991 if (ipprintfs) 992 printf(" %x\n", ntohl(q->s_addr)); 993 #endif 994 return (m); 995 } 996 997 /* 998 * Strip out IP options, at higher 999 * level protocol in the kernel. 1000 * Second argument is buffer to which options 1001 * will be moved, and return value is their length. 1002 * XXX should be deleted; last arg currently ignored. 1003 */ 1004 void 1005 ip_stripoptions(m, mopt) 1006 register struct mbuf *m; 1007 struct mbuf *mopt; 1008 { 1009 register int i; 1010 struct ip *ip = mtod(m, struct ip *); 1011 register caddr_t opts; 1012 int olen; 1013 1014 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1015 opts = (caddr_t)(ip + 1); 1016 i = m->m_len - (sizeof (struct ip) + olen); 1017 bcopy(opts + olen, opts, (unsigned)i); 1018 m->m_len -= olen; 1019 if (m->m_flags & M_PKTHDR) 1020 m->m_pkthdr.len -= olen; 1021 ip->ip_hl = sizeof(struct ip) >> 2; 1022 } 1023 1024 int inetctlerrmap[PRC_NCMDS] = { 1025 0, 0, 0, 0, 1026 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1027 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1028 EMSGSIZE, EHOSTUNREACH, 0, 0, 1029 0, 0, 0, 0, 1030 ENOPROTOOPT 1031 }; 1032 1033 /* 1034 * Forward a packet. If some error occurs return the sender 1035 * an icmp packet. Note we can't always generate a meaningful 1036 * icmp message because icmp doesn't have a large enough repertoire 1037 * of codes and types. 1038 * 1039 * If not forwarding, just drop the packet. This could be confusing 1040 * if ipforwarding was zero but some routing protocol was advancing 1041 * us as a gateway to somewhere. However, we must let the routing 1042 * protocol deal with that. 1043 * 1044 * The srcrt parameter indicates whether the packet is being forwarded 1045 * via a source route. 1046 */ 1047 void 1048 ip_forward(m, srcrt) 1049 struct mbuf *m; 1050 int srcrt; 1051 { 1052 register struct ip *ip = mtod(m, struct ip *); 1053 register struct sockaddr_in *sin; 1054 register struct rtentry *rt; 1055 int error, type = 0, code = 0; 1056 struct mbuf *mcopy; 1057 n_long dest; 1058 struct ifnet *destifp; 1059 1060 dest = 0; 1061 #ifdef DIAGNOSTIC 1062 if (ipprintfs) 1063 printf("forward: src %x dst %x ttl %x\n", 1064 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1065 #endif 1066 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1067 ipstat.ips_cantforward++; 1068 m_freem(m); 1069 return; 1070 } 1071 HTONS(ip->ip_id); 1072 if (ip->ip_ttl <= IPTTLDEC) { 1073 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1074 return; 1075 } 1076 ip->ip_ttl -= IPTTLDEC; 1077 1078 sin = satosin(&ipforward_rt.ro_dst); 1079 if ((rt = ipforward_rt.ro_rt) == 0 || 1080 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1081 if (ipforward_rt.ro_rt) { 1082 RTFREE(ipforward_rt.ro_rt); 1083 ipforward_rt.ro_rt = 0; 1084 } 1085 sin->sin_family = AF_INET; 1086 sin->sin_len = sizeof(struct sockaddr_in); 1087 sin->sin_addr = ip->ip_dst; 1088 1089 rtalloc(&ipforward_rt); 1090 if (ipforward_rt.ro_rt == 0) { 1091 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1092 return; 1093 } 1094 rt = ipforward_rt.ro_rt; 1095 } 1096 1097 /* 1098 * Save at most 68 bytes of the packet in case 1099 * we need to generate an ICMP message to the src. 1100 */ 1101 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1102 1103 /* 1104 * If forwarding packet using same interface that it came in on, 1105 * perhaps should send a redirect to sender to shortcut a hop. 1106 * Only send redirect if source is sending directly to us, 1107 * and if packet was not source routed (or has any options). 1108 * Also, don't send redirect if forwarding using a default route 1109 * or a route modified by a redirect. 1110 */ 1111 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1112 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1113 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1114 ipsendredirects && !srcrt) { 1115 if (rt->rt_ifa && 1116 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1117 ifatoia(rt->rt_ifa)->ia_subnet) { 1118 if (rt->rt_flags & RTF_GATEWAY) 1119 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1120 else 1121 dest = ip->ip_dst.s_addr; 1122 /* Router requirements says to only send host redirects */ 1123 type = ICMP_REDIRECT; 1124 code = ICMP_REDIRECT_HOST; 1125 #ifdef DIAGNOSTIC 1126 if (ipprintfs) 1127 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1128 #endif 1129 } 1130 } 1131 1132 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1133 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1134 if (error) 1135 ipstat.ips_cantforward++; 1136 else { 1137 ipstat.ips_forward++; 1138 if (type) 1139 ipstat.ips_redirectsent++; 1140 else { 1141 if (mcopy) 1142 m_freem(mcopy); 1143 return; 1144 } 1145 } 1146 if (mcopy == NULL) 1147 return; 1148 destifp = NULL; 1149 1150 switch (error) { 1151 1152 case 0: /* forwarded, but need redirect */ 1153 /* type, code set above */ 1154 break; 1155 1156 case ENETUNREACH: /* shouldn't happen, checked above */ 1157 case EHOSTUNREACH: 1158 case ENETDOWN: 1159 case EHOSTDOWN: 1160 default: 1161 type = ICMP_UNREACH; 1162 code = ICMP_UNREACH_HOST; 1163 break; 1164 1165 case EMSGSIZE: 1166 type = ICMP_UNREACH; 1167 code = ICMP_UNREACH_NEEDFRAG; 1168 if (ipforward_rt.ro_rt) 1169 destifp = ipforward_rt.ro_rt->rt_ifp; 1170 ipstat.ips_cantfrag++; 1171 break; 1172 1173 case ENOBUFS: 1174 type = ICMP_SOURCEQUENCH; 1175 code = 0; 1176 break; 1177 } 1178 icmp_error(mcopy, type, code, dest, destifp); 1179 } 1180 1181 void 1182 ip_savecontrol(inp, mp, ip, m) 1183 register struct inpcb *inp; 1184 register struct mbuf **mp; 1185 register struct ip *ip; 1186 register struct mbuf *m; 1187 { 1188 1189 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1190 struct timeval tv; 1191 1192 microtime(&tv); 1193 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1194 SCM_TIMESTAMP, SOL_SOCKET); 1195 if (*mp) 1196 mp = &(*mp)->m_next; 1197 } 1198 if (inp->inp_flags & INP_RECVDSTADDR) { 1199 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1200 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1201 if (*mp) 1202 mp = &(*mp)->m_next; 1203 } 1204 #ifdef notyet 1205 /* 1206 * XXX 1207 * Moving these out of udp_input() made them even more broken 1208 * than they already were. 1209 * - fenner@parc.xerox.com 1210 */ 1211 /* options were tossed already */ 1212 if (inp->inp_flags & INP_RECVOPTS) { 1213 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1214 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1215 if (*mp) 1216 mp = &(*mp)->m_next; 1217 } 1218 /* ip_srcroute doesn't do what we want here, need to fix */ 1219 if (inp->inp_flags & INP_RECVRETOPTS) { 1220 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1221 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1222 if (*mp) 1223 mp = &(*mp)->m_next; 1224 } 1225 #endif 1226 if (inp->inp_flags & INP_RECVIF) { 1227 struct sockaddr_dl sdl; 1228 1229 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1230 sdl.sdl_family = AF_LINK; 1231 sdl.sdl_index = m->m_pkthdr.rcvif ? 1232 m->m_pkthdr.rcvif->if_index : 0; 1233 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1234 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1235 IP_RECVIF, IPPROTO_IP); 1236 if (*mp) 1237 mp = &(*mp)->m_next; 1238 } 1239 } 1240 1241 int 1242 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1243 int *name; 1244 u_int namelen; 1245 void *oldp; 1246 size_t *oldlenp; 1247 void *newp; 1248 size_t newlen; 1249 { 1250 extern int subnetsarelocal; 1251 1252 int error, old; 1253 1254 /* All sysctl names at this level are terminal. */ 1255 if (namelen != 1) 1256 return (ENOTDIR); 1257 1258 switch (name[0]) { 1259 case IPCTL_FORWARDING: 1260 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1261 case IPCTL_SENDREDIRECTS: 1262 return (sysctl_int(oldp, oldlenp, newp, newlen, 1263 &ipsendredirects)); 1264 case IPCTL_DEFTTL: 1265 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1266 #ifdef notyet 1267 case IPCTL_DEFMTU: 1268 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1269 #endif 1270 case IPCTL_FORWSRCRT: 1271 /* Don't allow this to change in a secure environment. */ 1272 if (securelevel > 0) 1273 return (sysctl_rdint(oldp, oldlenp, newp, 1274 ip_forwsrcrt)); 1275 else 1276 return (sysctl_int(oldp, oldlenp, newp, newlen, 1277 &ip_forwsrcrt)); 1278 case IPCTL_DIRECTEDBCAST: 1279 return (sysctl_int(oldp, oldlenp, newp, newlen, 1280 &ip_directedbcast)); 1281 case IPCTL_ALLOWSRCRT: 1282 return (sysctl_int(oldp, oldlenp, newp, newlen, 1283 &ip_allowsrcrt)); 1284 case IPCTL_SUBNETSARELOCAL: 1285 return (sysctl_int(oldp, oldlenp, newp, newlen, 1286 &subnetsarelocal)); 1287 case IPCTL_MTUDISC: 1288 return (sysctl_int(oldp, oldlenp, newp, newlen, 1289 &ip_mtudisc)); 1290 case IPCTL_ANONPORTMIN: 1291 old = anonportmin; 1292 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1293 if (anonportmin >= anonportmax || anonportmin > 65535 1294 #ifndef IPNOPRIVPORTS 1295 || anonportmin < IPPORT_RESERVED 1296 #endif 1297 ) { 1298 anonportmin = old; 1299 return (EINVAL); 1300 } 1301 return (error); 1302 case IPCTL_ANONPORTMAX: 1303 old = anonportmax; 1304 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1305 if (anonportmin >= anonportmax || anonportmax > 65535 1306 #ifndef IPNOPRIVPORTS 1307 || anonportmax < IPPORT_RESERVED 1308 #endif 1309 ) { 1310 anonportmax = old; 1311 return (EINVAL); 1312 } 1313 return (error); 1314 default: 1315 return (EOPNOTSUPP); 1316 } 1317 /* NOTREACHED */ 1318 } 1319