ip_input.c revision 19183
1/* 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 34 * $Id: ip_input.c,v 1.49 1996/10/22 22:25:58 sos Exp $ 35 * $ANA: ip_input.c,v 1.5 1996/09/18 14:34:59 wollman Exp $ 36 */ 37 38#define _IP_VHL 39 40#include "opt_ipfw.h" 41 42#include <sys/param.h> 43#include <sys/systm.h> 44#include <sys/malloc.h> 45#include <sys/mbuf.h> 46#include <sys/domain.h> 47#include <sys/protosw.h> 48#include <sys/socket.h> 49#include <sys/errno.h> 50#include <sys/time.h> 51#include <sys/kernel.h> 52#include <sys/syslog.h> 53#include <sys/sysctl.h> 54 55#include <net/if.h> 56#include <net/route.h> 57#include <net/netisr.h> 58 59#include <netinet/in.h> 60#include <netinet/in_systm.h> 61#include <netinet/in_var.h> 62#include <netinet/ip.h> 63#include <netinet/in_pcb.h> 64#include <netinet/in_var.h> 65#include <netinet/ip_var.h> 66#include <netinet/ip_icmp.h> 67#include <machine/in_cksum.h> 68 69#include <sys/socketvar.h> 70 71#ifdef IPFIREWALL 72#include <netinet/ip_fw.h> 73#endif 74 75int rsvp_on = 0; 76static int ip_rsvp_on; 77struct socket *ip_rsvpd; 78 79static int ipforwarding = 0; 80SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW, 81 &ipforwarding, 0, ""); 82 83static int ipsendredirects = 1; /* XXX */ 84SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW, 85 &ipsendredirects, 0, ""); 86 87int ip_defttl = IPDEFTTL; 88SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW, 89 &ip_defttl, 0, ""); 90 91static int ip_dosourceroute = 0; 92SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW, 93 &ip_dosourceroute, 0, ""); 94#ifdef DIAGNOSTIC 95static int ipprintfs = 0; 96#endif 97 98extern struct domain inetdomain; 99extern struct protosw inetsw[]; 100u_char ip_protox[IPPROTO_MAX]; 101static int ipqmaxlen = IFQ_MAXLEN; 102struct in_ifaddr *in_ifaddr; /* first inet address */ 103struct ifqueue ipintrq; 104SYSCTL_INT(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLFLAG_RD, 105 &ipintrq.ifq_maxlen, 0, ""); 106SYSCTL_INT(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLFLAG_RD, 107 &ipintrq.ifq_drops, 0, ""); 108 109struct ipstat ipstat; 110static struct ipq ipq; 111 112#ifdef IPCTL_DEFMTU 113SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW, 114 &ip_mtu, 0, ""); 115#endif 116 117#if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1 118#undef COMPAT_IPFW 119#define COMPAT_IPFW 1 120#else 121#undef COMPAT_IPFW 122#endif 123 124#ifdef COMPAT_IPFW 125/* Firewall hooks */ 126ip_fw_chk_t *ip_fw_chk_ptr; 127ip_fw_ctl_t *ip_fw_ctl_ptr; 128 129/* IP Network Address Translation (NAT) hooks */ 130ip_nat_t *ip_nat_ptr; 131ip_nat_ctl_t *ip_nat_ctl_ptr; 132#endif 133 134/* 135 * We need to save the IP options in case a protocol wants to respond 136 * to an incoming packet over the same route if the packet got here 137 * using IP source routing. This allows connection establishment and 138 * maintenance when the remote end is on a network that is not known 139 * to us. 140 */ 141static int ip_nhops = 0; 142static struct ip_srcrt { 143 struct in_addr dst; /* final destination */ 144 char nop; /* one NOP to align */ 145 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 146 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 147} ip_srcrt; 148 149#ifdef IPDIVERT 150/* 151 * Shared variable between ip_input() and ip_reass() to communicate 152 * about which packets, once assembled from fragments, get diverted, 153 * and to which port. 154 */ 155static u_short frag_divert_port; 156#endif 157 158static void save_rte __P((u_char *, struct in_addr)); 159static void ip_deq __P((struct ipasfrag *)); 160static int ip_dooptions __P((struct mbuf *)); 161static void ip_enq __P((struct ipasfrag *, struct ipasfrag *)); 162static void ip_forward __P((struct mbuf *, int)); 163static void ip_freef __P((struct ipq *)); 164static struct ip * 165 ip_reass __P((struct ipasfrag *, struct ipq *)); 166static struct in_ifaddr * 167 ip_rtaddr __P((struct in_addr)); 168static void ipintr __P((void)); 169/* 170 * IP initialization: fill in IP protocol switch table. 171 * All protocols not implemented in kernel go to raw IP protocol handler. 172 */ 173void 174ip_init() 175{ 176 register struct protosw *pr; 177 register int i; 178 179 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 180 if (pr == 0) 181 panic("ip_init"); 182 for (i = 0; i < IPPROTO_MAX; i++) 183 ip_protox[i] = pr - inetsw; 184 for (pr = inetdomain.dom_protosw; 185 pr < inetdomain.dom_protoswNPROTOSW; pr++) 186 if (pr->pr_domain->dom_family == PF_INET && 187 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 188 ip_protox[pr->pr_protocol] = pr - inetsw; 189 ipq.next = ipq.prev = &ipq; 190 ip_id = time.tv_sec & 0xffff; 191 ipintrq.ifq_maxlen = ipqmaxlen; 192#ifdef IPFIREWALL 193 ip_fw_init(); 194#endif 195#ifdef IPNAT 196 ip_nat_init(); 197#endif 198 199} 200 201static struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 202static struct route ipforward_rt; 203 204/* 205 * Ip input routine. Checksum and byte swap header. If fragmented 206 * try to reassemble. Process options. Pass to next level. 207 */ 208void 209ip_input(struct mbuf *m) 210{ 211 struct ip *ip; 212 struct ipq *fp; 213 struct in_ifaddr *ia; 214 int hlen; 215 216#ifdef DIAGNOSTIC 217 if ((m->m_flags & M_PKTHDR) == 0) 218 panic("ip_input no HDR"); 219#endif 220 /* 221 * If no IP addresses have been set yet but the interfaces 222 * are receiving, can't do anything with incoming packets yet. 223 */ 224 if (in_ifaddr == NULL) 225 goto bad; 226 ipstat.ips_total++; 227 228 if (m->m_pkthdr.len < sizeof(struct ip)) 229 goto tooshort; 230 231#ifdef DIAGNOSTIC 232 if (m->m_len < sizeof(struct ip)) 233 panic("ipintr mbuf too short"); 234#endif 235 236 if (m->m_len < sizeof (struct ip) && 237 (m = m_pullup(m, sizeof (struct ip))) == 0) { 238 ipstat.ips_toosmall++; 239 return; 240 } 241 ip = mtod(m, struct ip *); 242 243 if (IP_VHL_V(ip->ip_vhl) != IPVERSION) { 244 ipstat.ips_badvers++; 245 goto bad; 246 } 247 248 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 249 if (hlen < sizeof(struct ip)) { /* minimum header length */ 250 ipstat.ips_badhlen++; 251 goto bad; 252 } 253 if (hlen > m->m_len) { 254 if ((m = m_pullup(m, hlen)) == 0) { 255 ipstat.ips_badhlen++; 256 return; 257 } 258 ip = mtod(m, struct ip *); 259 } 260 if (hlen == sizeof(struct ip)) { 261 ip->ip_sum = in_cksum_hdr(ip); 262 } else { 263 ip->ip_sum = in_cksum(m, hlen); 264 } 265 if (ip->ip_sum) { 266 ipstat.ips_badsum++; 267 goto bad; 268 } 269 270 /* 271 * Convert fields to host representation. 272 */ 273 NTOHS(ip->ip_len); 274 if (ip->ip_len < hlen) { 275 ipstat.ips_badlen++; 276 goto bad; 277 } 278 NTOHS(ip->ip_id); 279 NTOHS(ip->ip_off); 280 281 /* 282 * Check that the amount of data in the buffers 283 * is as at least much as the IP header would have us expect. 284 * Trim mbufs if longer than we expect. 285 * Drop packet if shorter than we expect. 286 */ 287 if (m->m_pkthdr.len < ip->ip_len) { 288tooshort: 289 ipstat.ips_tooshort++; 290 goto bad; 291 } 292 if (m->m_pkthdr.len > ip->ip_len) { 293 if (m->m_len == m->m_pkthdr.len) { 294 m->m_len = ip->ip_len; 295 m->m_pkthdr.len = ip->ip_len; 296 } else 297 m_adj(m, ip->ip_len - m->m_pkthdr.len); 298 } 299 /* 300 * IpHack's section. 301 * Right now when no processing on packet has done 302 * and it is still fresh out of network we do our black 303 * deals with it. 304 * - Firewall: deny/allow/divert 305 * - Xlate: translate packet's addr/port (NAT). 306 * - Wrap: fake packet's addr/port <unimpl.> 307 * - Encapsulate: put it in another IP and send out. <unimp.> 308 */ 309 310#ifdef COMPAT_IPFW 311 if (ip_fw_chk_ptr) { 312 int action; 313 314#ifdef IPDIVERT 315 action = (*ip_fw_chk_ptr)(&ip, hlen, 316 m->m_pkthdr.rcvif, ip_divert_ignore, &m); 317#else 318 action = (*ip_fw_chk_ptr)(&ip, hlen, m->m_pkthdr.rcvif, 0, &m); 319#endif 320 if (action == -1) 321 return; 322 if (action != 0) { 323#ifdef IPDIVERT 324 frag_divert_port = action; 325 goto ours; 326#else 327 goto bad; /* ipfw said divert but we can't */ 328#endif 329 } 330 } 331 332 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, m->m_pkthdr.rcvif, IP_NAT_IN)) 333 return; 334#endif 335 336 /* 337 * Process options and, if not destined for us, 338 * ship it on. ip_dooptions returns 1 when an 339 * error was detected (causing an icmp message 340 * to be sent and the original packet to be freed). 341 */ 342 ip_nhops = 0; /* for source routed packets */ 343 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 344 return; 345 346 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no 347 * matter if it is destined to another node, or whether it is 348 * a multicast one, RSVP wants it! and prevents it from being forwarded 349 * anywhere else. Also checks if the rsvp daemon is running before 350 * grabbing the packet. 351 */ 352 if (rsvp_on && ip->ip_p==IPPROTO_RSVP) 353 goto ours; 354 355 /* 356 * Check our list of addresses, to see if the packet is for us. 357 */ 358 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 359#define satosin(sa) ((struct sockaddr_in *)(sa)) 360 361 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 362 goto ours; 363 if (ia->ia_ifp && ia->ia_ifp->if_flags & IFF_BROADCAST) { 364 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 365 ip->ip_dst.s_addr) 366 goto ours; 367 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 368 goto ours; 369 } 370 } 371 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 372 struct in_multi *inm; 373 if (ip_mrouter) { 374 /* 375 * If we are acting as a multicast router, all 376 * incoming multicast packets are passed to the 377 * kernel-level multicast forwarding function. 378 * The packet is returned (relatively) intact; if 379 * ip_mforward() returns a non-zero value, the packet 380 * must be discarded, else it may be accepted below. 381 * 382 * (The IP ident field is put in the same byte order 383 * as expected when ip_mforward() is called from 384 * ip_output().) 385 */ 386 ip->ip_id = htons(ip->ip_id); 387 if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) { 388 ipstat.ips_cantforward++; 389 m_freem(m); 390 return; 391 } 392 ip->ip_id = ntohs(ip->ip_id); 393 394 /* 395 * The process-level routing demon needs to receive 396 * all multicast IGMP packets, whether or not this 397 * host belongs to their destination groups. 398 */ 399 if (ip->ip_p == IPPROTO_IGMP) 400 goto ours; 401 ipstat.ips_forward++; 402 } 403 /* 404 * See if we belong to the destination multicast group on the 405 * arrival interface. 406 */ 407 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 408 if (inm == NULL) { 409 ipstat.ips_cantforward++; 410 m_freem(m); 411 return; 412 } 413 goto ours; 414 } 415 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 416 goto ours; 417 if (ip->ip_dst.s_addr == INADDR_ANY) 418 goto ours; 419 420 /* 421 * Not for us; forward if possible and desirable. 422 */ 423 if (ipforwarding == 0) { 424 ipstat.ips_cantforward++; 425 m_freem(m); 426 } else 427 ip_forward(m, 0); 428 return; 429 430ours: 431 432 /* 433 * If offset or IP_MF are set, must reassemble. 434 * Otherwise, nothing need be done. 435 * (We could look in the reassembly queue to see 436 * if the packet was previously fragmented, 437 * but it's not worth the time; just let them time out.) 438 */ 439 if (ip->ip_off &~ IP_DF) { 440 if (m->m_flags & M_EXT) { /* XXX */ 441 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 442 ipstat.ips_toosmall++; 443#ifdef IPDIVERT 444 frag_divert_port = 0; 445#endif 446 return; 447 } 448 ip = mtod(m, struct ip *); 449 } 450 /* 451 * Look for queue of fragments 452 * of this datagram. 453 */ 454 for (fp = ipq.next; fp != &ipq; fp = fp->next) 455 if (ip->ip_id == fp->ipq_id && 456 ip->ip_src.s_addr == fp->ipq_src.s_addr && 457 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 458 ip->ip_p == fp->ipq_p) 459 goto found; 460 fp = 0; 461found: 462 463 /* 464 * Adjust ip_len to not reflect header, 465 * set ip_mff if more fragments are expected, 466 * convert offset of this to bytes. 467 */ 468 ip->ip_len -= hlen; 469 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 470 if (ip->ip_off & IP_MF) 471 ((struct ipasfrag *)ip)->ipf_mff |= 1; 472 ip->ip_off <<= 3; 473 474 /* 475 * If datagram marked as having more fragments 476 * or if this is not the first fragment, 477 * attempt reassembly; if it succeeds, proceed. 478 */ 479 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 480 ipstat.ips_fragments++; 481 ip = ip_reass((struct ipasfrag *)ip, fp); 482 if (ip == 0) 483 return; 484 ipstat.ips_reassembled++; 485 m = dtom(ip); 486 } else 487 if (fp) 488 ip_freef(fp); 489 } else 490 ip->ip_len -= hlen; 491 492#ifdef IPDIVERT 493 /* 494 * Divert packets here to the divert protocol if required 495 */ 496 if (frag_divert_port) { 497 ip_divert_port = frag_divert_port; 498 frag_divert_port = 0; 499 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, hlen); 500 return; 501 } 502#endif 503 504 /* 505 * Switch out to protocol's input routine. 506 */ 507 ipstat.ips_delivered++; 508 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 509 return; 510bad: 511 m_freem(m); 512} 513 514/* 515 * IP software interrupt routine - to go away sometime soon 516 */ 517static void 518ipintr(void) 519{ 520 int s; 521 struct mbuf *m; 522 523 while(1) { 524 s = splimp(); 525 IF_DEQUEUE(&ipintrq, m); 526 splx(s); 527 if (m == 0) 528 return; 529 ip_input(m); 530 } 531} 532 533NETISR_SET(NETISR_IP, ipintr); 534 535/* 536 * Take incoming datagram fragment and try to 537 * reassemble it into whole datagram. If a chain for 538 * reassembly of this datagram already exists, then it 539 * is given as fp; otherwise have to make a chain. 540 */ 541static struct ip * 542ip_reass(ip, fp) 543 register struct ipasfrag *ip; 544 register struct ipq *fp; 545{ 546 register struct mbuf *m = dtom(ip); 547 register struct ipasfrag *q; 548 struct mbuf *t; 549 int hlen = ip->ip_hl << 2; 550 int i, next; 551 552 /* 553 * Presence of header sizes in mbufs 554 * would confuse code below. 555 */ 556 m->m_data += hlen; 557 m->m_len -= hlen; 558 559 /* 560 * If first fragment to arrive, create a reassembly queue. 561 */ 562 if (fp == 0) { 563 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 564 goto dropfrag; 565 fp = mtod(t, struct ipq *); 566 insque(fp, &ipq); 567 fp->ipq_ttl = IPFRAGTTL; 568 fp->ipq_p = ip->ip_p; 569 fp->ipq_id = ip->ip_id; 570 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 571 fp->ipq_src = ((struct ip *)ip)->ip_src; 572 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 573#ifdef IPDIVERT 574 fp->ipq_divert = 0; 575#endif 576 q = (struct ipasfrag *)fp; 577 goto insert; 578 } 579 580 /* 581 * Find a segment which begins after this one does. 582 */ 583 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 584 if (q->ip_off > ip->ip_off) 585 break; 586 587 /* 588 * If there is a preceding segment, it may provide some of 589 * our data already. If so, drop the data from the incoming 590 * segment. If it provides all of our data, drop us. 591 */ 592 if (q->ipf_prev != (struct ipasfrag *)fp) { 593 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 594 if (i > 0) { 595 if (i >= ip->ip_len) 596 goto dropfrag; 597 m_adj(dtom(ip), i); 598 ip->ip_off += i; 599 ip->ip_len -= i; 600 } 601 } 602 603 /* 604 * While we overlap succeeding segments trim them or, 605 * if they are completely covered, dequeue them. 606 */ 607 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 608 struct mbuf *m0; 609 610 i = (ip->ip_off + ip->ip_len) - q->ip_off; 611 if (i < q->ip_len) { 612 q->ip_len -= i; 613 q->ip_off += i; 614 m_adj(dtom(q), i); 615 break; 616 } 617 m0 = dtom(q); 618 q = q->ipf_next; 619 ip_deq(q->ipf_prev); 620 m_freem(m0); 621 } 622 623insert: 624 625#ifdef IPDIVERT 626 /* 627 * Any fragment diverting causes the whole packet to divert 628 */ 629 if (frag_divert_port != 0) 630 fp->ipq_divert = frag_divert_port; 631 frag_divert_port = 0; 632#endif 633 634 /* 635 * Stick new segment in its place; 636 * check for complete reassembly. 637 */ 638 ip_enq(ip, q->ipf_prev); 639 next = 0; 640 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 641 if (q->ip_off != next) 642 return (0); 643 next += q->ip_len; 644 } 645 if (q->ipf_prev->ipf_mff & 1) 646 return (0); 647 648 /* 649 * Reassembly is complete. Make sure the packet is a sane size. 650 */ 651 if (next + (IP_VHL_HL(((struct ip *)fp->ipq_next)->ip_vhl) << 2) 652 > IP_MAXPACKET) { 653 ipstat.ips_toolong++; 654 ip_freef(fp); 655 return (0); 656 } 657 658 /* 659 * Concatenate fragments. 660 */ 661 q = fp->ipq_next; 662 m = dtom(q); 663 t = m->m_next; 664 m->m_next = 0; 665 m_cat(m, t); 666 q = q->ipf_next; 667 while (q != (struct ipasfrag *)fp) { 668 t = dtom(q); 669 q = q->ipf_next; 670 m_cat(m, t); 671 } 672 673#ifdef IPDIVERT 674 /* 675 * Record divert port for packet, if any 676 */ 677 frag_divert_port = fp->ipq_divert; 678#endif 679 680 /* 681 * Create header for new ip packet by 682 * modifying header of first packet; 683 * dequeue and discard fragment reassembly header. 684 * Make header visible. 685 */ 686 ip = fp->ipq_next; 687 ip->ip_len = next; 688 ip->ipf_mff &= ~1; 689 ((struct ip *)ip)->ip_src = fp->ipq_src; 690 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 691 remque(fp); 692 (void) m_free(dtom(fp)); 693 m = dtom(ip); 694 m->m_len += (ip->ip_hl << 2); 695 m->m_data -= (ip->ip_hl << 2); 696 /* some debugging cruft by sklower, below, will go away soon */ 697 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 698 register int plen = 0; 699 for (t = m; m; m = m->m_next) 700 plen += m->m_len; 701 t->m_pkthdr.len = plen; 702 } 703 return ((struct ip *)ip); 704 705dropfrag: 706 ipstat.ips_fragdropped++; 707 m_freem(m); 708 return (0); 709} 710 711/* 712 * Free a fragment reassembly header and all 713 * associated datagrams. 714 */ 715static void 716ip_freef(fp) 717 struct ipq *fp; 718{ 719 register struct ipasfrag *q, *p; 720 721 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 722 p = q->ipf_next; 723 ip_deq(q); 724 m_freem(dtom(q)); 725 } 726 remque(fp); 727 (void) m_free(dtom(fp)); 728} 729 730/* 731 * Put an ip fragment on a reassembly chain. 732 * Like insque, but pointers in middle of structure. 733 */ 734static void 735ip_enq(p, prev) 736 register struct ipasfrag *p, *prev; 737{ 738 739 p->ipf_prev = prev; 740 p->ipf_next = prev->ipf_next; 741 prev->ipf_next->ipf_prev = p; 742 prev->ipf_next = p; 743} 744 745/* 746 * To ip_enq as remque is to insque. 747 */ 748static void 749ip_deq(p) 750 register struct ipasfrag *p; 751{ 752 753 p->ipf_prev->ipf_next = p->ipf_next; 754 p->ipf_next->ipf_prev = p->ipf_prev; 755} 756 757/* 758 * IP timer processing; 759 * if a timer expires on a reassembly 760 * queue, discard it. 761 */ 762void 763ip_slowtimo() 764{ 765 register struct ipq *fp; 766 int s = splnet(); 767 768 fp = ipq.next; 769 if (fp == 0) { 770 splx(s); 771 return; 772 } 773 while (fp != &ipq) { 774 --fp->ipq_ttl; 775 fp = fp->next; 776 if (fp->prev->ipq_ttl == 0) { 777 ipstat.ips_fragtimeout++; 778 ip_freef(fp->prev); 779 } 780 } 781 splx(s); 782} 783 784/* 785 * Drain off all datagram fragments. 786 */ 787void 788ip_drain() 789{ 790 while (ipq.next != &ipq) { 791 ipstat.ips_fragdropped++; 792 ip_freef(ipq.next); 793 } 794 795 in_rtqdrain(); 796} 797 798/* 799 * Do option processing on a datagram, 800 * possibly discarding it if bad options are encountered, 801 * or forwarding it if source-routed. 802 * Returns 1 if packet has been forwarded/freed, 803 * 0 if the packet should be processed further. 804 */ 805static int 806ip_dooptions(m) 807 struct mbuf *m; 808{ 809 register struct ip *ip = mtod(m, struct ip *); 810 register u_char *cp; 811 register struct ip_timestamp *ipt; 812 register struct in_ifaddr *ia; 813 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 814 struct in_addr *sin, dst; 815 n_time ntime; 816 817 dst = ip->ip_dst; 818 cp = (u_char *)(ip + 1); 819 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 820 for (; cnt > 0; cnt -= optlen, cp += optlen) { 821 opt = cp[IPOPT_OPTVAL]; 822 if (opt == IPOPT_EOL) 823 break; 824 if (opt == IPOPT_NOP) 825 optlen = 1; 826 else { 827 optlen = cp[IPOPT_OLEN]; 828 if (optlen <= 0 || optlen > cnt) { 829 code = &cp[IPOPT_OLEN] - (u_char *)ip; 830 goto bad; 831 } 832 } 833 switch (opt) { 834 835 default: 836 break; 837 838 /* 839 * Source routing with record. 840 * Find interface with current destination address. 841 * If none on this machine then drop if strictly routed, 842 * or do nothing if loosely routed. 843 * Record interface address and bring up next address 844 * component. If strictly routed make sure next 845 * address is on directly accessible net. 846 */ 847 case IPOPT_LSRR: 848 case IPOPT_SSRR: 849 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 850 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 851 goto bad; 852 } 853 ipaddr.sin_addr = ip->ip_dst; 854 ia = (struct in_ifaddr *) 855 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 856 if (ia == 0) { 857 if (opt == IPOPT_SSRR) { 858 type = ICMP_UNREACH; 859 code = ICMP_UNREACH_SRCFAIL; 860 goto bad; 861 } 862 /* 863 * Loose routing, and not at next destination 864 * yet; nothing to do except forward. 865 */ 866 break; 867 } 868 off--; /* 0 origin */ 869 if (off > optlen - sizeof(struct in_addr)) { 870 /* 871 * End of source route. Should be for us. 872 */ 873 save_rte(cp, ip->ip_src); 874 break; 875 } 876 877 if (!ip_dosourceroute) { 878 char buf[4*sizeof "123"]; 879 strcpy(buf, inet_ntoa(ip->ip_dst)); 880 881 log(LOG_WARNING, 882 "attempted source route from %s to %s\n", 883 inet_ntoa(ip->ip_src), buf); 884 type = ICMP_UNREACH; 885 code = ICMP_UNREACH_SRCFAIL; 886 goto bad; 887 } 888 889 /* 890 * locate outgoing interface 891 */ 892 (void)memcpy(&ipaddr.sin_addr, cp + off, 893 sizeof(ipaddr.sin_addr)); 894 895 if (opt == IPOPT_SSRR) { 896#define INA struct in_ifaddr * 897#define SA struct sockaddr * 898 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 899 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 900 } else 901 ia = ip_rtaddr(ipaddr.sin_addr); 902 if (ia == 0) { 903 type = ICMP_UNREACH; 904 code = ICMP_UNREACH_SRCFAIL; 905 goto bad; 906 } 907 ip->ip_dst = ipaddr.sin_addr; 908 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 909 sizeof(struct in_addr)); 910 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 911 /* 912 * Let ip_intr's mcast routing check handle mcast pkts 913 */ 914 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 915 break; 916 917 case IPOPT_RR: 918 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 919 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 920 goto bad; 921 } 922 /* 923 * If no space remains, ignore. 924 */ 925 off--; /* 0 origin */ 926 if (off > optlen - sizeof(struct in_addr)) 927 break; 928 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst, 929 sizeof(ipaddr.sin_addr)); 930 /* 931 * locate outgoing interface; if we're the destination, 932 * use the incoming interface (should be same). 933 */ 934 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 935 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 936 type = ICMP_UNREACH; 937 code = ICMP_UNREACH_HOST; 938 goto bad; 939 } 940 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 941 sizeof(struct in_addr)); 942 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 943 break; 944 945 case IPOPT_TS: 946 code = cp - (u_char *)ip; 947 ipt = (struct ip_timestamp *)cp; 948 if (ipt->ipt_len < 5) 949 goto bad; 950 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 951 if (++ipt->ipt_oflw == 0) 952 goto bad; 953 break; 954 } 955 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 956 switch (ipt->ipt_flg) { 957 958 case IPOPT_TS_TSONLY: 959 break; 960 961 case IPOPT_TS_TSANDADDR: 962 if (ipt->ipt_ptr + sizeof(n_time) + 963 sizeof(struct in_addr) > ipt->ipt_len) 964 goto bad; 965 ipaddr.sin_addr = dst; 966 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 967 m->m_pkthdr.rcvif); 968 if (ia == 0) 969 continue; 970 (void)memcpy(sin, &IA_SIN(ia)->sin_addr, 971 sizeof(struct in_addr)); 972 ipt->ipt_ptr += sizeof(struct in_addr); 973 break; 974 975 case IPOPT_TS_PRESPEC: 976 if (ipt->ipt_ptr + sizeof(n_time) + 977 sizeof(struct in_addr) > ipt->ipt_len) 978 goto bad; 979 (void)memcpy(&ipaddr.sin_addr, sin, 980 sizeof(struct in_addr)); 981 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 982 continue; 983 ipt->ipt_ptr += sizeof(struct in_addr); 984 break; 985 986 default: 987 goto bad; 988 } 989 ntime = iptime(); 990 (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime, 991 sizeof(n_time)); 992 ipt->ipt_ptr += sizeof(n_time); 993 } 994 } 995 if (forward) { 996 ip_forward(m, 1); 997 return (1); 998 } 999 return (0); 1000bad: 1001 ip->ip_len -= IP_VHL_HL(ip->ip_vhl) << 2; /* XXX icmp_error adds in hdr length */ 1002 icmp_error(m, type, code, 0, 0); 1003 ipstat.ips_badoptions++; 1004 return (1); 1005} 1006 1007/* 1008 * Given address of next destination (final or next hop), 1009 * return internet address info of interface to be used to get there. 1010 */ 1011static struct in_ifaddr * 1012ip_rtaddr(dst) 1013 struct in_addr dst; 1014{ 1015 register struct sockaddr_in *sin; 1016 1017 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 1018 1019 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 1020 if (ipforward_rt.ro_rt) { 1021 RTFREE(ipforward_rt.ro_rt); 1022 ipforward_rt.ro_rt = 0; 1023 } 1024 sin->sin_family = AF_INET; 1025 sin->sin_len = sizeof(*sin); 1026 sin->sin_addr = dst; 1027 1028 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1029 } 1030 if (ipforward_rt.ro_rt == 0) 1031 return ((struct in_ifaddr *)0); 1032 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 1033} 1034 1035/* 1036 * Save incoming source route for use in replies, 1037 * to be picked up later by ip_srcroute if the receiver is interested. 1038 */ 1039void 1040save_rte(option, dst) 1041 u_char *option; 1042 struct in_addr dst; 1043{ 1044 unsigned olen; 1045 1046 olen = option[IPOPT_OLEN]; 1047#ifdef DIAGNOSTIC 1048 if (ipprintfs) 1049 printf("save_rte: olen %d\n", olen); 1050#endif 1051 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1052 return; 1053 bcopy(option, ip_srcrt.srcopt, olen); 1054 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1055 ip_srcrt.dst = dst; 1056} 1057 1058/* 1059 * Retrieve incoming source route for use in replies, 1060 * in the same form used by setsockopt. 1061 * The first hop is placed before the options, will be removed later. 1062 */ 1063struct mbuf * 1064ip_srcroute() 1065{ 1066 register struct in_addr *p, *q; 1067 register struct mbuf *m; 1068 1069 if (ip_nhops == 0) 1070 return ((struct mbuf *)0); 1071 m = m_get(M_DONTWAIT, MT_SOOPTS); 1072 if (m == 0) 1073 return ((struct mbuf *)0); 1074 1075#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1076 1077 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1078 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1079 OPTSIZ; 1080#ifdef DIAGNOSTIC 1081 if (ipprintfs) 1082 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1083#endif 1084 1085 /* 1086 * First save first hop for return route 1087 */ 1088 p = &ip_srcrt.route[ip_nhops - 1]; 1089 *(mtod(m, struct in_addr *)) = *p--; 1090#ifdef DIAGNOSTIC 1091 if (ipprintfs) 1092 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 1093#endif 1094 1095 /* 1096 * Copy option fields and padding (nop) to mbuf. 1097 */ 1098 ip_srcrt.nop = IPOPT_NOP; 1099 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1100 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), 1101 &ip_srcrt.nop, OPTSIZ); 1102 q = (struct in_addr *)(mtod(m, caddr_t) + 1103 sizeof(struct in_addr) + OPTSIZ); 1104#undef OPTSIZ 1105 /* 1106 * Record return path as an IP source route, 1107 * reversing the path (pointers are now aligned). 1108 */ 1109 while (p >= ip_srcrt.route) { 1110#ifdef DIAGNOSTIC 1111 if (ipprintfs) 1112 printf(" %lx", ntohl(q->s_addr)); 1113#endif 1114 *q++ = *p--; 1115 } 1116 /* 1117 * Last hop goes to final destination. 1118 */ 1119 *q = ip_srcrt.dst; 1120#ifdef DIAGNOSTIC 1121 if (ipprintfs) 1122 printf(" %lx\n", ntohl(q->s_addr)); 1123#endif 1124 return (m); 1125} 1126 1127/* 1128 * Strip out IP options, at higher 1129 * level protocol in the kernel. 1130 * Second argument is buffer to which options 1131 * will be moved, and return value is their length. 1132 * XXX should be deleted; last arg currently ignored. 1133 */ 1134void 1135ip_stripoptions(m, mopt) 1136 register struct mbuf *m; 1137 struct mbuf *mopt; 1138{ 1139 register int i; 1140 struct ip *ip = mtod(m, struct ip *); 1141 register caddr_t opts; 1142 int olen; 1143 1144 olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 1145 opts = (caddr_t)(ip + 1); 1146 i = m->m_len - (sizeof (struct ip) + olen); 1147 bcopy(opts + olen, opts, (unsigned)i); 1148 m->m_len -= olen; 1149 if (m->m_flags & M_PKTHDR) 1150 m->m_pkthdr.len -= olen; 1151 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2); 1152} 1153 1154u_char inetctlerrmap[PRC_NCMDS] = { 1155 0, 0, 0, 0, 1156 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1157 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1158 EMSGSIZE, EHOSTUNREACH, 0, 0, 1159 0, 0, 0, 0, 1160 ENOPROTOOPT 1161}; 1162 1163/* 1164 * Forward a packet. If some error occurs return the sender 1165 * an icmp packet. Note we can't always generate a meaningful 1166 * icmp message because icmp doesn't have a large enough repertoire 1167 * of codes and types. 1168 * 1169 * If not forwarding, just drop the packet. This could be confusing 1170 * if ipforwarding was zero but some routing protocol was advancing 1171 * us as a gateway to somewhere. However, we must let the routing 1172 * protocol deal with that. 1173 * 1174 * The srcrt parameter indicates whether the packet is being forwarded 1175 * via a source route. 1176 */ 1177static void 1178ip_forward(m, srcrt) 1179 struct mbuf *m; 1180 int srcrt; 1181{ 1182 register struct ip *ip = mtod(m, struct ip *); 1183 register struct sockaddr_in *sin; 1184 register struct rtentry *rt; 1185 int error, type = 0, code = 0; 1186 struct mbuf *mcopy; 1187 n_long dest; 1188 struct ifnet *destifp; 1189 1190 dest = 0; 1191#ifdef DIAGNOSTIC 1192 if (ipprintfs) 1193 printf("forward: src %lx dst %lx ttl %x\n", 1194 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1195#endif 1196 1197 1198 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1199 ipstat.ips_cantforward++; 1200 m_freem(m); 1201 return; 1202 } 1203 HTONS(ip->ip_id); 1204 if (ip->ip_ttl <= IPTTLDEC) { 1205 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1206 return; 1207 } 1208 ip->ip_ttl -= IPTTLDEC; 1209 1210 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1211 if ((rt = ipforward_rt.ro_rt) == 0 || 1212 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1213 if (ipforward_rt.ro_rt) { 1214 RTFREE(ipforward_rt.ro_rt); 1215 ipforward_rt.ro_rt = 0; 1216 } 1217 sin->sin_family = AF_INET; 1218 sin->sin_len = sizeof(*sin); 1219 sin->sin_addr = ip->ip_dst; 1220 1221 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1222 if (ipforward_rt.ro_rt == 0) { 1223 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1224 return; 1225 } 1226 rt = ipforward_rt.ro_rt; 1227 } 1228 1229 /* 1230 * Save at most 64 bytes of the packet in case 1231 * we need to generate an ICMP message to the src. 1232 */ 1233 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1234 1235 /* 1236 * If forwarding packet using same interface that it came in on, 1237 * perhaps should send a redirect to sender to shortcut a hop. 1238 * Only send redirect if source is sending directly to us, 1239 * and if packet was not source routed (or has any options). 1240 * Also, don't send redirect if forwarding using a default route 1241 * or a route modified by a redirect. 1242 */ 1243#define satosin(sa) ((struct sockaddr_in *)(sa)) 1244 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1245 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1246 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1247 ipsendredirects && !srcrt) { 1248#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1249 u_long src = ntohl(ip->ip_src.s_addr); 1250 1251 if (RTA(rt) && 1252 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1253 if (rt->rt_flags & RTF_GATEWAY) 1254 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1255 else 1256 dest = ip->ip_dst.s_addr; 1257 /* Router requirements says to only send host redirects */ 1258 type = ICMP_REDIRECT; 1259 code = ICMP_REDIRECT_HOST; 1260#ifdef DIAGNOSTIC 1261 if (ipprintfs) 1262 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1263#endif 1264 } 1265 } 1266 1267 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1268 IP_FORWARDING, 0); 1269 if (error) 1270 ipstat.ips_cantforward++; 1271 else { 1272 ipstat.ips_forward++; 1273 if (type) 1274 ipstat.ips_redirectsent++; 1275 else { 1276 if (mcopy) 1277 m_freem(mcopy); 1278 return; 1279 } 1280 } 1281 if (mcopy == NULL) 1282 return; 1283 destifp = NULL; 1284 1285 switch (error) { 1286 1287 case 0: /* forwarded, but need redirect */ 1288 /* type, code set above */ 1289 break; 1290 1291 case ENETUNREACH: /* shouldn't happen, checked above */ 1292 case EHOSTUNREACH: 1293 case ENETDOWN: 1294 case EHOSTDOWN: 1295 default: 1296 type = ICMP_UNREACH; 1297 code = ICMP_UNREACH_HOST; 1298 break; 1299 1300 case EMSGSIZE: 1301 type = ICMP_UNREACH; 1302 code = ICMP_UNREACH_NEEDFRAG; 1303 if (ipforward_rt.ro_rt) 1304 destifp = ipforward_rt.ro_rt->rt_ifp; 1305 ipstat.ips_cantfrag++; 1306 break; 1307 1308 case ENOBUFS: 1309 type = ICMP_SOURCEQUENCH; 1310 code = 0; 1311 break; 1312 } 1313 icmp_error(mcopy, type, code, dest, destifp); 1314} 1315 1316int 1317ip_rsvp_init(struct socket *so) 1318{ 1319 if (so->so_type != SOCK_RAW || 1320 so->so_proto->pr_protocol != IPPROTO_RSVP) 1321 return EOPNOTSUPP; 1322 1323 if (ip_rsvpd != NULL) 1324 return EADDRINUSE; 1325 1326 ip_rsvpd = so; 1327 /* 1328 * This may seem silly, but we need to be sure we don't over-increment 1329 * the RSVP counter, in case something slips up. 1330 */ 1331 if (!ip_rsvp_on) { 1332 ip_rsvp_on = 1; 1333 rsvp_on++; 1334 } 1335 1336 return 0; 1337} 1338 1339int 1340ip_rsvp_done(void) 1341{ 1342 ip_rsvpd = NULL; 1343 /* 1344 * This may seem silly, but we need to be sure we don't over-decrement 1345 * the RSVP counter, in case something slips up. 1346 */ 1347 if (ip_rsvp_on) { 1348 ip_rsvp_on = 0; 1349 rsvp_on--; 1350 } 1351 return 0; 1352} 1353