1 /* $NetBSD: ip_input.c,v 1.15 1995/04/13 06:33:21 cgd 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/errno.h> 46 #include <sys/time.h> 47 #include <sys/kernel.h> 48 49 #include <net/if.h> 50 #include <net/route.h> 51 52 #include <netinet/in.h> 53 #include <netinet/in_systm.h> 54 #include <netinet/ip.h> 55 #include <netinet/in_pcb.h> 56 #include <netinet/in_var.h> 57 #include <netinet/ip_var.h> 58 #include <netinet/ip_icmp.h> 59 60 #ifndef IPFORWARDING 61 #ifdef GATEWAY 62 #define IPFORWARDING 1 /* forward IP packets not for us */ 63 #else /* GATEWAY */ 64 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 65 #endif /* GATEWAY */ 66 #endif /* IPFORWARDING */ 67 #ifndef IPSENDREDIRECTS 68 #define IPSENDREDIRECTS 1 69 #endif 70 int ipforwarding = IPFORWARDING; 71 int ipsendredirects = IPSENDREDIRECTS; 72 int ip_defttl = IPDEFTTL; 73 #ifdef DIAGNOSTIC 74 int ipprintfs = 0; 75 #endif 76 77 extern struct domain inetdomain; 78 extern struct protosw inetsw[]; 79 u_char ip_protox[IPPROTO_MAX]; 80 int ipqmaxlen = IFQ_MAXLEN; 81 struct in_ifaddr *in_ifaddr; /* first inet address */ 82 struct ifqueue ipintrq; 83 84 /* 85 * We need to save the IP options in case a protocol wants to respond 86 * to an incoming packet over the same route if the packet got here 87 * using IP source routing. This allows connection establishment and 88 * maintenance when the remote end is on a network that is not known 89 * to us. 90 */ 91 int ip_nhops = 0; 92 static struct ip_srcrt { 93 struct in_addr dst; /* final destination */ 94 char nop; /* one NOP to align */ 95 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 96 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 97 } ip_srcrt; 98 99 #ifdef GATEWAY 100 extern int if_index; 101 u_int32_t *ip_ifmatrix; 102 #endif 103 104 static void save_rte __P((u_char *, struct in_addr)); 105 /* 106 * IP initialization: fill in IP protocol switch table. 107 * All protocols not implemented in kernel go to raw IP protocol handler. 108 */ 109 void 110 ip_init() 111 { 112 register struct protosw *pr; 113 register int i; 114 115 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 116 if (pr == 0) 117 panic("ip_init"); 118 for (i = 0; i < IPPROTO_MAX; i++) 119 ip_protox[i] = pr - inetsw; 120 for (pr = inetdomain.dom_protosw; 121 pr < inetdomain.dom_protoswNPROTOSW; pr++) 122 if (pr->pr_domain->dom_family == PF_INET && 123 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 124 ip_protox[pr->pr_protocol] = pr - inetsw; 125 ipq.next = ipq.prev = &ipq; 126 ip_id = time.tv_sec & 0xffff; 127 ipintrq.ifq_maxlen = ipqmaxlen; 128 #ifdef GATEWAY 129 i = (if_index + 1) * (if_index + 1) * sizeof (u_int32_t); 130 ip_ifmatrix = (u_int32_t *) malloc(i, M_RTABLE, M_WAITOK); 131 bzero((char *)ip_ifmatrix, i); 132 #endif 133 } 134 135 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 136 struct route ipforward_rt; 137 138 /* 139 * Ip input routine. Checksum and byte swap header. If fragmented 140 * try to reassemble. Process options. Pass to next level. 141 */ 142 void 143 ipintr() 144 { 145 register struct ip *ip; 146 register struct mbuf *m; 147 register struct ipq *fp; 148 register struct in_ifaddr *ia; 149 int hlen, s; 150 151 next: 152 /* 153 * Get next datagram off input queue and get IP header 154 * in first mbuf. 155 */ 156 s = splimp(); 157 IF_DEQUEUE(&ipintrq, m); 158 splx(s); 159 if (m == 0) 160 return; 161 #ifdef DIAGNOSTIC 162 if ((m->m_flags & M_PKTHDR) == 0) 163 panic("ipintr no HDR"); 164 #endif 165 /* 166 * If no IP addresses have been set yet but the interfaces 167 * are receiving, can't do anything with incoming packets yet. 168 */ 169 if (in_ifaddr == NULL) 170 goto bad; 171 ipstat.ips_total++; 172 if (m->m_len < sizeof (struct ip) && 173 (m = m_pullup(m, sizeof (struct ip))) == 0) { 174 ipstat.ips_toosmall++; 175 goto next; 176 } 177 ip = mtod(m, struct ip *); 178 if (ip->ip_v != IPVERSION) { 179 ipstat.ips_badvers++; 180 goto bad; 181 } 182 hlen = ip->ip_hl << 2; 183 if (hlen < sizeof(struct ip)) { /* minimum header length */ 184 ipstat.ips_badhlen++; 185 goto bad; 186 } 187 if (hlen > m->m_len) { 188 if ((m = m_pullup(m, hlen)) == 0) { 189 ipstat.ips_badhlen++; 190 goto next; 191 } 192 ip = mtod(m, struct ip *); 193 } 194 if (ip->ip_sum = in_cksum(m, hlen)) { 195 ipstat.ips_badsum++; 196 goto bad; 197 } 198 199 /* 200 * Convert fields to host representation. 201 */ 202 NTOHS(ip->ip_len); 203 if (ip->ip_len < hlen) { 204 ipstat.ips_badlen++; 205 goto bad; 206 } 207 NTOHS(ip->ip_id); 208 NTOHS(ip->ip_off); 209 210 /* 211 * Check that the amount of data in the buffers 212 * is as at least much as the IP header would have us expect. 213 * Trim mbufs if longer than we expect. 214 * Drop packet if shorter than we expect. 215 */ 216 if (m->m_pkthdr.len < ip->ip_len) { 217 ipstat.ips_tooshort++; 218 goto bad; 219 } 220 if (m->m_pkthdr.len > ip->ip_len) { 221 if (m->m_len == m->m_pkthdr.len) { 222 m->m_len = ip->ip_len; 223 m->m_pkthdr.len = ip->ip_len; 224 } else 225 m_adj(m, ip->ip_len - m->m_pkthdr.len); 226 } 227 228 /* 229 * Process options and, if not destined for us, 230 * ship it on. ip_dooptions returns 1 when an 231 * error was detected (causing an icmp message 232 * to be sent and the original packet to be freed). 233 */ 234 ip_nhops = 0; /* for source routed packets */ 235 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 236 goto next; 237 238 /* 239 * Check our list of addresses, to see if the packet is for us. 240 */ 241 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 242 #define satosin(sa) ((struct sockaddr_in *)(sa)) 243 244 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 245 goto ours; 246 if ( 247 #ifdef DIRECTED_BROADCAST 248 ia->ia_ifp == m->m_pkthdr.rcvif && 249 #endif 250 (ia->ia_ifp->if_flags & IFF_BROADCAST)) { 251 u_int32_t t; 252 253 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 254 ip->ip_dst.s_addr) 255 goto ours; 256 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 257 goto ours; 258 /* 259 * Look for all-0's host part (old broadcast addr), 260 * either for subnet or net. 261 */ 262 t = ntohl(ip->ip_dst.s_addr); 263 if (t == ia->ia_subnet) 264 goto ours; 265 if (t == ia->ia_net) 266 goto ours; 267 } 268 } 269 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 270 struct in_multi *inm; 271 #ifdef MROUTING 272 extern struct socket *ip_mrouter; 273 274 if (m->m_flags & M_EXT) { 275 if ((m = m_pullup(m, hlen)) == 0) { 276 ipstat.ips_toosmall++; 277 goto next; 278 } 279 ip = mtod(m, struct ip *); 280 } 281 282 if (ip_mrouter) { 283 /* 284 * If we are acting as a multicast router, all 285 * incoming multicast packets are passed to the 286 * kernel-level multicast forwarding function. 287 * The packet is returned (relatively) intact; if 288 * ip_mforward() returns a non-zero value, the packet 289 * must be discarded, else it may be accepted below. 290 * 291 * (The IP ident field is put in the same byte order 292 * as expected when ip_mforward() is called from 293 * ip_output().) 294 */ 295 ip->ip_id = htons(ip->ip_id); 296 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 297 ipstat.ips_cantforward++; 298 m_freem(m); 299 goto next; 300 } 301 ip->ip_id = ntohs(ip->ip_id); 302 303 /* 304 * The process-level routing demon needs to receive 305 * all multicast IGMP packets, whether or not this 306 * host belongs to their destination groups. 307 */ 308 if (ip->ip_p == IPPROTO_IGMP) 309 goto ours; 310 ipstat.ips_forward++; 311 } 312 #endif 313 /* 314 * See if we belong to the destination multicast group on the 315 * arrival interface. 316 */ 317 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 318 if (inm == NULL) { 319 ipstat.ips_cantforward++; 320 m_freem(m); 321 goto next; 322 } 323 goto ours; 324 } 325 if (ip->ip_dst.s_addr == (u_int32_t)INADDR_BROADCAST) 326 goto ours; 327 if (ip->ip_dst.s_addr == INADDR_ANY) 328 goto ours; 329 330 /* 331 * Not for us; forward if possible and desirable. 332 */ 333 if (ipforwarding == 0) { 334 ipstat.ips_cantforward++; 335 m_freem(m); 336 } else 337 ip_forward(m, 0); 338 goto next; 339 340 ours: 341 /* 342 * If offset or IP_MF are set, must reassemble. 343 * Otherwise, nothing need be done. 344 * (We could look in the reassembly queue to see 345 * if the packet was previously fragmented, 346 * but it's not worth the time; just let them time out.) 347 */ 348 if (ip->ip_off &~ IP_DF) { 349 if (m->m_flags & M_EXT) { /* XXX */ 350 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 351 ipstat.ips_toosmall++; 352 goto next; 353 } 354 ip = mtod(m, struct ip *); 355 } 356 /* 357 * Look for queue of fragments 358 * of this datagram. 359 */ 360 for (fp = ipq.next; fp != &ipq; fp = fp->next) 361 if (ip->ip_id == fp->ipq_id && 362 ip->ip_src.s_addr == fp->ipq_src.s_addr && 363 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 364 ip->ip_p == fp->ipq_p) 365 goto found; 366 fp = 0; 367 found: 368 369 /* 370 * Adjust ip_len to not reflect header, 371 * set ip_mff if more fragments are expected, 372 * convert offset of this to bytes. 373 */ 374 ip->ip_len -= hlen; 375 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 376 if (ip->ip_off & IP_MF) 377 ((struct ipasfrag *)ip)->ipf_mff |= 1; 378 ip->ip_off <<= 3; 379 380 /* 381 * If datagram marked as having more fragments 382 * or if this is not the first fragment, 383 * attempt reassembly; if it succeeds, proceed. 384 */ 385 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 386 ipstat.ips_fragments++; 387 ip = ip_reass((struct ipasfrag *)ip, fp); 388 if (ip == 0) 389 goto next; 390 ipstat.ips_reassembled++; 391 m = dtom(ip); 392 } else 393 if (fp) 394 ip_freef(fp); 395 } else 396 ip->ip_len -= hlen; 397 398 /* 399 * Switch out to protocol's input routine. 400 */ 401 ipstat.ips_delivered++; 402 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 403 goto next; 404 bad: 405 m_freem(m); 406 goto next; 407 } 408 409 /* 410 * Take incoming datagram fragment and try to 411 * reassemble it into whole datagram. If a chain for 412 * reassembly of this datagram already exists, then it 413 * is given as fp; otherwise have to make a chain. 414 */ 415 struct ip * 416 ip_reass(ip, fp) 417 register struct ipasfrag *ip; 418 register struct ipq *fp; 419 { 420 register struct mbuf *m = dtom(ip); 421 register struct ipasfrag *q; 422 struct mbuf *t; 423 int hlen = ip->ip_hl << 2; 424 int i, next; 425 426 /* 427 * Presence of header sizes in mbufs 428 * would confuse code below. 429 */ 430 m->m_data += hlen; 431 m->m_len -= hlen; 432 433 /* 434 * If first fragment to arrive, create a reassembly queue. 435 */ 436 if (fp == 0) { 437 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 438 goto dropfrag; 439 fp = mtod(t, struct ipq *); 440 insque(fp, &ipq); 441 fp->ipq_ttl = IPFRAGTTL; 442 fp->ipq_p = ip->ip_p; 443 fp->ipq_id = ip->ip_id; 444 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 445 fp->ipq_src = ((struct ip *)ip)->ip_src; 446 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 447 q = (struct ipasfrag *)fp; 448 goto insert; 449 } 450 451 /* 452 * Find a segment which begins after this one does. 453 */ 454 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 455 if (q->ip_off > ip->ip_off) 456 break; 457 458 /* 459 * If there is a preceding segment, it may provide some of 460 * our data already. If so, drop the data from the incoming 461 * segment. If it provides all of our data, drop us. 462 */ 463 if (q->ipf_prev != (struct ipasfrag *)fp) { 464 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 465 if (i > 0) { 466 if (i >= ip->ip_len) 467 goto dropfrag; 468 m_adj(dtom(ip), i); 469 ip->ip_off += i; 470 ip->ip_len -= i; 471 } 472 } 473 474 /* 475 * While we overlap succeeding segments trim them or, 476 * if they are completely covered, dequeue them. 477 */ 478 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 479 i = (ip->ip_off + ip->ip_len) - q->ip_off; 480 if (i < q->ip_len) { 481 q->ip_len -= i; 482 q->ip_off += i; 483 m_adj(dtom(q), i); 484 break; 485 } 486 q = q->ipf_next; 487 m_freem(dtom(q->ipf_prev)); 488 ip_deq(q->ipf_prev); 489 } 490 491 insert: 492 /* 493 * Stick new segment in its place; 494 * check for complete reassembly. 495 */ 496 ip_enq(ip, q->ipf_prev); 497 next = 0; 498 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 499 if (q->ip_off != next) 500 return (0); 501 next += q->ip_len; 502 } 503 if (q->ipf_prev->ipf_mff & 1) 504 return (0); 505 506 /* 507 * Reassembly is complete; concatenate fragments. 508 */ 509 q = fp->ipq_next; 510 m = dtom(q); 511 t = m->m_next; 512 m->m_next = 0; 513 m_cat(m, t); 514 q = q->ipf_next; 515 while (q != (struct ipasfrag *)fp) { 516 t = dtom(q); 517 q = q->ipf_next; 518 m_cat(m, t); 519 } 520 521 /* 522 * Create header for new ip packet by 523 * modifying header of first packet; 524 * dequeue and discard fragment reassembly header. 525 * Make header visible. 526 */ 527 ip = fp->ipq_next; 528 ip->ip_len = next; 529 ip->ipf_mff &= ~1; 530 ((struct ip *)ip)->ip_src = fp->ipq_src; 531 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 532 remque(fp); 533 (void) m_free(dtom(fp)); 534 m = dtom(ip); 535 m->m_len += (ip->ip_hl << 2); 536 m->m_data -= (ip->ip_hl << 2); 537 /* some debugging cruft by sklower, below, will go away soon */ 538 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 539 register int plen = 0; 540 for (t = m; m; m = m->m_next) 541 plen += m->m_len; 542 t->m_pkthdr.len = plen; 543 } 544 return ((struct ip *)ip); 545 546 dropfrag: 547 ipstat.ips_fragdropped++; 548 m_freem(m); 549 return (0); 550 } 551 552 /* 553 * Free a fragment reassembly header and all 554 * associated datagrams. 555 */ 556 void 557 ip_freef(fp) 558 struct ipq *fp; 559 { 560 register struct ipasfrag *q, *p; 561 562 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 563 p = q->ipf_next; 564 ip_deq(q); 565 m_freem(dtom(q)); 566 } 567 remque(fp); 568 (void) m_free(dtom(fp)); 569 } 570 571 /* 572 * Put an ip fragment on a reassembly chain. 573 * Like insque, but pointers in middle of structure. 574 */ 575 void 576 ip_enq(p, prev) 577 register struct ipasfrag *p, *prev; 578 { 579 580 p->ipf_prev = prev; 581 p->ipf_next = prev->ipf_next; 582 prev->ipf_next->ipf_prev = p; 583 prev->ipf_next = p; 584 } 585 586 /* 587 * To ip_enq as remque is to insque. 588 */ 589 void 590 ip_deq(p) 591 register struct ipasfrag *p; 592 { 593 594 p->ipf_prev->ipf_next = p->ipf_next; 595 p->ipf_next->ipf_prev = p->ipf_prev; 596 } 597 598 /* 599 * IP timer processing; 600 * if a timer expires on a reassembly 601 * queue, discard it. 602 */ 603 void 604 ip_slowtimo() 605 { 606 register struct ipq *fp; 607 int s = splnet(); 608 609 fp = ipq.next; 610 if (fp == 0) { 611 splx(s); 612 return; 613 } 614 while (fp != &ipq) { 615 --fp->ipq_ttl; 616 fp = fp->next; 617 if (fp->prev->ipq_ttl == 0) { 618 ipstat.ips_fragtimeout++; 619 ip_freef(fp->prev); 620 } 621 } 622 splx(s); 623 } 624 625 /* 626 * Drain off all datagram fragments. 627 */ 628 void 629 ip_drain() 630 { 631 632 while (ipq.next != &ipq) { 633 ipstat.ips_fragdropped++; 634 ip_freef(ipq.next); 635 } 636 } 637 638 /* 639 * Do option processing on a datagram, 640 * possibly discarding it if bad options are encountered, 641 * or forwarding it if source-routed. 642 * Returns 1 if packet has been forwarded/freed, 643 * 0 if the packet should be processed further. 644 */ 645 int 646 ip_dooptions(m) 647 struct mbuf *m; 648 { 649 register struct ip *ip = mtod(m, struct ip *); 650 register u_char *cp; 651 register struct ip_timestamp *ipt; 652 register struct in_ifaddr *ia; 653 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 654 struct in_addr *sin, dst; 655 n_time ntime; 656 657 dst = ip->ip_dst; 658 cp = (u_char *)(ip + 1); 659 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 660 for (; cnt > 0; cnt -= optlen, cp += optlen) { 661 opt = cp[IPOPT_OPTVAL]; 662 if (opt == IPOPT_EOL) 663 break; 664 if (opt == IPOPT_NOP) 665 optlen = 1; 666 else { 667 optlen = cp[IPOPT_OLEN]; 668 if (optlen <= 0 || optlen > cnt) { 669 code = &cp[IPOPT_OLEN] - (u_char *)ip; 670 goto bad; 671 } 672 } 673 switch (opt) { 674 675 default: 676 break; 677 678 /* 679 * Source routing with record. 680 * Find interface with current destination address. 681 * If none on this machine then drop if strictly routed, 682 * or do nothing if loosely routed. 683 * Record interface address and bring up next address 684 * component. If strictly routed make sure next 685 * address is on directly accessible net. 686 */ 687 case IPOPT_LSRR: 688 case IPOPT_SSRR: 689 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 690 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 691 goto bad; 692 } 693 ipaddr.sin_addr = ip->ip_dst; 694 ia = (struct in_ifaddr *) 695 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 696 if (ia == 0) { 697 if (opt == IPOPT_SSRR) { 698 type = ICMP_UNREACH; 699 code = ICMP_UNREACH_SRCFAIL; 700 goto bad; 701 } 702 /* 703 * Loose routing, and not at next destination 704 * yet; nothing to do except forward. 705 */ 706 break; 707 } 708 off--; /* 0 origin */ 709 if (off > optlen - sizeof(struct in_addr)) { 710 /* 711 * End of source route. Should be for us. 712 */ 713 save_rte(cp, ip->ip_src); 714 break; 715 } 716 /* 717 * locate outgoing interface 718 */ 719 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 720 sizeof(ipaddr.sin_addr)); 721 if (opt == IPOPT_SSRR) { 722 #define INA struct in_ifaddr * 723 #define SA struct sockaddr * 724 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 725 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 726 } else 727 ia = ip_rtaddr(ipaddr.sin_addr); 728 if (ia == 0) { 729 type = ICMP_UNREACH; 730 code = ICMP_UNREACH_SRCFAIL; 731 goto bad; 732 } 733 ip->ip_dst = ipaddr.sin_addr; 734 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 735 (caddr_t)(cp + off), sizeof(struct in_addr)); 736 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 737 /* 738 * Let ip_intr's mcast routing check handle mcast pkts 739 */ 740 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 741 break; 742 743 case IPOPT_RR: 744 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 745 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 746 goto bad; 747 } 748 /* 749 * If no space remains, ignore. 750 */ 751 off--; /* 0 origin */ 752 if (off > optlen - sizeof(struct in_addr)) 753 break; 754 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 755 sizeof(ipaddr.sin_addr)); 756 /* 757 * locate outgoing interface; if we're the destination, 758 * use the incoming interface (should be same). 759 */ 760 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 761 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 762 type = ICMP_UNREACH; 763 code = ICMP_UNREACH_HOST; 764 goto bad; 765 } 766 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), 767 (caddr_t)(cp + off), sizeof(struct in_addr)); 768 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 769 break; 770 771 case IPOPT_TS: 772 code = cp - (u_char *)ip; 773 ipt = (struct ip_timestamp *)cp; 774 if (ipt->ipt_len < 5) 775 goto bad; 776 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 777 if (++ipt->ipt_oflw == 0) 778 goto bad; 779 break; 780 } 781 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 782 switch (ipt->ipt_flg) { 783 784 case IPOPT_TS_TSONLY: 785 break; 786 787 case IPOPT_TS_TSANDADDR: 788 if (ipt->ipt_ptr + sizeof(n_time) + 789 sizeof(struct in_addr) > ipt->ipt_len) 790 goto bad; 791 ipaddr.sin_addr = dst; 792 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 793 m->m_pkthdr.rcvif); 794 if (ia == 0) 795 continue; 796 bcopy((caddr_t)&IA_SIN(ia)->sin_addr, 797 (caddr_t)sin, sizeof(struct in_addr)); 798 ipt->ipt_ptr += sizeof(struct in_addr); 799 break; 800 801 case IPOPT_TS_PRESPEC: 802 if (ipt->ipt_ptr + sizeof(n_time) + 803 sizeof(struct in_addr) > ipt->ipt_len) 804 goto bad; 805 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 806 sizeof(struct in_addr)); 807 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 808 continue; 809 ipt->ipt_ptr += sizeof(struct in_addr); 810 break; 811 812 default: 813 goto bad; 814 } 815 ntime = iptime(); 816 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 817 sizeof(n_time)); 818 ipt->ipt_ptr += sizeof(n_time); 819 } 820 } 821 if (forward) { 822 ip_forward(m, 1); 823 return (1); 824 } 825 return (0); 826 bad: 827 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 828 icmp_error(m, type, code, 0, 0); 829 ipstat.ips_badoptions++; 830 return (1); 831 } 832 833 /* 834 * Given address of next destination (final or next hop), 835 * return internet address info of interface to be used to get there. 836 */ 837 struct in_ifaddr * 838 ip_rtaddr(dst) 839 struct in_addr dst; 840 { 841 register struct sockaddr_in *sin; 842 843 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 844 845 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 846 if (ipforward_rt.ro_rt) { 847 RTFREE(ipforward_rt.ro_rt); 848 ipforward_rt.ro_rt = 0; 849 } 850 sin->sin_family = AF_INET; 851 sin->sin_len = sizeof(*sin); 852 sin->sin_addr = dst; 853 854 rtalloc(&ipforward_rt); 855 } 856 if (ipforward_rt.ro_rt == 0) 857 return ((struct in_ifaddr *)0); 858 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 859 } 860 861 /* 862 * Save incoming source route for use in replies, 863 * to be picked up later by ip_srcroute if the receiver is interested. 864 */ 865 void 866 save_rte(option, dst) 867 u_char *option; 868 struct in_addr dst; 869 { 870 unsigned olen; 871 872 olen = option[IPOPT_OLEN]; 873 #ifdef DIAGNOSTIC 874 if (ipprintfs) 875 printf("save_rte: olen %d\n", olen); 876 #endif 877 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 878 return; 879 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 880 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 881 ip_srcrt.dst = dst; 882 } 883 884 /* 885 * Retrieve incoming source route for use in replies, 886 * in the same form used by setsockopt. 887 * The first hop is placed before the options, will be removed later. 888 */ 889 struct mbuf * 890 ip_srcroute() 891 { 892 register struct in_addr *p, *q; 893 register struct mbuf *m; 894 895 if (ip_nhops == 0) 896 return ((struct mbuf *)0); 897 m = m_get(M_DONTWAIT, MT_SOOPTS); 898 if (m == 0) 899 return ((struct mbuf *)0); 900 901 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 902 903 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 904 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 905 OPTSIZ; 906 #ifdef DIAGNOSTIC 907 if (ipprintfs) 908 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 909 #endif 910 911 /* 912 * First save first hop for return route 913 */ 914 p = &ip_srcrt.route[ip_nhops - 1]; 915 *(mtod(m, struct in_addr *)) = *p--; 916 #ifdef DIAGNOSTIC 917 if (ipprintfs) 918 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 919 #endif 920 921 /* 922 * Copy option fields and padding (nop) to mbuf. 923 */ 924 ip_srcrt.nop = IPOPT_NOP; 925 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 926 bcopy((caddr_t)&ip_srcrt.nop, 927 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 928 q = (struct in_addr *)(mtod(m, caddr_t) + 929 sizeof(struct in_addr) + OPTSIZ); 930 #undef OPTSIZ 931 /* 932 * Record return path as an IP source route, 933 * reversing the path (pointers are now aligned). 934 */ 935 while (p >= ip_srcrt.route) { 936 #ifdef DIAGNOSTIC 937 if (ipprintfs) 938 printf(" %lx", ntohl(q->s_addr)); 939 #endif 940 *q++ = *p--; 941 } 942 /* 943 * Last hop goes to final destination. 944 */ 945 *q = ip_srcrt.dst; 946 #ifdef DIAGNOSTIC 947 if (ipprintfs) 948 printf(" %lx\n", ntohl(q->s_addr)); 949 #endif 950 return (m); 951 } 952 953 /* 954 * Strip out IP options, at higher 955 * level protocol in the kernel. 956 * Second argument is buffer to which options 957 * will be moved, and return value is their length. 958 * XXX should be deleted; last arg currently ignored. 959 */ 960 void 961 ip_stripoptions(m, mopt) 962 register struct mbuf *m; 963 struct mbuf *mopt; 964 { 965 register int i; 966 struct ip *ip = mtod(m, struct ip *); 967 register caddr_t opts; 968 int olen; 969 970 olen = (ip->ip_hl<<2) - sizeof (struct ip); 971 opts = (caddr_t)(ip + 1); 972 i = m->m_len - (sizeof (struct ip) + olen); 973 bcopy(opts + olen, opts, (unsigned)i); 974 m->m_len -= olen; 975 if (m->m_flags & M_PKTHDR) 976 m->m_pkthdr.len -= olen; 977 ip->ip_hl = sizeof(struct ip) >> 2; 978 } 979 980 u_char inetctlerrmap[PRC_NCMDS] = { 981 0, 0, 0, 0, 982 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 983 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 984 EMSGSIZE, EHOSTUNREACH, 0, 0, 985 0, 0, 0, 0, 986 ENOPROTOOPT 987 }; 988 989 /* 990 * Forward a packet. If some error occurs return the sender 991 * an icmp packet. Note we can't always generate a meaningful 992 * icmp message because icmp doesn't have a large enough repertoire 993 * of codes and types. 994 * 995 * If not forwarding, just drop the packet. This could be confusing 996 * if ipforwarding was zero but some routing protocol was advancing 997 * us as a gateway to somewhere. However, we must let the routing 998 * protocol deal with that. 999 * 1000 * The srcrt parameter indicates whether the packet is being forwarded 1001 * via a source route. 1002 */ 1003 void 1004 ip_forward(m, srcrt) 1005 struct mbuf *m; 1006 int srcrt; 1007 { 1008 register struct ip *ip = mtod(m, struct ip *); 1009 register struct sockaddr_in *sin; 1010 register struct rtentry *rt; 1011 int error, type = 0, code; 1012 struct mbuf *mcopy; 1013 n_long dest; 1014 struct ifnet *destifp; 1015 1016 dest = 0; 1017 #ifdef DIAGNOSTIC 1018 if (ipprintfs) 1019 printf("forward: src %x dst %x ttl %x\n", ip->ip_src, 1020 ip->ip_dst, ip->ip_ttl); 1021 #endif 1022 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1023 ipstat.ips_cantforward++; 1024 m_freem(m); 1025 return; 1026 } 1027 HTONS(ip->ip_id); 1028 if (ip->ip_ttl <= IPTTLDEC) { 1029 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1030 return; 1031 } 1032 ip->ip_ttl -= IPTTLDEC; 1033 1034 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1035 if ((rt = ipforward_rt.ro_rt) == 0 || 1036 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1037 if (ipforward_rt.ro_rt) { 1038 RTFREE(ipforward_rt.ro_rt); 1039 ipforward_rt.ro_rt = 0; 1040 } 1041 sin->sin_family = AF_INET; 1042 sin->sin_len = sizeof(*sin); 1043 sin->sin_addr = ip->ip_dst; 1044 1045 rtalloc(&ipforward_rt); 1046 if (ipforward_rt.ro_rt == 0) { 1047 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1048 return; 1049 } 1050 rt = ipforward_rt.ro_rt; 1051 } 1052 1053 /* 1054 * Save at most 64 bytes of the packet in case 1055 * we need to generate an ICMP message to the src. 1056 */ 1057 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1058 1059 #ifdef GATEWAY 1060 ip_ifmatrix[rt->rt_ifp->if_index + 1061 if_index * m->m_pkthdr.rcvif->if_index]++; 1062 #endif 1063 /* 1064 * If forwarding packet using same interface that it came in on, 1065 * perhaps should send a redirect to sender to shortcut a hop. 1066 * Only send redirect if source is sending directly to us, 1067 * and if packet was not source routed (or has any options). 1068 * Also, don't send redirect if forwarding using a default route 1069 * or a route modified by a redirect. 1070 */ 1071 #define satosin(sa) ((struct sockaddr_in *)(sa)) 1072 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1073 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1074 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1075 ipsendredirects && !srcrt) { 1076 #define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1077 u_int32_t src = ntohl(ip->ip_src.s_addr); 1078 1079 if (RTA(rt) && 1080 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1081 if (rt->rt_flags & RTF_GATEWAY) 1082 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1083 else 1084 dest = ip->ip_dst.s_addr; 1085 /* Router requirements says to only send host redirects */ 1086 type = ICMP_REDIRECT; 1087 code = ICMP_REDIRECT_HOST; 1088 #ifdef DIAGNOSTIC 1089 if (ipprintfs) 1090 printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); 1091 #endif 1092 } 1093 } 1094 1095 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING 1096 #ifdef DIRECTED_BROADCAST 1097 | IP_ALLOWBROADCAST 1098 #endif 1099 , 0); 1100 if (error) 1101 ipstat.ips_cantforward++; 1102 else { 1103 ipstat.ips_forward++; 1104 if (type) 1105 ipstat.ips_redirectsent++; 1106 else { 1107 if (mcopy) 1108 m_freem(mcopy); 1109 return; 1110 } 1111 } 1112 if (mcopy == NULL) 1113 return; 1114 destifp = NULL; 1115 1116 switch (error) { 1117 1118 case 0: /* forwarded, but need redirect */ 1119 /* type, code set above */ 1120 break; 1121 1122 case ENETUNREACH: /* shouldn't happen, checked above */ 1123 case EHOSTUNREACH: 1124 case ENETDOWN: 1125 case EHOSTDOWN: 1126 default: 1127 type = ICMP_UNREACH; 1128 code = ICMP_UNREACH_HOST; 1129 break; 1130 1131 case EMSGSIZE: 1132 type = ICMP_UNREACH; 1133 code = ICMP_UNREACH_NEEDFRAG; 1134 if (ipforward_rt.ro_rt) 1135 destifp = ipforward_rt.ro_rt->rt_ifp; 1136 ipstat.ips_cantfrag++; 1137 break; 1138 1139 case ENOBUFS: 1140 type = ICMP_SOURCEQUENCH; 1141 code = 0; 1142 break; 1143 } 1144 icmp_error(mcopy, type, code, dest, destifp); 1145 } 1146 1147 int 1148 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1149 int *name; 1150 u_int namelen; 1151 void *oldp; 1152 size_t *oldlenp; 1153 void *newp; 1154 size_t newlen; 1155 { 1156 /* All sysctl names at this level are terminal. */ 1157 if (namelen != 1) 1158 return (ENOTDIR); 1159 1160 switch (name[0]) { 1161 case IPCTL_FORWARDING: 1162 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1163 case IPCTL_SENDREDIRECTS: 1164 return (sysctl_int(oldp, oldlenp, newp, newlen, 1165 &ipsendredirects)); 1166 case IPCTL_DEFTTL: 1167 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1168 #ifdef notyet 1169 case IPCTL_DEFMTU: 1170 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1171 #endif 1172 default: 1173 return (EOPNOTSUPP); 1174 } 1175 /* NOTREACHED */ 1176 } 1177