ip_input.c revision 18160
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 1.46 1996/08/21 21:37:00 sos Exp $ 35 */ 36 37#include "opt_ipfw.h" 38 39#include <sys/param.h> 40#include <sys/systm.h> 41#include <sys/malloc.h> 42#include <sys/mbuf.h> 43#include <sys/domain.h> 44#include <sys/protosw.h> 45#include <sys/socket.h> 46#include <sys/errno.h> 47#include <sys/time.h> 48#include <sys/kernel.h> 49#include <sys/syslog.h> 50#include <sys/sysctl.h> 51 52#include <net/if.h> 53#include <net/route.h> 54#include <net/netisr.h> 55 56#include <netinet/in.h> 57#include <netinet/in_systm.h> 58#include <netinet/in_var.h> 59#include <netinet/ip.h> 60#include <netinet/in_pcb.h> 61#include <netinet/in_var.h> 62#include <netinet/ip_var.h> 63#include <netinet/ip_icmp.h> 64 65#include <sys/socketvar.h> 66 67#ifdef IPFIREWALL 68#include <netinet/ip_fw.h> 69#endif 70 71int rsvp_on = 0; 72static int ip_rsvp_on; 73struct socket *ip_rsvpd; 74 75static int ipforwarding = 0; 76SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW, 77 &ipforwarding, 0, ""); 78 79static int ipsendredirects = 1; /* XXX */ 80SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW, 81 &ipsendredirects, 0, ""); 82 83int ip_defttl = IPDEFTTL; 84SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW, 85 &ip_defttl, 0, ""); 86 87static int ip_dosourceroute = 0; 88SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW, 89 &ip_dosourceroute, 0, ""); 90#ifdef DIAGNOSTIC 91static int ipprintfs = 0; 92#endif 93 94extern struct domain inetdomain; 95extern struct protosw inetsw[]; 96u_char ip_protox[IPPROTO_MAX]; 97static int ipqmaxlen = IFQ_MAXLEN; 98struct in_ifaddr *in_ifaddr; /* first inet address */ 99struct ifqueue ipintrq; 100SYSCTL_INT(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLFLAG_RD, 101 &ipintrq.ifq_maxlen, 0, ""); 102SYSCTL_INT(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLFLAG_RD, 103 &ipintrq.ifq_drops, 0, ""); 104 105struct ipstat ipstat; 106static struct ipq ipq; 107 108#ifdef IPCTL_DEFMTU 109SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW, 110 &ip_mtu, 0, ""); 111#endif 112 113/* Firewall hooks */ 114ip_fw_chk_t *ip_fw_chk_ptr; 115ip_fw_ctl_t *ip_fw_ctl_ptr; 116 117/* IP Network Address Translation (NAT) hooks */ 118ip_nat_t *ip_nat_ptr; 119ip_nat_ctl_t *ip_nat_ctl_ptr; 120 121/* 122 * We need to save the IP options in case a protocol wants to respond 123 * to an incoming packet over the same route if the packet got here 124 * using IP source routing. This allows connection establishment and 125 * maintenance when the remote end is on a network that is not known 126 * to us. 127 */ 128static int ip_nhops = 0; 129static struct ip_srcrt { 130 struct in_addr dst; /* final destination */ 131 char nop; /* one NOP to align */ 132 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 133 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 134} ip_srcrt; 135 136#ifdef IPDIVERT 137/* 138 * Shared variable between ip_input() and ip_reass() to communicate 139 * about which packets, once assembled from fragments, get diverted, 140 * and to which port. 141 */ 142static u_short frag_divert_port; 143#endif 144 145static void save_rte __P((u_char *, struct in_addr)); 146static void ip_deq __P((struct ipasfrag *)); 147static int ip_dooptions __P((struct mbuf *)); 148static void ip_enq __P((struct ipasfrag *, struct ipasfrag *)); 149static void ip_forward __P((struct mbuf *, int)); 150static void ip_freef __P((struct ipq *)); 151static struct ip * 152 ip_reass __P((struct ipasfrag *, struct ipq *)); 153static struct in_ifaddr * 154 ip_rtaddr __P((struct in_addr)); 155static void ipintr __P((void)); 156/* 157 * IP initialization: fill in IP protocol switch table. 158 * All protocols not implemented in kernel go to raw IP protocol handler. 159 */ 160void 161ip_init() 162{ 163 register struct protosw *pr; 164 register int i; 165 166 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 167 if (pr == 0) 168 panic("ip_init"); 169 for (i = 0; i < IPPROTO_MAX; i++) 170 ip_protox[i] = pr - inetsw; 171 for (pr = inetdomain.dom_protosw; 172 pr < inetdomain.dom_protoswNPROTOSW; pr++) 173 if (pr->pr_domain->dom_family == PF_INET && 174 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 175 ip_protox[pr->pr_protocol] = pr - inetsw; 176 ipq.next = ipq.prev = &ipq; 177 ip_id = time.tv_sec & 0xffff; 178 ipintrq.ifq_maxlen = ipqmaxlen; 179#ifdef IPFIREWALL 180 ip_fw_init(); 181#endif 182#ifdef IPNAT 183 ip_nat_init(); 184#endif 185 186} 187 188static struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 189static struct route ipforward_rt; 190 191/* 192 * Ip input routine. Checksum and byte swap header. If fragmented 193 * try to reassemble. Process options. Pass to next level. 194 */ 195void 196ip_input(struct mbuf *m) 197{ 198 struct ip *ip; 199 struct ipq *fp; 200 struct in_ifaddr *ia; 201 int hlen; 202 203#ifdef DIAGNOSTIC 204 if ((m->m_flags & M_PKTHDR) == 0) 205 panic("ipintr no HDR"); 206#endif 207 /* 208 * If no IP addresses have been set yet but the interfaces 209 * are receiving, can't do anything with incoming packets yet. 210 */ 211 if (in_ifaddr == NULL) 212 goto bad; 213 ipstat.ips_total++; 214 if (m->m_len < sizeof (struct ip) && 215 (m = m_pullup(m, sizeof (struct ip))) == 0) { 216 ipstat.ips_toosmall++; 217 return; 218 } 219 ip = mtod(m, struct ip *); 220 if (ip->ip_v != IPVERSION) { 221 ipstat.ips_badvers++; 222 goto bad; 223 } 224 hlen = ip->ip_hl << 2; 225 if (hlen < sizeof(struct ip)) { /* minimum header length */ 226 ipstat.ips_badhlen++; 227 goto bad; 228 } 229 if (hlen > m->m_len) { 230 if ((m = m_pullup(m, hlen)) == 0) { 231 ipstat.ips_badhlen++; 232 return; 233 } 234 ip = mtod(m, struct ip *); 235 } 236 ip->ip_sum = in_cksum(m, hlen); 237 if (ip->ip_sum) { 238 ipstat.ips_badsum++; 239 goto bad; 240 } 241 242 /* 243 * Convert fields to host representation. 244 */ 245 NTOHS(ip->ip_len); 246 if (ip->ip_len < hlen) { 247 ipstat.ips_badlen++; 248 goto bad; 249 } 250 NTOHS(ip->ip_id); 251 NTOHS(ip->ip_off); 252 253 /* 254 * Check that the amount of data in the buffers 255 * is as at least much as the IP header would have us expect. 256 * Trim mbufs if longer than we expect. 257 * Drop packet if shorter than we expect. 258 */ 259 if (m->m_pkthdr.len < ip->ip_len) { 260 ipstat.ips_tooshort++; 261 goto bad; 262 } 263 if (m->m_pkthdr.len > ip->ip_len) { 264 if (m->m_len == m->m_pkthdr.len) { 265 m->m_len = ip->ip_len; 266 m->m_pkthdr.len = ip->ip_len; 267 } else 268 m_adj(m, ip->ip_len - m->m_pkthdr.len); 269 } 270 /* 271 * IpHack's section. 272 * Right now when no processing on packet has done 273 * and it is still fresh out of network we do our black 274 * deals with it. 275 * - Firewall: deny/allow/divert 276 * - Xlate: translate packet's addr/port (NAT). 277 * - Wrap: fake packet's addr/port <unimpl.> 278 * - Encapsulate: put it in another IP and send out. <unimp.> 279 */ 280 281 if (ip_fw_chk_ptr) { 282 int action; 283 284#ifdef IPDIVERT 285 action = (*ip_fw_chk_ptr)(&ip, hlen, 286 m->m_pkthdr.rcvif, ip_divert_ignore, &m); 287#else 288 action = (*ip_fw_chk_ptr)(&ip, hlen, m->m_pkthdr.rcvif, 0, &m); 289#endif 290 if (action == -1) 291 return; 292 if (action != 0) { 293#ifdef IPDIVERT 294 frag_divert_port = action; 295 goto ours; 296#else 297 goto bad; /* ipfw said divert but we can't */ 298#endif 299 } 300 } 301 302 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, IP_NAT_IN)) 303 return; 304 305 /* 306 * Process options and, if not destined for us, 307 * ship it on. ip_dooptions returns 1 when an 308 * error was detected (causing an icmp message 309 * to be sent and the original packet to be freed). 310 */ 311 ip_nhops = 0; /* for source routed packets */ 312 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 313 return; 314 315 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no 316 * matter if it is destined to another node, or whether it is 317 * a multicast one, RSVP wants it! and prevents it from being forwarded 318 * anywhere else. Also checks if the rsvp daemon is running before 319 * grabbing the packet. 320 */ 321 if (rsvp_on && ip->ip_p==IPPROTO_RSVP) 322 goto ours; 323 324 /* 325 * Check our list of addresses, to see if the packet is for us. 326 */ 327 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 328#define satosin(sa) ((struct sockaddr_in *)(sa)) 329 330 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 331 goto ours; 332 if (ia->ia_ifp && ia->ia_ifp->if_flags & IFF_BROADCAST) { 333#if 0 334 u_long t; 335#endif 336 337 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 338 ip->ip_dst.s_addr) 339 goto ours; 340 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 341 goto ours; 342#if 0 /* XXX - this should go away */ 343 /* 344 * Look for all-0's host part (old broadcast addr), 345 * either for subnet or net. 346 */ 347 t = ntohl(ip->ip_dst.s_addr); 348 if (t == ia->ia_subnet) 349 goto ours; 350 if (t == ia->ia_net) 351 goto ours; 352#endif /* compatibility cruft */ 353 } 354 } 355 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 356 struct in_multi *inm; 357 if (ip_mrouter) { 358 /* 359 * If we are acting as a multicast router, all 360 * incoming multicast packets are passed to the 361 * kernel-level multicast forwarding function. 362 * The packet is returned (relatively) intact; if 363 * ip_mforward() returns a non-zero value, the packet 364 * must be discarded, else it may be accepted below. 365 * 366 * (The IP ident field is put in the same byte order 367 * as expected when ip_mforward() is called from 368 * ip_output().) 369 */ 370 ip->ip_id = htons(ip->ip_id); 371 if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) { 372 ipstat.ips_cantforward++; 373 m_freem(m); 374 return; 375 } 376 ip->ip_id = ntohs(ip->ip_id); 377 378 /* 379 * The process-level routing demon needs to receive 380 * all multicast IGMP packets, whether or not this 381 * host belongs to their destination groups. 382 */ 383 if (ip->ip_p == IPPROTO_IGMP) 384 goto ours; 385 ipstat.ips_forward++; 386 } 387 /* 388 * See if we belong to the destination multicast group on the 389 * arrival interface. 390 */ 391 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 392 if (inm == NULL) { 393 ipstat.ips_cantforward++; 394 m_freem(m); 395 return; 396 } 397 goto ours; 398 } 399 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 400 goto ours; 401 if (ip->ip_dst.s_addr == INADDR_ANY) 402 goto ours; 403 404 /* 405 * Not for us; forward if possible and desirable. 406 */ 407 if (ipforwarding == 0) { 408 ipstat.ips_cantforward++; 409 m_freem(m); 410 } else 411 ip_forward(m, 0); 412 return; 413 414ours: 415 416 /* 417 * If offset or IP_MF are set, must reassemble. 418 * Otherwise, nothing need be done. 419 * (We could look in the reassembly queue to see 420 * if the packet was previously fragmented, 421 * but it's not worth the time; just let them time out.) 422 */ 423 if (ip->ip_off &~ IP_DF) { 424 if (m->m_flags & M_EXT) { /* XXX */ 425 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 426 ipstat.ips_toosmall++; 427#ifdef IPDIVERT 428 frag_divert_port = 0; 429#endif 430 return; 431 } 432 ip = mtod(m, struct ip *); 433 } 434 /* 435 * Look for queue of fragments 436 * of this datagram. 437 */ 438 for (fp = ipq.next; fp != &ipq; fp = fp->next) 439 if (ip->ip_id == fp->ipq_id && 440 ip->ip_src.s_addr == fp->ipq_src.s_addr && 441 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 442 ip->ip_p == fp->ipq_p) 443 goto found; 444 fp = 0; 445found: 446 447 /* 448 * Adjust ip_len to not reflect header, 449 * set ip_mff if more fragments are expected, 450 * convert offset of this to bytes. 451 */ 452 ip->ip_len -= hlen; 453 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 454 if (ip->ip_off & IP_MF) 455 ((struct ipasfrag *)ip)->ipf_mff |= 1; 456 ip->ip_off <<= 3; 457 458 /* 459 * If datagram marked as having more fragments 460 * or if this is not the first fragment, 461 * attempt reassembly; if it succeeds, proceed. 462 */ 463 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 464 ipstat.ips_fragments++; 465 ip = ip_reass((struct ipasfrag *)ip, fp); 466 if (ip == 0) 467 return; 468 ipstat.ips_reassembled++; 469 m = dtom(ip); 470 } else 471 if (fp) 472 ip_freef(fp); 473 } else 474 ip->ip_len -= hlen; 475 476#ifdef IPDIVERT 477 /* 478 * Divert packets here to the divert protocol if required 479 */ 480 if (frag_divert_port) { 481 ip_divert_port = frag_divert_port; 482 frag_divert_port = 0; 483 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, hlen); 484 return; 485 } 486#endif 487 488 /* 489 * Switch out to protocol's input routine. 490 */ 491 ipstat.ips_delivered++; 492 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 493 return; 494bad: 495 m_freem(m); 496} 497 498/* 499 * IP software interrupt routine - to go away sometime soon 500 */ 501static void 502ipintr(void) 503{ 504 int s; 505 struct mbuf *m; 506 507 while(1) { 508 s = splimp(); 509 IF_DEQUEUE(&ipintrq, m); 510 splx(s); 511 if (m == 0) 512 return; 513 ip_input(m); 514 } 515} 516 517NETISR_SET(NETISR_IP, ipintr); 518 519/* 520 * Take incoming datagram fragment and try to 521 * reassemble it into whole datagram. If a chain for 522 * reassembly of this datagram already exists, then it 523 * is given as fp; otherwise have to make a chain. 524 */ 525static struct ip * 526ip_reass(ip, fp) 527 register struct ipasfrag *ip; 528 register struct ipq *fp; 529{ 530 register struct mbuf *m = dtom(ip); 531 register struct ipasfrag *q; 532 struct mbuf *t; 533 int hlen = ip->ip_hl << 2; 534 int i, next; 535 536 /* 537 * Presence of header sizes in mbufs 538 * would confuse code below. 539 */ 540 m->m_data += hlen; 541 m->m_len -= hlen; 542 543 /* 544 * If first fragment to arrive, create a reassembly queue. 545 */ 546 if (fp == 0) { 547 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 548 goto dropfrag; 549 fp = mtod(t, struct ipq *); 550 insque(fp, &ipq); 551 fp->ipq_ttl = IPFRAGTTL; 552 fp->ipq_p = ip->ip_p; 553 fp->ipq_id = ip->ip_id; 554 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 555 fp->ipq_src = ((struct ip *)ip)->ip_src; 556 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 557#ifdef IPDIVERT 558 fp->ipq_divert = 0; 559#endif 560 q = (struct ipasfrag *)fp; 561 goto insert; 562 } 563 564 /* 565 * Find a segment which begins after this one does. 566 */ 567 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 568 if (q->ip_off > ip->ip_off) 569 break; 570 571 /* 572 * If there is a preceding segment, it may provide some of 573 * our data already. If so, drop the data from the incoming 574 * segment. If it provides all of our data, drop us. 575 */ 576 if (q->ipf_prev != (struct ipasfrag *)fp) { 577 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 578 if (i > 0) { 579 if (i >= ip->ip_len) 580 goto dropfrag; 581 m_adj(dtom(ip), i); 582 ip->ip_off += i; 583 ip->ip_len -= i; 584 } 585 } 586 587 /* 588 * While we overlap succeeding segments trim them or, 589 * if they are completely covered, dequeue them. 590 */ 591 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 592 struct mbuf *m0; 593 594 i = (ip->ip_off + ip->ip_len) - q->ip_off; 595 if (i < q->ip_len) { 596 q->ip_len -= i; 597 q->ip_off += i; 598 m_adj(dtom(q), i); 599 break; 600 } 601 m0 = dtom(q); 602 q = q->ipf_next; 603 ip_deq(q->ipf_prev); 604 m_freem(m0); 605 } 606 607insert: 608 609#ifdef IPDIVERT 610 /* 611 * Any fragment diverting causes the whole packet to divert 612 */ 613 if (frag_divert_port != 0) 614 fp->ipq_divert = frag_divert_port; 615 frag_divert_port = 0; 616#endif 617 618 /* 619 * Stick new segment in its place; 620 * check for complete reassembly. 621 */ 622 ip_enq(ip, q->ipf_prev); 623 next = 0; 624 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 625 if (q->ip_off != next) 626 return (0); 627 next += q->ip_len; 628 } 629 if (q->ipf_prev->ipf_mff & 1) 630 return (0); 631 632 /* 633 * Reassembly is complete; concatenate fragments. 634 */ 635 q = fp->ipq_next; 636 m = dtom(q); 637 t = m->m_next; 638 m->m_next = 0; 639 m_cat(m, t); 640 q = q->ipf_next; 641 while (q != (struct ipasfrag *)fp) { 642 t = dtom(q); 643 q = q->ipf_next; 644 m_cat(m, t); 645 } 646 647#ifdef IPDIVERT 648 /* 649 * Record divert port for packet, if any 650 */ 651 frag_divert_port = fp->ipq_divert; 652#endif 653 654 /* 655 * Create header for new ip packet by 656 * modifying header of first packet; 657 * dequeue and discard fragment reassembly header. 658 * Make header visible. 659 */ 660 ip = fp->ipq_next; 661 ip->ip_len = next; 662 ip->ipf_mff &= ~1; 663 ((struct ip *)ip)->ip_src = fp->ipq_src; 664 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 665 remque(fp); 666 (void) m_free(dtom(fp)); 667 m = dtom(ip); 668 m->m_len += (ip->ip_hl << 2); 669 m->m_data -= (ip->ip_hl << 2); 670 /* some debugging cruft by sklower, below, will go away soon */ 671 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 672 register int plen = 0; 673 for (t = m; m; m = m->m_next) 674 plen += m->m_len; 675 t->m_pkthdr.len = plen; 676 } 677 return ((struct ip *)ip); 678 679dropfrag: 680 ipstat.ips_fragdropped++; 681 m_freem(m); 682 return (0); 683} 684 685/* 686 * Free a fragment reassembly header and all 687 * associated datagrams. 688 */ 689static void 690ip_freef(fp) 691 struct ipq *fp; 692{ 693 register struct ipasfrag *q, *p; 694 695 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 696 p = q->ipf_next; 697 ip_deq(q); 698 m_freem(dtom(q)); 699 } 700 remque(fp); 701 (void) m_free(dtom(fp)); 702} 703 704/* 705 * Put an ip fragment on a reassembly chain. 706 * Like insque, but pointers in middle of structure. 707 */ 708static void 709ip_enq(p, prev) 710 register struct ipasfrag *p, *prev; 711{ 712 713 p->ipf_prev = prev; 714 p->ipf_next = prev->ipf_next; 715 prev->ipf_next->ipf_prev = p; 716 prev->ipf_next = p; 717} 718 719/* 720 * To ip_enq as remque is to insque. 721 */ 722static void 723ip_deq(p) 724 register struct ipasfrag *p; 725{ 726 727 p->ipf_prev->ipf_next = p->ipf_next; 728 p->ipf_next->ipf_prev = p->ipf_prev; 729} 730 731/* 732 * IP timer processing; 733 * if a timer expires on a reassembly 734 * queue, discard it. 735 */ 736void 737ip_slowtimo() 738{ 739 register struct ipq *fp; 740 int s = splnet(); 741 742 fp = ipq.next; 743 if (fp == 0) { 744 splx(s); 745 return; 746 } 747 while (fp != &ipq) { 748 --fp->ipq_ttl; 749 fp = fp->next; 750 if (fp->prev->ipq_ttl == 0) { 751 ipstat.ips_fragtimeout++; 752 ip_freef(fp->prev); 753 } 754 } 755 splx(s); 756} 757 758/* 759 * Drain off all datagram fragments. 760 */ 761void 762ip_drain() 763{ 764 while (ipq.next != &ipq) { 765 ipstat.ips_fragdropped++; 766 ip_freef(ipq.next); 767 } 768 769 in_rtqdrain(); 770} 771 772/* 773 * Do option processing on a datagram, 774 * possibly discarding it if bad options are encountered, 775 * or forwarding it if source-routed. 776 * Returns 1 if packet has been forwarded/freed, 777 * 0 if the packet should be processed further. 778 */ 779static int 780ip_dooptions(m) 781 struct mbuf *m; 782{ 783 register struct ip *ip = mtod(m, struct ip *); 784 register u_char *cp; 785 register struct ip_timestamp *ipt; 786 register struct in_ifaddr *ia; 787 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 788 struct in_addr *sin, dst; 789 n_time ntime; 790 791 dst = ip->ip_dst; 792 cp = (u_char *)(ip + 1); 793 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 794 for (; cnt > 0; cnt -= optlen, cp += optlen) { 795 opt = cp[IPOPT_OPTVAL]; 796 if (opt == IPOPT_EOL) 797 break; 798 if (opt == IPOPT_NOP) 799 optlen = 1; 800 else { 801 optlen = cp[IPOPT_OLEN]; 802 if (optlen <= 0 || optlen > cnt) { 803 code = &cp[IPOPT_OLEN] - (u_char *)ip; 804 goto bad; 805 } 806 } 807 switch (opt) { 808 809 default: 810 break; 811 812 /* 813 * Source routing with record. 814 * Find interface with current destination address. 815 * If none on this machine then drop if strictly routed, 816 * or do nothing if loosely routed. 817 * Record interface address and bring up next address 818 * component. If strictly routed make sure next 819 * address is on directly accessible net. 820 */ 821 case IPOPT_LSRR: 822 case IPOPT_SSRR: 823 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 824 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 825 goto bad; 826 } 827 ipaddr.sin_addr = ip->ip_dst; 828 ia = (struct in_ifaddr *) 829 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 830 if (ia == 0) { 831 if (opt == IPOPT_SSRR) { 832 type = ICMP_UNREACH; 833 code = ICMP_UNREACH_SRCFAIL; 834 goto bad; 835 } 836 /* 837 * Loose routing, and not at next destination 838 * yet; nothing to do except forward. 839 */ 840 break; 841 } 842 off--; /* 0 origin */ 843 if (off > optlen - sizeof(struct in_addr)) { 844 /* 845 * End of source route. Should be for us. 846 */ 847 save_rte(cp, ip->ip_src); 848 break; 849 } 850 851 if (!ip_dosourceroute) { 852 char buf[4*sizeof "123"]; 853 strcpy(buf, inet_ntoa(ip->ip_dst)); 854 855 log(LOG_WARNING, 856 "attempted source route from %s to %s\n", 857 inet_ntoa(ip->ip_src), buf); 858 type = ICMP_UNREACH; 859 code = ICMP_UNREACH_SRCFAIL; 860 goto bad; 861 } 862 863 /* 864 * locate outgoing interface 865 */ 866 (void)memcpy(&ipaddr.sin_addr, cp + off, 867 sizeof(ipaddr.sin_addr)); 868 869 if (opt == IPOPT_SSRR) { 870#define INA struct in_ifaddr * 871#define SA struct sockaddr * 872 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 873 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 874 } else 875 ia = ip_rtaddr(ipaddr.sin_addr); 876 if (ia == 0) { 877 type = ICMP_UNREACH; 878 code = ICMP_UNREACH_SRCFAIL; 879 goto bad; 880 } 881 ip->ip_dst = ipaddr.sin_addr; 882 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 883 sizeof(struct in_addr)); 884 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 885 /* 886 * Let ip_intr's mcast routing check handle mcast pkts 887 */ 888 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 889 break; 890 891 case IPOPT_RR: 892 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 893 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 894 goto bad; 895 } 896 /* 897 * If no space remains, ignore. 898 */ 899 off--; /* 0 origin */ 900 if (off > optlen - sizeof(struct in_addr)) 901 break; 902 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst, 903 sizeof(ipaddr.sin_addr)); 904 /* 905 * locate outgoing interface; if we're the destination, 906 * use the incoming interface (should be same). 907 */ 908 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 909 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 910 type = ICMP_UNREACH; 911 code = ICMP_UNREACH_HOST; 912 goto bad; 913 } 914 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 915 sizeof(struct in_addr)); 916 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 917 break; 918 919 case IPOPT_TS: 920 code = cp - (u_char *)ip; 921 ipt = (struct ip_timestamp *)cp; 922 if (ipt->ipt_len < 5) 923 goto bad; 924 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 925 if (++ipt->ipt_oflw == 0) 926 goto bad; 927 break; 928 } 929 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 930 switch (ipt->ipt_flg) { 931 932 case IPOPT_TS_TSONLY: 933 break; 934 935 case IPOPT_TS_TSANDADDR: 936 if (ipt->ipt_ptr + sizeof(n_time) + 937 sizeof(struct in_addr) > ipt->ipt_len) 938 goto bad; 939 ipaddr.sin_addr = dst; 940 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 941 m->m_pkthdr.rcvif); 942 if (ia == 0) 943 continue; 944 (void)memcpy(sin, &IA_SIN(ia)->sin_addr, 945 sizeof(struct in_addr)); 946 ipt->ipt_ptr += sizeof(struct in_addr); 947 break; 948 949 case IPOPT_TS_PRESPEC: 950 if (ipt->ipt_ptr + sizeof(n_time) + 951 sizeof(struct in_addr) > ipt->ipt_len) 952 goto bad; 953 (void)memcpy(&ipaddr.sin_addr, sin, 954 sizeof(struct in_addr)); 955 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 956 continue; 957 ipt->ipt_ptr += sizeof(struct in_addr); 958 break; 959 960 default: 961 goto bad; 962 } 963 ntime = iptime(); 964 (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime, 965 sizeof(n_time)); 966 ipt->ipt_ptr += sizeof(n_time); 967 } 968 } 969 if (forward) { 970 ip_forward(m, 1); 971 return (1); 972 } 973 return (0); 974bad: 975 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 976 icmp_error(m, type, code, 0, 0); 977 ipstat.ips_badoptions++; 978 return (1); 979} 980 981/* 982 * Given address of next destination (final or next hop), 983 * return internet address info of interface to be used to get there. 984 */ 985static struct in_ifaddr * 986ip_rtaddr(dst) 987 struct in_addr dst; 988{ 989 register struct sockaddr_in *sin; 990 991 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 992 993 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 994 if (ipforward_rt.ro_rt) { 995 RTFREE(ipforward_rt.ro_rt); 996 ipforward_rt.ro_rt = 0; 997 } 998 sin->sin_family = AF_INET; 999 sin->sin_len = sizeof(*sin); 1000 sin->sin_addr = dst; 1001 1002 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1003 } 1004 if (ipforward_rt.ro_rt == 0) 1005 return ((struct in_ifaddr *)0); 1006 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 1007} 1008 1009/* 1010 * Save incoming source route for use in replies, 1011 * to be picked up later by ip_srcroute if the receiver is interested. 1012 */ 1013void 1014save_rte(option, dst) 1015 u_char *option; 1016 struct in_addr dst; 1017{ 1018 unsigned olen; 1019 1020 olen = option[IPOPT_OLEN]; 1021#ifdef DIAGNOSTIC 1022 if (ipprintfs) 1023 printf("save_rte: olen %d\n", olen); 1024#endif 1025 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1026 return; 1027 bcopy(option, ip_srcrt.srcopt, olen); 1028 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1029 ip_srcrt.dst = dst; 1030} 1031 1032/* 1033 * Retrieve incoming source route for use in replies, 1034 * in the same form used by setsockopt. 1035 * The first hop is placed before the options, will be removed later. 1036 */ 1037struct mbuf * 1038ip_srcroute() 1039{ 1040 register struct in_addr *p, *q; 1041 register struct mbuf *m; 1042 1043 if (ip_nhops == 0) 1044 return ((struct mbuf *)0); 1045 m = m_get(M_DONTWAIT, MT_SOOPTS); 1046 if (m == 0) 1047 return ((struct mbuf *)0); 1048 1049#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1050 1051 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1052 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1053 OPTSIZ; 1054#ifdef DIAGNOSTIC 1055 if (ipprintfs) 1056 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1057#endif 1058 1059 /* 1060 * First save first hop for return route 1061 */ 1062 p = &ip_srcrt.route[ip_nhops - 1]; 1063 *(mtod(m, struct in_addr *)) = *p--; 1064#ifdef DIAGNOSTIC 1065 if (ipprintfs) 1066 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 1067#endif 1068 1069 /* 1070 * Copy option fields and padding (nop) to mbuf. 1071 */ 1072 ip_srcrt.nop = IPOPT_NOP; 1073 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1074 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), 1075 &ip_srcrt.nop, OPTSIZ); 1076 q = (struct in_addr *)(mtod(m, caddr_t) + 1077 sizeof(struct in_addr) + OPTSIZ); 1078#undef OPTSIZ 1079 /* 1080 * Record return path as an IP source route, 1081 * reversing the path (pointers are now aligned). 1082 */ 1083 while (p >= ip_srcrt.route) { 1084#ifdef DIAGNOSTIC 1085 if (ipprintfs) 1086 printf(" %lx", ntohl(q->s_addr)); 1087#endif 1088 *q++ = *p--; 1089 } 1090 /* 1091 * Last hop goes to final destination. 1092 */ 1093 *q = ip_srcrt.dst; 1094#ifdef DIAGNOSTIC 1095 if (ipprintfs) 1096 printf(" %lx\n", ntohl(q->s_addr)); 1097#endif 1098 return (m); 1099} 1100 1101/* 1102 * Strip out IP options, at higher 1103 * level protocol in the kernel. 1104 * Second argument is buffer to which options 1105 * will be moved, and return value is their length. 1106 * XXX should be deleted; last arg currently ignored. 1107 */ 1108void 1109ip_stripoptions(m, mopt) 1110 register struct mbuf *m; 1111 struct mbuf *mopt; 1112{ 1113 register int i; 1114 struct ip *ip = mtod(m, struct ip *); 1115 register caddr_t opts; 1116 int olen; 1117 1118 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1119 opts = (caddr_t)(ip + 1); 1120 i = m->m_len - (sizeof (struct ip) + olen); 1121 bcopy(opts + olen, opts, (unsigned)i); 1122 m->m_len -= olen; 1123 if (m->m_flags & M_PKTHDR) 1124 m->m_pkthdr.len -= olen; 1125 ip->ip_hl = sizeof(struct ip) >> 2; 1126} 1127 1128u_char inetctlerrmap[PRC_NCMDS] = { 1129 0, 0, 0, 0, 1130 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1131 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1132 EMSGSIZE, EHOSTUNREACH, 0, 0, 1133 0, 0, 0, 0, 1134 ENOPROTOOPT 1135}; 1136 1137/* 1138 * Forward a packet. If some error occurs return the sender 1139 * an icmp packet. Note we can't always generate a meaningful 1140 * icmp message because icmp doesn't have a large enough repertoire 1141 * of codes and types. 1142 * 1143 * If not forwarding, just drop the packet. This could be confusing 1144 * if ipforwarding was zero but some routing protocol was advancing 1145 * us as a gateway to somewhere. However, we must let the routing 1146 * protocol deal with that. 1147 * 1148 * The srcrt parameter indicates whether the packet is being forwarded 1149 * via a source route. 1150 */ 1151static void 1152ip_forward(m, srcrt) 1153 struct mbuf *m; 1154 int srcrt; 1155{ 1156 register struct ip *ip = mtod(m, struct ip *); 1157 register struct sockaddr_in *sin; 1158 register struct rtentry *rt; 1159 int error, type = 0, code = 0; 1160 struct mbuf *mcopy; 1161 n_long dest; 1162 struct ifnet *destifp; 1163 1164 dest = 0; 1165#ifdef DIAGNOSTIC 1166 if (ipprintfs) 1167 printf("forward: src %lx dst %lx ttl %x\n", 1168 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1169#endif 1170 1171 1172 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1173 ipstat.ips_cantforward++; 1174 m_freem(m); 1175 return; 1176 } 1177 HTONS(ip->ip_id); 1178 if (ip->ip_ttl <= IPTTLDEC) { 1179 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1180 return; 1181 } 1182 ip->ip_ttl -= IPTTLDEC; 1183 1184 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1185 if ((rt = ipforward_rt.ro_rt) == 0 || 1186 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1187 if (ipforward_rt.ro_rt) { 1188 RTFREE(ipforward_rt.ro_rt); 1189 ipforward_rt.ro_rt = 0; 1190 } 1191 sin->sin_family = AF_INET; 1192 sin->sin_len = sizeof(*sin); 1193 sin->sin_addr = ip->ip_dst; 1194 1195 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1196 if (ipforward_rt.ro_rt == 0) { 1197 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1198 return; 1199 } 1200 rt = ipforward_rt.ro_rt; 1201 } 1202 1203 /* 1204 * Save at most 64 bytes of the packet in case 1205 * we need to generate an ICMP message to the src. 1206 */ 1207 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1208 1209 /* 1210 * If forwarding packet using same interface that it came in on, 1211 * perhaps should send a redirect to sender to shortcut a hop. 1212 * Only send redirect if source is sending directly to us, 1213 * and if packet was not source routed (or has any options). 1214 * Also, don't send redirect if forwarding using a default route 1215 * or a route modified by a redirect. 1216 */ 1217#define satosin(sa) ((struct sockaddr_in *)(sa)) 1218 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1219 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1220 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1221 ipsendredirects && !srcrt) { 1222#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1223 u_long src = ntohl(ip->ip_src.s_addr); 1224 1225 if (RTA(rt) && 1226 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1227 if (rt->rt_flags & RTF_GATEWAY) 1228 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1229 else 1230 dest = ip->ip_dst.s_addr; 1231 /* Router requirements says to only send host redirects */ 1232 type = ICMP_REDIRECT; 1233 code = ICMP_REDIRECT_HOST; 1234#ifdef DIAGNOSTIC 1235 if (ipprintfs) 1236 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1237#endif 1238 } 1239 } 1240 1241 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1242 IP_FORWARDING, 0); 1243 if (error) 1244 ipstat.ips_cantforward++; 1245 else { 1246 ipstat.ips_forward++; 1247 if (type) 1248 ipstat.ips_redirectsent++; 1249 else { 1250 if (mcopy) 1251 m_freem(mcopy); 1252 return; 1253 } 1254 } 1255 if (mcopy == NULL) 1256 return; 1257 destifp = NULL; 1258 1259 switch (error) { 1260 1261 case 0: /* forwarded, but need redirect */ 1262 /* type, code set above */ 1263 break; 1264 1265 case ENETUNREACH: /* shouldn't happen, checked above */ 1266 case EHOSTUNREACH: 1267 case ENETDOWN: 1268 case EHOSTDOWN: 1269 default: 1270 type = ICMP_UNREACH; 1271 code = ICMP_UNREACH_HOST; 1272 break; 1273 1274 case EMSGSIZE: 1275 type = ICMP_UNREACH; 1276 code = ICMP_UNREACH_NEEDFRAG; 1277 if (ipforward_rt.ro_rt) 1278 destifp = ipforward_rt.ro_rt->rt_ifp; 1279 ipstat.ips_cantfrag++; 1280 break; 1281 1282 case ENOBUFS: 1283 type = ICMP_SOURCEQUENCH; 1284 code = 0; 1285 break; 1286 } 1287 icmp_error(mcopy, type, code, dest, destifp); 1288} 1289 1290int 1291ip_rsvp_init(struct socket *so) 1292{ 1293 if (so->so_type != SOCK_RAW || 1294 so->so_proto->pr_protocol != IPPROTO_RSVP) 1295 return EOPNOTSUPP; 1296 1297 if (ip_rsvpd != NULL) 1298 return EADDRINUSE; 1299 1300 ip_rsvpd = so; 1301 /* 1302 * This may seem silly, but we need to be sure we don't over-increment 1303 * the RSVP counter, in case something slips up. 1304 */ 1305 if (!ip_rsvp_on) { 1306 ip_rsvp_on = 1; 1307 rsvp_on++; 1308 } 1309 1310 return 0; 1311} 1312 1313int 1314ip_rsvp_done(void) 1315{ 1316 ip_rsvpd = NULL; 1317 /* 1318 * This may seem silly, but we need to be sure we don't over-decrement 1319 * the RSVP counter, in case something slips up. 1320 */ 1321 if (ip_rsvp_on) { 1322 ip_rsvp_on = 0; 1323 rsvp_on--; 1324 } 1325 return 0; 1326} 1327