1 /* $NetBSD: ip_input.c,v 1.51 1997/08/29 16:02:42 gwr 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 /* 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 */ 97 int ipforwarding = IPFORWARDING; 98 int ipsendredirects = IPSENDREDIRECTS; 99 int ip_defttl = IPDEFTTL; 100 int ip_forwsrcrt = IPFORWSRCRT; 101 int ip_directedbcast = IPDIRECTEDBCAST; 102 int ip_allowsrcrt = IPALLOWSRCRT; 103 #ifdef DIAGNOSTIC 104 int ipprintfs = 0; 105 #endif 106 107 extern struct domain inetdomain; 108 extern struct protosw inetsw[]; 109 u_char ip_protox[IPPROTO_MAX]; 110 int ipqmaxlen = IFQ_MAXLEN; 111 struct in_ifaddrhead in_ifaddr; 112 struct ifqueue ipintrq; 113 114 /* 115 * We need to save the IP options in case a protocol wants to respond 116 * to an incoming packet over the same route if the packet got here 117 * using IP source routing. This allows connection establishment and 118 * maintenance when the remote end is on a network that is not known 119 * to us. 120 */ 121 int ip_nhops = 0; 122 static struct ip_srcrt { 123 struct in_addr dst; /* final destination */ 124 char nop; /* one NOP to align */ 125 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 126 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 127 } ip_srcrt; 128 129 static void save_rte __P((u_char *, struct in_addr)); 130 131 /* 132 * IP initialization: fill in IP protocol switch table. 133 * All protocols not implemented in kernel go to raw IP protocol handler. 134 */ 135 void 136 ip_init() 137 { 138 register struct protosw *pr; 139 register int i; 140 141 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 142 if (pr == 0) 143 panic("ip_init"); 144 for (i = 0; i < IPPROTO_MAX; i++) 145 ip_protox[i] = pr - inetsw; 146 for (pr = inetdomain.dom_protosw; 147 pr < inetdomain.dom_protoswNPROTOSW; pr++) 148 if (pr->pr_domain->dom_family == PF_INET && 149 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 150 ip_protox[pr->pr_protocol] = pr - inetsw; 151 LIST_INIT(&ipq); 152 ip_id = time.tv_sec & 0xffff; 153 ipintrq.ifq_maxlen = ipqmaxlen; 154 TAILQ_INIT(&in_ifaddr); 155 } 156 157 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 158 struct route ipforward_rt; 159 160 /* 161 * Ip input routine. Checksum and byte swap header. If fragmented 162 * try to reassemble. Process options. Pass to next level. 163 */ 164 void 165 ipintr() 166 { 167 register struct ip *ip = NULL; 168 register struct mbuf *m; 169 register struct ipq *fp; 170 register struct in_ifaddr *ia; 171 struct ipqent *ipqe; 172 int hlen = 0, mff, len, s; 173 #ifdef PFIL_HOOKS 174 struct packet_filter_hook *pfh; 175 struct mbuf *m0; 176 int rv; 177 #endif /* PFIL_HOOKS */ 178 179 next: 180 /* 181 * Get next datagram off input queue and get IP header 182 * in first mbuf. 183 */ 184 s = splimp(); 185 IF_DEQUEUE(&ipintrq, m); 186 splx(s); 187 if (m == 0) 188 return; 189 #ifdef DIAGNOSTIC 190 if ((m->m_flags & M_PKTHDR) == 0) 191 panic("ipintr no HDR"); 192 #endif 193 /* 194 * If no IP addresses have been set yet but the interfaces 195 * are receiving, can't do anything with incoming packets yet. 196 */ 197 if (in_ifaddr.tqh_first == 0) 198 goto bad; 199 ipstat.ips_total++; 200 if (m->m_len < sizeof (struct ip) && 201 (m = m_pullup(m, sizeof (struct ip))) == 0) { 202 ipstat.ips_toosmall++; 203 goto next; 204 } 205 ip = mtod(m, struct ip *); 206 if (ip->ip_v != IPVERSION) { 207 ipstat.ips_badvers++; 208 goto bad; 209 } 210 hlen = ip->ip_hl << 2; 211 if (hlen < sizeof(struct ip)) { /* minimum header length */ 212 ipstat.ips_badhlen++; 213 goto bad; 214 } 215 if (hlen > m->m_len) { 216 if ((m = m_pullup(m, hlen)) == 0) { 217 ipstat.ips_badhlen++; 218 goto next; 219 } 220 ip = mtod(m, struct ip *); 221 } 222 if ((ip->ip_sum = in_cksum(m, hlen)) != 0) { 223 ipstat.ips_badsum++; 224 goto bad; 225 } 226 227 /* 228 * Convert fields to host representation. 229 */ 230 NTOHS(ip->ip_len); 231 NTOHS(ip->ip_id); 232 NTOHS(ip->ip_off); 233 len = ip->ip_len; 234 235 /* 236 * Check that the amount of data in the buffers 237 * is as at least much as the IP header would have us expect. 238 * Trim mbufs if longer than we expect. 239 * Drop packet if shorter than we expect. 240 */ 241 if (m->m_pkthdr.len < len) { 242 ipstat.ips_tooshort++; 243 goto bad; 244 } 245 if (m->m_pkthdr.len > len) { 246 if (m->m_len == m->m_pkthdr.len) { 247 m->m_len = len; 248 m->m_pkthdr.len = len; 249 } else 250 m_adj(m, len - m->m_pkthdr.len); 251 } 252 253 #ifdef PFIL_HOOKS 254 /* 255 * Run through list of hooks for input packets. 256 */ 257 m0 = m; 258 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.le_next) 259 if (pfh->pfil_func) { 260 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 261 if (rv) 262 goto next; 263 ip = mtod(m = m0, struct ip *); 264 } 265 #endif /* PFIL_HOOKS */ 266 267 /* 268 * Process options and, if not destined for us, 269 * ship it on. ip_dooptions returns 1 when an 270 * error was detected (causing an icmp message 271 * to be sent and the original packet to be freed). 272 */ 273 ip_nhops = 0; /* for source routed packets */ 274 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 275 goto next; 276 277 /* 278 * Check our list of addresses, to see if the packet is for us. 279 */ 280 for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) { 281 if (in_hosteq(ip->ip_dst, ia->ia_addr.sin_addr)) 282 goto ours; 283 if (((ip_directedbcast == 0) || (ip_directedbcast && 284 ia->ia_ifp == m->m_pkthdr.rcvif)) && 285 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 286 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 287 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 288 /* 289 * Look for all-0's host part (old broadcast addr), 290 * either for subnet or net. 291 */ 292 ip->ip_dst.s_addr == ia->ia_subnet || 293 ip->ip_dst.s_addr == ia->ia_net) 294 goto ours; 295 } 296 /* 297 * An interface with IP address zero accepts 298 * all packets that arrive on that interface. 299 */ 300 if ((ia->ia_ifp == m->m_pkthdr.rcvif) && 301 in_nullhost(ia->ia_addr.sin_addr)) 302 goto ours; 303 } 304 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 305 struct in_multi *inm; 306 #ifdef MROUTING 307 extern struct socket *ip_mrouter; 308 309 if (m->m_flags & M_EXT) { 310 if ((m = m_pullup(m, hlen)) == 0) { 311 ipstat.ips_toosmall++; 312 goto next; 313 } 314 ip = mtod(m, struct ip *); 315 } 316 317 if (ip_mrouter) { 318 /* 319 * If we are acting as a multicast router, all 320 * incoming multicast packets are passed to the 321 * kernel-level multicast forwarding function. 322 * The packet is returned (relatively) intact; if 323 * ip_mforward() returns a non-zero value, the packet 324 * must be discarded, else it may be accepted below. 325 * 326 * (The IP ident field is put in the same byte order 327 * as expected when ip_mforward() is called from 328 * ip_output().) 329 */ 330 ip->ip_id = htons(ip->ip_id); 331 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 332 ipstat.ips_cantforward++; 333 m_freem(m); 334 goto next; 335 } 336 ip->ip_id = ntohs(ip->ip_id); 337 338 /* 339 * The process-level routing demon needs to receive 340 * all multicast IGMP packets, whether or not this 341 * host belongs to their destination groups. 342 */ 343 if (ip->ip_p == IPPROTO_IGMP) 344 goto ours; 345 ipstat.ips_forward++; 346 } 347 #endif 348 /* 349 * See if we belong to the destination multicast group on the 350 * arrival interface. 351 */ 352 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 353 if (inm == NULL) { 354 ipstat.ips_cantforward++; 355 m_freem(m); 356 goto next; 357 } 358 goto ours; 359 } 360 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 361 in_nullhost(ip->ip_dst)) 362 goto ours; 363 364 /* 365 * Not for us; forward if possible and desirable. 366 */ 367 if (ipforwarding == 0) { 368 ipstat.ips_cantforward++; 369 m_freem(m); 370 } else 371 ip_forward(m, 0); 372 goto next; 373 374 ours: 375 /* 376 * If offset or IP_MF are set, must reassemble. 377 * Otherwise, nothing need be done. 378 * (We could look in the reassembly queue to see 379 * if the packet was previously fragmented, 380 * but it's not worth the time; just let them time out.) 381 */ 382 if (ip->ip_off & ~(IP_DF|IP_RF)) { 383 /* 384 * Look for queue of fragments 385 * of this datagram. 386 */ 387 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 388 if (ip->ip_id == fp->ipq_id && 389 in_hosteq(ip->ip_src, fp->ipq_src) && 390 in_hosteq(ip->ip_dst, fp->ipq_dst) && 391 ip->ip_p == fp->ipq_p) 392 goto found; 393 fp = 0; 394 found: 395 396 /* 397 * Adjust ip_len to not reflect header, 398 * set ipqe_mff if more fragments are expected, 399 * convert offset of this to bytes. 400 */ 401 ip->ip_len -= hlen; 402 mff = (ip->ip_off & IP_MF) != 0; 403 if (mff) { 404 /* 405 * Make sure that fragments have a data length 406 * that's a non-zero multiple of 8 bytes. 407 */ 408 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 409 ipstat.ips_badfrags++; 410 goto bad; 411 } 412 } 413 ip->ip_off <<= 3; 414 415 /* 416 * If datagram marked as having more fragments 417 * or if this is not the first fragment, 418 * attempt reassembly; if it succeeds, proceed. 419 */ 420 if (mff || ip->ip_off) { 421 ipstat.ips_fragments++; 422 MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent), 423 M_IPQ, M_NOWAIT); 424 if (ipqe == NULL) { 425 ipstat.ips_rcvmemdrop++; 426 goto bad; 427 } 428 ipqe->ipqe_mff = mff; 429 ipqe->ipqe_m = m; 430 ipqe->ipqe_ip = ip; 431 m = ip_reass(ipqe, fp); 432 if (m == 0) 433 goto next; 434 ipstat.ips_reassembled++; 435 ip = mtod(m, struct ip *); 436 } else 437 if (fp) 438 ip_freef(fp); 439 } else 440 ip->ip_len -= hlen; 441 442 /* 443 * Switch out to protocol's input routine. 444 */ 445 ipstat.ips_delivered++; 446 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 447 goto next; 448 bad: 449 m_freem(m); 450 goto next; 451 } 452 453 /* 454 * Take incoming datagram fragment and try to 455 * reassemble it into whole datagram. If a chain for 456 * reassembly of this datagram already exists, then it 457 * is given as fp; otherwise have to make a chain. 458 */ 459 struct mbuf * 460 ip_reass(ipqe, fp) 461 register struct ipqent *ipqe; 462 register struct ipq *fp; 463 { 464 register struct mbuf *m = ipqe->ipqe_m; 465 register struct ipqent *nq, *p, *q; 466 struct ip *ip; 467 struct mbuf *t; 468 int hlen = ipqe->ipqe_ip->ip_hl << 2; 469 int i, next; 470 471 /* 472 * Presence of header sizes in mbufs 473 * would confuse code below. 474 */ 475 m->m_data += hlen; 476 m->m_len -= hlen; 477 478 /* 479 * If first fragment to arrive, create a reassembly queue. 480 */ 481 if (fp == 0) { 482 MALLOC(fp, struct ipq *, sizeof (struct ipq), 483 M_FTABLE, M_NOWAIT); 484 if (fp == NULL) 485 goto dropfrag; 486 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 487 fp->ipq_ttl = IPFRAGTTL; 488 fp->ipq_p = ipqe->ipqe_ip->ip_p; 489 fp->ipq_id = ipqe->ipqe_ip->ip_id; 490 LIST_INIT(&fp->ipq_fragq); 491 fp->ipq_src = ipqe->ipqe_ip->ip_src; 492 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 493 p = NULL; 494 goto insert; 495 } 496 497 /* 498 * Find a segment which begins after this one does. 499 */ 500 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 501 p = q, q = q->ipqe_q.le_next) 502 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 503 break; 504 505 /* 506 * If there is a preceding segment, it may provide some of 507 * our data already. If so, drop the data from the incoming 508 * segment. If it provides all of our data, drop us. 509 */ 510 if (p != NULL) { 511 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 512 ipqe->ipqe_ip->ip_off; 513 if (i > 0) { 514 if (i >= ipqe->ipqe_ip->ip_len) 515 goto dropfrag; 516 m_adj(ipqe->ipqe_m, i); 517 ipqe->ipqe_ip->ip_off += i; 518 ipqe->ipqe_ip->ip_len -= i; 519 } 520 } 521 522 /* 523 * While we overlap succeeding segments trim them or, 524 * if they are completely covered, dequeue them. 525 */ 526 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 527 q->ipqe_ip->ip_off; q = nq) { 528 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 529 q->ipqe_ip->ip_off; 530 if (i < q->ipqe_ip->ip_len) { 531 q->ipqe_ip->ip_len -= i; 532 q->ipqe_ip->ip_off += i; 533 m_adj(q->ipqe_m, i); 534 break; 535 } 536 nq = q->ipqe_q.le_next; 537 m_freem(q->ipqe_m); 538 LIST_REMOVE(q, ipqe_q); 539 FREE(q, M_IPQ); 540 } 541 542 insert: 543 /* 544 * Stick new segment in its place; 545 * check for complete reassembly. 546 */ 547 if (p == NULL) { 548 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 549 } else { 550 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 551 } 552 next = 0; 553 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 554 p = q, q = q->ipqe_q.le_next) { 555 if (q->ipqe_ip->ip_off != next) 556 return (0); 557 next += q->ipqe_ip->ip_len; 558 } 559 if (p->ipqe_mff) 560 return (0); 561 562 /* 563 * Reassembly is complete. Check for a bogus message size and 564 * concatenate fragments. 565 */ 566 q = fp->ipq_fragq.lh_first; 567 ip = q->ipqe_ip; 568 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 569 ipstat.ips_toolong++; 570 ip_freef(fp); 571 return (0); 572 } 573 m = q->ipqe_m; 574 t = m->m_next; 575 m->m_next = 0; 576 m_cat(m, t); 577 nq = q->ipqe_q.le_next; 578 FREE(q, M_IPQ); 579 for (q = nq; q != NULL; q = nq) { 580 t = q->ipqe_m; 581 nq = q->ipqe_q.le_next; 582 FREE(q, M_IPQ); 583 m_cat(m, t); 584 } 585 586 /* 587 * Create header for new ip packet by 588 * modifying header of first packet; 589 * dequeue and discard fragment reassembly header. 590 * Make header visible. 591 */ 592 ip->ip_len = next; 593 ip->ip_src = fp->ipq_src; 594 ip->ip_dst = fp->ipq_dst; 595 LIST_REMOVE(fp, ipq_q); 596 FREE(fp, M_FTABLE); 597 m->m_len += (ip->ip_hl << 2); 598 m->m_data -= (ip->ip_hl << 2); 599 /* some debugging cruft by sklower, below, will go away soon */ 600 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 601 register int plen = 0; 602 for (t = m; t; t = t->m_next) 603 plen += t->m_len; 604 m->m_pkthdr.len = plen; 605 } 606 return (m); 607 608 dropfrag: 609 ipstat.ips_fragdropped++; 610 m_freem(m); 611 FREE(ipqe, M_IPQ); 612 return (0); 613 } 614 615 /* 616 * Free a fragment reassembly header and all 617 * associated datagrams. 618 */ 619 void 620 ip_freef(fp) 621 struct ipq *fp; 622 { 623 register struct ipqent *q, *p; 624 625 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 626 p = q->ipqe_q.le_next; 627 m_freem(q->ipqe_m); 628 LIST_REMOVE(q, ipqe_q); 629 FREE(q, M_IPQ); 630 } 631 LIST_REMOVE(fp, ipq_q); 632 FREE(fp, M_FTABLE); 633 } 634 635 /* 636 * IP timer processing; 637 * if a timer expires on a reassembly 638 * queue, discard it. 639 */ 640 void 641 ip_slowtimo() 642 { 643 register struct ipq *fp, *nfp; 644 int s = splsoftnet(); 645 646 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 647 nfp = fp->ipq_q.le_next; 648 if (--fp->ipq_ttl == 0) { 649 ipstat.ips_fragtimeout++; 650 ip_freef(fp); 651 } 652 } 653 splx(s); 654 } 655 656 /* 657 * Drain off all datagram fragments. 658 */ 659 void 660 ip_drain() 661 { 662 663 while (ipq.lh_first != NULL) { 664 ipstat.ips_fragdropped++; 665 ip_freef(ipq.lh_first); 666 } 667 } 668 669 /* 670 * Do option processing on a datagram, 671 * possibly discarding it if bad options are encountered, 672 * or forwarding it if source-routed. 673 * Returns 1 if packet has been forwarded/freed, 674 * 0 if the packet should be processed further. 675 */ 676 int 677 ip_dooptions(m) 678 struct mbuf *m; 679 { 680 register struct ip *ip = mtod(m, struct ip *); 681 register u_char *cp; 682 register struct ip_timestamp *ipt; 683 register struct in_ifaddr *ia; 684 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 685 struct in_addr *sin, dst; 686 n_time ntime; 687 688 dst = ip->ip_dst; 689 cp = (u_char *)(ip + 1); 690 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 691 for (; cnt > 0; cnt -= optlen, cp += optlen) { 692 opt = cp[IPOPT_OPTVAL]; 693 if (opt == IPOPT_EOL) 694 break; 695 if (opt == IPOPT_NOP) 696 optlen = 1; 697 else { 698 optlen = cp[IPOPT_OLEN]; 699 if (optlen <= 0 || optlen > cnt) { 700 code = &cp[IPOPT_OLEN] - (u_char *)ip; 701 goto bad; 702 } 703 } 704 switch (opt) { 705 706 default: 707 break; 708 709 /* 710 * Source routing with record. 711 * Find interface with current destination address. 712 * If none on this machine then drop if strictly routed, 713 * or do nothing if loosely routed. 714 * Record interface address and bring up next address 715 * component. If strictly routed make sure next 716 * address is on directly accessible net. 717 */ 718 case IPOPT_LSRR: 719 case IPOPT_SSRR: 720 if (ip_allowsrcrt == 0) { 721 type = ICMP_UNREACH; 722 code = ICMP_UNREACH_NET_PROHIB; 723 goto bad; 724 } 725 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 726 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 727 goto bad; 728 } 729 ipaddr.sin_addr = ip->ip_dst; 730 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 731 if (ia == 0) { 732 if (opt == IPOPT_SSRR) { 733 type = ICMP_UNREACH; 734 code = ICMP_UNREACH_SRCFAIL; 735 goto bad; 736 } 737 /* 738 * Loose routing, and not at next destination 739 * yet; nothing to do except forward. 740 */ 741 break; 742 } 743 off--; /* 0 origin */ 744 if (off > optlen - sizeof(struct in_addr)) { 745 /* 746 * End of source route. Should be for us. 747 */ 748 save_rte(cp, ip->ip_src); 749 break; 750 } 751 /* 752 * locate outgoing interface 753 */ 754 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 755 sizeof(ipaddr.sin_addr)); 756 if (opt == IPOPT_SSRR) { 757 #define INA struct in_ifaddr * 758 #define SA struct sockaddr * 759 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 760 } else 761 ia = ip_rtaddr(ipaddr.sin_addr); 762 if (ia == 0) { 763 type = ICMP_UNREACH; 764 code = ICMP_UNREACH_SRCFAIL; 765 goto bad; 766 } 767 ip->ip_dst = ipaddr.sin_addr; 768 bcopy((caddr_t)&ia->ia_addr.sin_addr, 769 (caddr_t)(cp + off), sizeof(struct in_addr)); 770 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 771 /* 772 * Let ip_intr's mcast routing check handle mcast pkts 773 */ 774 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 775 break; 776 777 case IPOPT_RR: 778 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 779 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 780 goto bad; 781 } 782 /* 783 * If no space remains, ignore. 784 */ 785 off--; /* 0 origin */ 786 if (off > optlen - sizeof(struct in_addr)) 787 break; 788 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 789 sizeof(ipaddr.sin_addr)); 790 /* 791 * locate outgoing interface; if we're the destination, 792 * use the incoming interface (should be same). 793 */ 794 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 795 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 796 type = ICMP_UNREACH; 797 code = ICMP_UNREACH_HOST; 798 goto bad; 799 } 800 bcopy((caddr_t)&ia->ia_addr.sin_addr, 801 (caddr_t)(cp + off), sizeof(struct in_addr)); 802 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 803 break; 804 805 case IPOPT_TS: 806 code = cp - (u_char *)ip; 807 ipt = (struct ip_timestamp *)cp; 808 if (ipt->ipt_len < 5) 809 goto bad; 810 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 811 if (++ipt->ipt_oflw == 0) 812 goto bad; 813 break; 814 } 815 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 816 switch (ipt->ipt_flg) { 817 818 case IPOPT_TS_TSONLY: 819 break; 820 821 case IPOPT_TS_TSANDADDR: 822 if (ipt->ipt_ptr + sizeof(n_time) + 823 sizeof(struct in_addr) > ipt->ipt_len) 824 goto bad; 825 ipaddr.sin_addr = dst; 826 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 827 m->m_pkthdr.rcvif); 828 if (ia == 0) 829 continue; 830 bcopy((caddr_t)&ia->ia_addr.sin_addr, 831 (caddr_t)sin, sizeof(struct in_addr)); 832 ipt->ipt_ptr += sizeof(struct in_addr); 833 break; 834 835 case IPOPT_TS_PRESPEC: 836 if (ipt->ipt_ptr + sizeof(n_time) + 837 sizeof(struct in_addr) > ipt->ipt_len) 838 goto bad; 839 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 840 sizeof(struct in_addr)); 841 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 842 continue; 843 ipt->ipt_ptr += sizeof(struct in_addr); 844 break; 845 846 default: 847 goto bad; 848 } 849 ntime = iptime(); 850 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 851 sizeof(n_time)); 852 ipt->ipt_ptr += sizeof(n_time); 853 } 854 } 855 if (forward) { 856 if (ip_forwsrcrt == 0) { 857 type = ICMP_UNREACH; 858 code = ICMP_UNREACH_SRCFAIL; 859 goto bad; 860 } 861 ip_forward(m, 1); 862 return (1); 863 } 864 return (0); 865 bad: 866 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 867 icmp_error(m, type, code, 0, 0); 868 ipstat.ips_badoptions++; 869 return (1); 870 } 871 872 /* 873 * Given address of next destination (final or next hop), 874 * return internet address info of interface to be used to get there. 875 */ 876 struct in_ifaddr * 877 ip_rtaddr(dst) 878 struct in_addr dst; 879 { 880 register struct sockaddr_in *sin; 881 882 sin = satosin(&ipforward_rt.ro_dst); 883 884 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 885 if (ipforward_rt.ro_rt) { 886 RTFREE(ipforward_rt.ro_rt); 887 ipforward_rt.ro_rt = 0; 888 } 889 sin->sin_family = AF_INET; 890 sin->sin_len = sizeof(*sin); 891 sin->sin_addr = dst; 892 893 rtalloc(&ipforward_rt); 894 } 895 if (ipforward_rt.ro_rt == 0) 896 return ((struct in_ifaddr *)0); 897 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 898 } 899 900 /* 901 * Save incoming source route for use in replies, 902 * to be picked up later by ip_srcroute if the receiver is interested. 903 */ 904 void 905 save_rte(option, dst) 906 u_char *option; 907 struct in_addr dst; 908 { 909 unsigned olen; 910 911 olen = option[IPOPT_OLEN]; 912 #ifdef DIAGNOSTIC 913 if (ipprintfs) 914 printf("save_rte: olen %d\n", olen); 915 #endif 916 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 917 return; 918 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 919 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 920 ip_srcrt.dst = dst; 921 } 922 923 /* 924 * Retrieve incoming source route for use in replies, 925 * in the same form used by setsockopt. 926 * The first hop is placed before the options, will be removed later. 927 */ 928 struct mbuf * 929 ip_srcroute() 930 { 931 register struct in_addr *p, *q; 932 register struct mbuf *m; 933 934 if (ip_nhops == 0) 935 return ((struct mbuf *)0); 936 m = m_get(M_DONTWAIT, MT_SOOPTS); 937 if (m == 0) 938 return ((struct mbuf *)0); 939 940 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 941 942 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 943 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 944 OPTSIZ; 945 #ifdef DIAGNOSTIC 946 if (ipprintfs) 947 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 948 #endif 949 950 /* 951 * First save first hop for return route 952 */ 953 p = &ip_srcrt.route[ip_nhops - 1]; 954 *(mtod(m, struct in_addr *)) = *p--; 955 #ifdef DIAGNOSTIC 956 if (ipprintfs) 957 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 958 #endif 959 960 /* 961 * Copy option fields and padding (nop) to mbuf. 962 */ 963 ip_srcrt.nop = IPOPT_NOP; 964 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 965 bcopy((caddr_t)&ip_srcrt.nop, 966 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 967 q = (struct in_addr *)(mtod(m, caddr_t) + 968 sizeof(struct in_addr) + OPTSIZ); 969 #undef OPTSIZ 970 /* 971 * Record return path as an IP source route, 972 * reversing the path (pointers are now aligned). 973 */ 974 while (p >= ip_srcrt.route) { 975 #ifdef DIAGNOSTIC 976 if (ipprintfs) 977 printf(" %x", ntohl(q->s_addr)); 978 #endif 979 *q++ = *p--; 980 } 981 /* 982 * Last hop goes to final destination. 983 */ 984 *q = ip_srcrt.dst; 985 #ifdef DIAGNOSTIC 986 if (ipprintfs) 987 printf(" %x\n", ntohl(q->s_addr)); 988 #endif 989 return (m); 990 } 991 992 /* 993 * Strip out IP options, at higher 994 * level protocol in the kernel. 995 * Second argument is buffer to which options 996 * will be moved, and return value is their length. 997 * XXX should be deleted; last arg currently ignored. 998 */ 999 void 1000 ip_stripoptions(m, mopt) 1001 register struct mbuf *m; 1002 struct mbuf *mopt; 1003 { 1004 register int i; 1005 struct ip *ip = mtod(m, struct ip *); 1006 register caddr_t opts; 1007 int olen; 1008 1009 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1010 opts = (caddr_t)(ip + 1); 1011 i = m->m_len - (sizeof (struct ip) + olen); 1012 bcopy(opts + olen, opts, (unsigned)i); 1013 m->m_len -= olen; 1014 if (m->m_flags & M_PKTHDR) 1015 m->m_pkthdr.len -= olen; 1016 ip->ip_hl = sizeof(struct ip) >> 2; 1017 } 1018 1019 int inetctlerrmap[PRC_NCMDS] = { 1020 0, 0, 0, 0, 1021 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1022 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1023 EMSGSIZE, EHOSTUNREACH, 0, 0, 1024 0, 0, 0, 0, 1025 ENOPROTOOPT 1026 }; 1027 1028 /* 1029 * Forward a packet. If some error occurs return the sender 1030 * an icmp packet. Note we can't always generate a meaningful 1031 * icmp message because icmp doesn't have a large enough repertoire 1032 * of codes and types. 1033 * 1034 * If not forwarding, just drop the packet. This could be confusing 1035 * if ipforwarding was zero but some routing protocol was advancing 1036 * us as a gateway to somewhere. However, we must let the routing 1037 * protocol deal with that. 1038 * 1039 * The srcrt parameter indicates whether the packet is being forwarded 1040 * via a source route. 1041 */ 1042 void 1043 ip_forward(m, srcrt) 1044 struct mbuf *m; 1045 int srcrt; 1046 { 1047 register struct ip *ip = mtod(m, struct ip *); 1048 register struct sockaddr_in *sin; 1049 register struct rtentry *rt; 1050 int error, type = 0, code = 0; 1051 struct mbuf *mcopy; 1052 n_long dest; 1053 struct ifnet *destifp; 1054 1055 dest = 0; 1056 #ifdef DIAGNOSTIC 1057 if (ipprintfs) 1058 printf("forward: src %x dst %x ttl %x\n", 1059 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1060 #endif 1061 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1062 ipstat.ips_cantforward++; 1063 m_freem(m); 1064 return; 1065 } 1066 HTONS(ip->ip_id); 1067 if (ip->ip_ttl <= IPTTLDEC) { 1068 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1069 return; 1070 } 1071 ip->ip_ttl -= IPTTLDEC; 1072 1073 sin = satosin(&ipforward_rt.ro_dst); 1074 if ((rt = ipforward_rt.ro_rt) == 0 || 1075 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1076 if (ipforward_rt.ro_rt) { 1077 RTFREE(ipforward_rt.ro_rt); 1078 ipforward_rt.ro_rt = 0; 1079 } 1080 sin->sin_family = AF_INET; 1081 sin->sin_len = sizeof(struct sockaddr_in); 1082 sin->sin_addr = ip->ip_dst; 1083 1084 rtalloc(&ipforward_rt); 1085 if (ipforward_rt.ro_rt == 0) { 1086 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1087 return; 1088 } 1089 rt = ipforward_rt.ro_rt; 1090 } 1091 1092 /* 1093 * Save at most 68 bytes of the packet in case 1094 * we need to generate an ICMP message to the src. 1095 */ 1096 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1097 1098 /* 1099 * If forwarding packet using same interface that it came in on, 1100 * perhaps should send a redirect to sender to shortcut a hop. 1101 * Only send redirect if source is sending directly to us, 1102 * and if packet was not source routed (or has any options). 1103 * Also, don't send redirect if forwarding using a default route 1104 * or a route modified by a redirect. 1105 */ 1106 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1107 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1108 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1109 ipsendredirects && !srcrt) { 1110 if (rt->rt_ifa && 1111 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1112 ifatoia(rt->rt_ifa)->ia_subnet) { 1113 if (rt->rt_flags & RTF_GATEWAY) 1114 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1115 else 1116 dest = ip->ip_dst.s_addr; 1117 /* Router requirements says to only send host redirects */ 1118 type = ICMP_REDIRECT; 1119 code = ICMP_REDIRECT_HOST; 1120 #ifdef DIAGNOSTIC 1121 if (ipprintfs) 1122 printf("redirect (%d) to %x\n", code, (u_int32_t)dest); 1123 #endif 1124 } 1125 } 1126 1127 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1128 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1129 if (error) 1130 ipstat.ips_cantforward++; 1131 else { 1132 ipstat.ips_forward++; 1133 if (type) 1134 ipstat.ips_redirectsent++; 1135 else { 1136 if (mcopy) 1137 m_freem(mcopy); 1138 return; 1139 } 1140 } 1141 if (mcopy == NULL) 1142 return; 1143 destifp = NULL; 1144 1145 switch (error) { 1146 1147 case 0: /* forwarded, but need redirect */ 1148 /* type, code set above */ 1149 break; 1150 1151 case ENETUNREACH: /* shouldn't happen, checked above */ 1152 case EHOSTUNREACH: 1153 case ENETDOWN: 1154 case EHOSTDOWN: 1155 default: 1156 type = ICMP_UNREACH; 1157 code = ICMP_UNREACH_HOST; 1158 break; 1159 1160 case EMSGSIZE: 1161 type = ICMP_UNREACH; 1162 code = ICMP_UNREACH_NEEDFRAG; 1163 if (ipforward_rt.ro_rt) 1164 destifp = ipforward_rt.ro_rt->rt_ifp; 1165 ipstat.ips_cantfrag++; 1166 break; 1167 1168 case ENOBUFS: 1169 type = ICMP_SOURCEQUENCH; 1170 code = 0; 1171 break; 1172 } 1173 icmp_error(mcopy, type, code, dest, destifp); 1174 } 1175 1176 void 1177 ip_savecontrol(inp, mp, ip, m) 1178 register struct inpcb *inp; 1179 register struct mbuf **mp; 1180 register struct ip *ip; 1181 register struct mbuf *m; 1182 { 1183 1184 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1185 struct timeval tv; 1186 1187 microtime(&tv); 1188 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1189 SCM_TIMESTAMP, SOL_SOCKET); 1190 if (*mp) 1191 mp = &(*mp)->m_next; 1192 } 1193 if (inp->inp_flags & INP_RECVDSTADDR) { 1194 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1195 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1196 if (*mp) 1197 mp = &(*mp)->m_next; 1198 } 1199 #ifdef notyet 1200 /* 1201 * XXX 1202 * Moving these out of udp_input() made them even more broken 1203 * than they already were. 1204 * - fenner@parc.xerox.com 1205 */ 1206 /* options were tossed already */ 1207 if (inp->inp_flags & INP_RECVOPTS) { 1208 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1209 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1210 if (*mp) 1211 mp = &(*mp)->m_next; 1212 } 1213 /* ip_srcroute doesn't do what we want here, need to fix */ 1214 if (inp->inp_flags & INP_RECVRETOPTS) { 1215 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1216 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1217 if (*mp) 1218 mp = &(*mp)->m_next; 1219 } 1220 #endif 1221 if (inp->inp_flags & INP_RECVIF) { 1222 struct sockaddr_dl sdl; 1223 1224 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1225 sdl.sdl_family = AF_LINK; 1226 sdl.sdl_index = m->m_pkthdr.rcvif ? 1227 m->m_pkthdr.rcvif->if_index : 0; 1228 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1229 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1230 IP_RECVIF, IPPROTO_IP); 1231 if (*mp) 1232 mp = &(*mp)->m_next; 1233 } 1234 } 1235 1236 int 1237 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1238 int *name; 1239 u_int namelen; 1240 void *oldp; 1241 size_t *oldlenp; 1242 void *newp; 1243 size_t newlen; 1244 { 1245 /* All sysctl names at this level are terminal. */ 1246 if (namelen != 1) 1247 return (ENOTDIR); 1248 1249 switch (name[0]) { 1250 case IPCTL_FORWARDING: 1251 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1252 case IPCTL_SENDREDIRECTS: 1253 return (sysctl_int(oldp, oldlenp, newp, newlen, 1254 &ipsendredirects)); 1255 case IPCTL_DEFTTL: 1256 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1257 #ifdef notyet 1258 case IPCTL_DEFMTU: 1259 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1260 #endif 1261 case IPCTL_FORWSRCRT: 1262 /* Don't allow this to change in a secure environment. */ 1263 if (securelevel > 0) 1264 return (sysctl_rdint(oldp, oldlenp, newp, 1265 ip_forwsrcrt)); 1266 else 1267 return (sysctl_int(oldp, oldlenp, newp, newlen, 1268 &ip_forwsrcrt)); 1269 case IPCTL_DIRECTEDBCAST: 1270 return (sysctl_int(oldp, oldlenp, newp, newlen, 1271 &ip_directedbcast)); 1272 case IPCTL_ALLOWSRCRT: 1273 return (sysctl_int(oldp, oldlenp, newp, newlen, 1274 &ip_allowsrcrt)); 1275 default: 1276 return (EOPNOTSUPP); 1277 } 1278 /* NOTREACHED */ 1279 } 1280