ip_input.c revision 13266
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 1.33 1995/12/21 21:12:22 wollman Exp $ 35 */ 36 37#include <sys/param.h> 38#include <sys/systm.h> 39#include <sys/malloc.h> 40#include <sys/mbuf.h> 41#include <sys/domain.h> 42#include <sys/protosw.h> 43#include <sys/socket.h> 44#include <sys/errno.h> 45#include <sys/time.h> 46#include <sys/kernel.h> 47#include <sys/syslog.h> 48#include <sys/sysctl.h> 49 50#include <net/if.h> 51#include <net/route.h> 52#include <net/netisr.h> 53 54#include <netinet/in.h> 55#include <netinet/in_systm.h> 56#include <netinet/in_var.h> 57#include <netinet/ip.h> 58#include <netinet/in_pcb.h> 59#include <netinet/in_var.h> 60#include <netinet/ip_var.h> 61#include <netinet/ip_icmp.h> 62 63#include <netinet/ip_fw.h> 64 65#include <sys/socketvar.h> 66int rsvp_on = 0; 67static int ip_rsvp_on; 68struct socket *ip_rsvpd; 69 70static int ipforwarding = 0; 71SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW, 72 &ipforwarding, 0, ""); 73 74static int ipsendredirects = 1; /* XXX */ 75SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW, 76 &ipsendredirects, 0, ""); 77 78int ip_defttl = IPDEFTTL; 79SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW, 80 &ip_defttl, 0, ""); 81 82static int ip_dosourceroute = 0; 83SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW, 84 &ip_dosourceroute, 0, ""); 85#ifdef DIAGNOSTIC 86static int ipprintfs = 0; 87#endif 88 89extern struct domain inetdomain; 90extern struct protosw inetsw[]; 91u_char ip_protox[IPPROTO_MAX]; 92static int ipqmaxlen = IFQ_MAXLEN; 93struct in_ifaddr *in_ifaddr; /* first inet address */ 94struct ifqueue ipintrq; 95SYSCTL_INT(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLFLAG_RD, 96 &ipintrq.ifq_maxlen, 0, ""); 97SYSCTL_INT(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLFLAG_RD, 98 &ipintrq.ifq_drops, 0, ""); 99 100struct ipstat ipstat; 101static struct ipq ipq; 102 103#ifdef IPCTL_DEFMTU 104SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW, 105 &ip_mtu, 0, ""); 106#endif 107 108/* 109 * We need to save the IP options in case a protocol wants to respond 110 * to an incoming packet over the same route if the packet got here 111 * using IP source routing. This allows connection establishment and 112 * maintenance when the remote end is on a network that is not known 113 * to us. 114 */ 115static int ip_nhops = 0; 116static struct ip_srcrt { 117 struct in_addr dst; /* final destination */ 118 char nop; /* one NOP to align */ 119 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 120 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 121} ip_srcrt; 122 123static void save_rte __P((u_char *, struct in_addr)); 124static void ip_deq __P((struct ipasfrag *)); 125static int ip_dooptions __P((struct mbuf *)); 126static void ip_enq __P((struct ipasfrag *, struct ipasfrag *)); 127static void ip_forward __P((struct mbuf *, int)); 128static void ip_freef __P((struct ipq *)); 129static struct ip * 130 ip_reass __P((struct ipasfrag *, struct ipq *)); 131static struct in_ifaddr * 132 ip_rtaddr __P((struct in_addr)); 133static void ipintr __P((void)); 134/* 135 * IP initialization: fill in IP protocol switch table. 136 * All protocols not implemented in kernel go to raw IP protocol handler. 137 */ 138void 139ip_init() 140{ 141 register struct protosw *pr; 142 register int i; 143 144 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 145 if (pr == 0) 146 panic("ip_init"); 147 for (i = 0; i < IPPROTO_MAX; i++) 148 ip_protox[i] = pr - inetsw; 149 for (pr = inetdomain.dom_protosw; 150 pr < inetdomain.dom_protoswNPROTOSW; pr++) 151 if (pr->pr_domain->dom_family == PF_INET && 152 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 153 ip_protox[pr->pr_protocol] = pr - inetsw; 154 ipq.next = ipq.prev = &ipq; 155 ip_id = time.tv_sec & 0xffff; 156 ipintrq.ifq_maxlen = ipqmaxlen; 157} 158 159static struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 160static struct route ipforward_rt; 161 162/* 163 * Ip input routine. Checksum and byte swap header. If fragmented 164 * try to reassemble. Process options. Pass to next level. 165 */ 166static void 167ipintr(void) 168{ 169 register struct ip *ip; 170 register struct mbuf *m; 171 register struct ipq *fp; 172 register struct in_ifaddr *ia; 173 int hlen, s; 174 175next: 176 /* 177 * Get next datagram off input queue and get IP header 178 * in first mbuf. 179 */ 180 s = splimp(); 181 IF_DEQUEUE(&ipintrq, m); 182 splx(s); 183 if (m == 0) 184 return; 185#ifdef DIAGNOSTIC 186 if ((m->m_flags & M_PKTHDR) == 0) 187 panic("ipintr no HDR"); 188#endif 189 /* 190 * If no IP addresses have been set yet but the interfaces 191 * are receiving, can't do anything with incoming packets yet. 192 */ 193 if (in_ifaddr == NULL) 194 goto bad; 195 ipstat.ips_total++; 196 if (m->m_len < sizeof (struct ip) && 197 (m = m_pullup(m, sizeof (struct ip))) == 0) { 198 ipstat.ips_toosmall++; 199 goto next; 200 } 201 ip = mtod(m, struct ip *); 202 if (ip->ip_v != IPVERSION) { 203 ipstat.ips_badvers++; 204 goto bad; 205 } 206 hlen = ip->ip_hl << 2; 207 if (hlen < sizeof(struct ip)) { /* minimum header length */ 208 ipstat.ips_badhlen++; 209 goto bad; 210 } 211 if (hlen > m->m_len) { 212 if ((m = m_pullup(m, hlen)) == 0) { 213 ipstat.ips_badhlen++; 214 goto next; 215 } 216 ip = mtod(m, struct ip *); 217 } 218 ip->ip_sum = in_cksum(m, hlen); 219 if (ip->ip_sum) { 220 ipstat.ips_badsum++; 221 goto bad; 222 } 223 224 /* 225 * Convert fields to host representation. 226 */ 227 NTOHS(ip->ip_len); 228 if (ip->ip_len < hlen) { 229 ipstat.ips_badlen++; 230 goto bad; 231 } 232 NTOHS(ip->ip_id); 233 NTOHS(ip->ip_off); 234 235 /* 236 * Check that the amount of data in the buffers 237 * is as at least much as the IP header would have us expect. 238 * Trim mbufs if longer than we expect. 239 * Drop packet if shorter than we expect. 240 */ 241 if (m->m_pkthdr.len < ip->ip_len) { 242 ipstat.ips_tooshort++; 243 goto bad; 244 } 245 if (m->m_pkthdr.len > ip->ip_len) { 246 if (m->m_len == m->m_pkthdr.len) { 247 m->m_len = ip->ip_len; 248 m->m_pkthdr.len = ip->ip_len; 249 } else 250 m_adj(m, ip->ip_len - m->m_pkthdr.len); 251 } 252 /* 253 * IpHack's section. 254 * Right now when no processing on packet has done 255 * and it is still fresh out of network we do our black 256 * deals with it. 257 * - Firewall: deny/allow 258 * - Wrap: fake packet's addr/port <unimpl.> 259 * - Encapsulate: put it in another IP and send out. <unimp.> 260 */ 261 262 if (ip_fw_chk_ptr!=NULL) 263 if (!(*ip_fw_chk_ptr)(m,ip,m->m_pkthdr.rcvif,ip_fw_chain) ) { 264 goto next; 265 } 266 267 /* 268 * Process options and, if not destined for us, 269 * ship it on. ip_dooptions returns 1 when an 270 * error was detected (causing an icmp message 271 * to be sent and the original packet to be freed). 272 */ 273 ip_nhops = 0; /* for source routed packets */ 274 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 275 goto next; 276 277 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no 278 * matter if it is destined to another node, or whether it is 279 * a multicast one, RSVP wants it! and prevents it from being forwarded 280 * anywhere else. Also checks if the rsvp daemon is running before 281 * grabbing the packet. 282 */ 283 if (rsvp_on && ip->ip_p==IPPROTO_RSVP) 284 goto ours; 285 286 /* 287 * Check our list of addresses, to see if the packet is for us. 288 */ 289 for (ia = in_ifaddr; ia; ia = ia->ia_next) { 290#define satosin(sa) ((struct sockaddr_in *)(sa)) 291 292 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 293 goto ours; 294 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 295#if 0 296 u_long t; 297#endif 298 299 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 300 ip->ip_dst.s_addr) 301 goto ours; 302 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 303 goto ours; 304#if 0 /* XXX - this should go away */ 305 /* 306 * Look for all-0's host part (old broadcast addr), 307 * either for subnet or net. 308 */ 309 t = ntohl(ip->ip_dst.s_addr); 310 if (t == ia->ia_subnet) 311 goto ours; 312 if (t == ia->ia_net) 313 goto ours; 314#endif /* compatibility cruft */ 315 } 316 } 317 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 318 struct in_multi *inm; 319 if (ip_mrouter) { 320 /* 321 * If we are acting as a multicast router, all 322 * incoming multicast packets are passed to the 323 * kernel-level multicast forwarding function. 324 * The packet is returned (relatively) intact; if 325 * ip_mforward() returns a non-zero value, the packet 326 * must be discarded, else it may be accepted below. 327 * 328 * (The IP ident field is put in the same byte order 329 * as expected when ip_mforward() is called from 330 * ip_output().) 331 */ 332 ip->ip_id = htons(ip->ip_id); 333 if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) { 334 ipstat.ips_cantforward++; 335 m_freem(m); 336 goto next; 337 } 338 ip->ip_id = ntohs(ip->ip_id); 339 340 /* 341 * The process-level routing demon needs to receive 342 * all multicast IGMP packets, whether or not this 343 * host belongs to their destination groups. 344 */ 345 if (ip->ip_p == IPPROTO_IGMP) 346 goto ours; 347 ipstat.ips_forward++; 348 } 349 /* 350 * See if we belong to the destination multicast group on the 351 * arrival interface. 352 */ 353 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 354 if (inm == NULL) { 355 ipstat.ips_cantforward++; 356 m_freem(m); 357 goto next; 358 } 359 goto ours; 360 } 361 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 362 goto ours; 363 if (ip->ip_dst.s_addr == INADDR_ANY) 364 goto ours; 365 366 /* 367 * Not for us; forward if possible and desirable. 368 */ 369 if (ipforwarding == 0) { 370 ipstat.ips_cantforward++; 371 m_freem(m); 372 } else 373 ip_forward(m, 0); 374 goto next; 375 376ours: 377 378 /* 379 * If packet came to us we count it... 380 * This way we count all incoming packets which has 381 * not been forwarded... 382 * Do not convert ip_len to host byte order when 383 * counting,ppl already made it for us before.. 384 */ 385 if (ip_acct_cnt_ptr!=NULL) 386 (*ip_acct_cnt_ptr)(ip,m->m_pkthdr.rcvif,ip_acct_chain,0); 387 388 /* 389 * If offset or IP_MF are set, must reassemble. 390 * Otherwise, nothing need be done. 391 * (We could look in the reassembly queue to see 392 * if the packet was previously fragmented, 393 * but it's not worth the time; just let them time out.) 394 */ 395 if (ip->ip_off &~ IP_DF) { 396 if (m->m_flags & M_EXT) { /* XXX */ 397 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 398 ipstat.ips_toosmall++; 399 goto next; 400 } 401 ip = mtod(m, struct ip *); 402 } 403 /* 404 * Look for queue of fragments 405 * of this datagram. 406 */ 407 for (fp = ipq.next; fp != &ipq; fp = fp->next) 408 if (ip->ip_id == fp->ipq_id && 409 ip->ip_src.s_addr == fp->ipq_src.s_addr && 410 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 411 ip->ip_p == fp->ipq_p) 412 goto found; 413 fp = 0; 414found: 415 416 /* 417 * Adjust ip_len to not reflect header, 418 * set ip_mff if more fragments are expected, 419 * convert offset of this to bytes. 420 */ 421 ip->ip_len -= hlen; 422 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 423 if (ip->ip_off & IP_MF) 424 ((struct ipasfrag *)ip)->ipf_mff |= 1; 425 ip->ip_off <<= 3; 426 427 /* 428 * If datagram marked as having more fragments 429 * or if this is not the first fragment, 430 * attempt reassembly; if it succeeds, proceed. 431 */ 432 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 433 ipstat.ips_fragments++; 434 ip = ip_reass((struct ipasfrag *)ip, fp); 435 if (ip == 0) 436 goto next; 437 ipstat.ips_reassembled++; 438 m = dtom(ip); 439 } else 440 if (fp) 441 ip_freef(fp); 442 } else 443 ip->ip_len -= hlen; 444 445 /* 446 * Switch out to protocol's input routine. 447 */ 448 ipstat.ips_delivered++; 449 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 450 goto next; 451bad: 452 m_freem(m); 453 goto next; 454} 455 456NETISR_SET(NETISR_IP, ipintr); 457 458/* 459 * Take incoming datagram fragment and try to 460 * reassemble it into whole datagram. If a chain for 461 * reassembly of this datagram already exists, then it 462 * is given as fp; otherwise have to make a chain. 463 */ 464static struct ip * 465ip_reass(ip, fp) 466 register struct ipasfrag *ip; 467 register struct ipq *fp; 468{ 469 register struct mbuf *m = dtom(ip); 470 register struct ipasfrag *q; 471 struct mbuf *t; 472 int hlen = ip->ip_hl << 2; 473 int i, next; 474 475 /* 476 * Presence of header sizes in mbufs 477 * would confuse code below. 478 */ 479 m->m_data += hlen; 480 m->m_len -= hlen; 481 482 /* 483 * If first fragment to arrive, create a reassembly queue. 484 */ 485 if (fp == 0) { 486 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 487 goto dropfrag; 488 fp = mtod(t, struct ipq *); 489 insque(fp, &ipq); 490 fp->ipq_ttl = IPFRAGTTL; 491 fp->ipq_p = ip->ip_p; 492 fp->ipq_id = ip->ip_id; 493 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 494 fp->ipq_src = ((struct ip *)ip)->ip_src; 495 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 496 q = (struct ipasfrag *)fp; 497 goto insert; 498 } 499 500 /* 501 * Find a segment which begins after this one does. 502 */ 503 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 504 if (q->ip_off > ip->ip_off) 505 break; 506 507 /* 508 * If there is a preceding segment, it may provide some of 509 * our data already. If so, drop the data from the incoming 510 * segment. If it provides all of our data, drop us. 511 */ 512 if (q->ipf_prev != (struct ipasfrag *)fp) { 513 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 514 if (i > 0) { 515 if (i >= ip->ip_len) 516 goto dropfrag; 517 m_adj(dtom(ip), i); 518 ip->ip_off += i; 519 ip->ip_len -= i; 520 } 521 } 522 523 /* 524 * While we overlap succeeding segments trim them or, 525 * if they are completely covered, dequeue them. 526 */ 527 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 528 i = (ip->ip_off + ip->ip_len) - q->ip_off; 529 if (i < q->ip_len) { 530 q->ip_len -= i; 531 q->ip_off += i; 532 m_adj(dtom(q), i); 533 break; 534 } 535 q = q->ipf_next; 536 m_freem(dtom(q->ipf_prev)); 537 ip_deq(q->ipf_prev); 538 } 539 540insert: 541 /* 542 * Stick new segment in its place; 543 * check for complete reassembly. 544 */ 545 ip_enq(ip, q->ipf_prev); 546 next = 0; 547 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 548 if (q->ip_off != next) 549 return (0); 550 next += q->ip_len; 551 } 552 if (q->ipf_prev->ipf_mff & 1) 553 return (0); 554 555 /* 556 * Reassembly is complete; concatenate fragments. 557 */ 558 q = fp->ipq_next; 559 m = dtom(q); 560 t = m->m_next; 561 m->m_next = 0; 562 m_cat(m, t); 563 q = q->ipf_next; 564 while (q != (struct ipasfrag *)fp) { 565 t = dtom(q); 566 q = q->ipf_next; 567 m_cat(m, t); 568 } 569 570 /* 571 * Create header for new ip packet by 572 * modifying header of first packet; 573 * dequeue and discard fragment reassembly header. 574 * Make header visible. 575 */ 576 ip = fp->ipq_next; 577 ip->ip_len = next; 578 ip->ipf_mff &= ~1; 579 ((struct ip *)ip)->ip_src = fp->ipq_src; 580 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 581 remque(fp); 582 (void) m_free(dtom(fp)); 583 m = dtom(ip); 584 m->m_len += (ip->ip_hl << 2); 585 m->m_data -= (ip->ip_hl << 2); 586 /* some debugging cruft by sklower, below, will go away soon */ 587 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 588 register int plen = 0; 589 for (t = m; m; m = m->m_next) 590 plen += m->m_len; 591 t->m_pkthdr.len = plen; 592 } 593 return ((struct ip *)ip); 594 595dropfrag: 596 ipstat.ips_fragdropped++; 597 m_freem(m); 598 return (0); 599} 600 601/* 602 * Free a fragment reassembly header and all 603 * associated datagrams. 604 */ 605static void 606ip_freef(fp) 607 struct ipq *fp; 608{ 609 register struct ipasfrag *q, *p; 610 611 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 612 p = q->ipf_next; 613 ip_deq(q); 614 m_freem(dtom(q)); 615 } 616 remque(fp); 617 (void) m_free(dtom(fp)); 618} 619 620/* 621 * Put an ip fragment on a reassembly chain. 622 * Like insque, but pointers in middle of structure. 623 */ 624static void 625ip_enq(p, prev) 626 register struct ipasfrag *p, *prev; 627{ 628 629 p->ipf_prev = prev; 630 p->ipf_next = prev->ipf_next; 631 prev->ipf_next->ipf_prev = p; 632 prev->ipf_next = p; 633} 634 635/* 636 * To ip_enq as remque is to insque. 637 */ 638static void 639ip_deq(p) 640 register struct ipasfrag *p; 641{ 642 643 p->ipf_prev->ipf_next = p->ipf_next; 644 p->ipf_next->ipf_prev = p->ipf_prev; 645} 646 647/* 648 * IP timer processing; 649 * if a timer expires on a reassembly 650 * queue, discard it. 651 */ 652void 653ip_slowtimo() 654{ 655 register struct ipq *fp; 656 int s = splnet(); 657 658 fp = ipq.next; 659 if (fp == 0) { 660 splx(s); 661 return; 662 } 663 while (fp != &ipq) { 664 --fp->ipq_ttl; 665 fp = fp->next; 666 if (fp->prev->ipq_ttl == 0) { 667 ipstat.ips_fragtimeout++; 668 ip_freef(fp->prev); 669 } 670 } 671 splx(s); 672} 673 674/* 675 * Drain off all datagram fragments. 676 */ 677void 678ip_drain() 679{ 680 while (ipq.next != &ipq) { 681 ipstat.ips_fragdropped++; 682 ip_freef(ipq.next); 683 } 684 685 in_rtqdrain(); 686} 687 688/* 689 * Do option processing on a datagram, 690 * possibly discarding it if bad options are encountered, 691 * or forwarding it if source-routed. 692 * Returns 1 if packet has been forwarded/freed, 693 * 0 if the packet should be processed further. 694 */ 695static int 696ip_dooptions(m) 697 struct mbuf *m; 698{ 699 register struct ip *ip = mtod(m, struct ip *); 700 register u_char *cp; 701 register struct ip_timestamp *ipt; 702 register struct in_ifaddr *ia; 703 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 704 struct in_addr *sin, dst; 705 n_time ntime; 706 707 dst = ip->ip_dst; 708 cp = (u_char *)(ip + 1); 709 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 710 for (; cnt > 0; cnt -= optlen, cp += optlen) { 711 opt = cp[IPOPT_OPTVAL]; 712 if (opt == IPOPT_EOL) 713 break; 714 if (opt == IPOPT_NOP) 715 optlen = 1; 716 else { 717 optlen = cp[IPOPT_OLEN]; 718 if (optlen <= 0 || optlen > cnt) { 719 code = &cp[IPOPT_OLEN] - (u_char *)ip; 720 goto bad; 721 } 722 } 723 switch (opt) { 724 725 default: 726 break; 727 728 /* 729 * Source routing with record. 730 * Find interface with current destination address. 731 * If none on this machine then drop if strictly routed, 732 * or do nothing if loosely routed. 733 * Record interface address and bring up next address 734 * component. If strictly routed make sure next 735 * address is on directly accessible net. 736 */ 737 case IPOPT_LSRR: 738 case IPOPT_SSRR: 739 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 740 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 741 goto bad; 742 } 743 ipaddr.sin_addr = ip->ip_dst; 744 ia = (struct in_ifaddr *) 745 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 746 if (ia == 0) { 747 if (opt == IPOPT_SSRR) { 748 type = ICMP_UNREACH; 749 code = ICMP_UNREACH_SRCFAIL; 750 goto bad; 751 } 752 /* 753 * Loose routing, and not at next destination 754 * yet; nothing to do except forward. 755 */ 756 break; 757 } 758 off--; /* 0 origin */ 759 if (off > optlen - sizeof(struct in_addr)) { 760 /* 761 * End of source route. Should be for us. 762 */ 763 save_rte(cp, ip->ip_src); 764 break; 765 } 766 767 if (!ip_dosourceroute) { 768 char buf[4*sizeof "123"]; 769 strcpy(buf, inet_ntoa(ip->ip_dst)); 770 771 log(LOG_WARNING, 772 "attempted source route from %s to %s\n", 773 inet_ntoa(ip->ip_src), buf); 774 type = ICMP_UNREACH; 775 code = ICMP_UNREACH_SRCFAIL; 776 goto bad; 777 } 778 779 /* 780 * locate outgoing interface 781 */ 782 (void)memcpy(&ipaddr.sin_addr, cp + off, 783 sizeof(ipaddr.sin_addr)); 784 785 if (opt == IPOPT_SSRR) { 786#define INA struct in_ifaddr * 787#define SA struct sockaddr * 788 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 789 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 790 } else 791 ia = ip_rtaddr(ipaddr.sin_addr); 792 if (ia == 0) { 793 type = ICMP_UNREACH; 794 code = ICMP_UNREACH_SRCFAIL; 795 goto bad; 796 } 797 ip->ip_dst = ipaddr.sin_addr; 798 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 799 sizeof(struct in_addr)); 800 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 801 /* 802 * Let ip_intr's mcast routing check handle mcast pkts 803 */ 804 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 805 break; 806 807 case IPOPT_RR: 808 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 809 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 810 goto bad; 811 } 812 /* 813 * If no space remains, ignore. 814 */ 815 off--; /* 0 origin */ 816 if (off > optlen - sizeof(struct in_addr)) 817 break; 818 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst, 819 sizeof(ipaddr.sin_addr)); 820 /* 821 * locate outgoing interface; if we're the destination, 822 * use the incoming interface (should be same). 823 */ 824 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 825 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 826 type = ICMP_UNREACH; 827 code = ICMP_UNREACH_HOST; 828 goto bad; 829 } 830 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 831 sizeof(struct in_addr)); 832 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 833 break; 834 835 case IPOPT_TS: 836 code = cp - (u_char *)ip; 837 ipt = (struct ip_timestamp *)cp; 838 if (ipt->ipt_len < 5) 839 goto bad; 840 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 841 if (++ipt->ipt_oflw == 0) 842 goto bad; 843 break; 844 } 845 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 846 switch (ipt->ipt_flg) { 847 848 case IPOPT_TS_TSONLY: 849 break; 850 851 case IPOPT_TS_TSANDADDR: 852 if (ipt->ipt_ptr + sizeof(n_time) + 853 sizeof(struct in_addr) > ipt->ipt_len) 854 goto bad; 855 ipaddr.sin_addr = dst; 856 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 857 m->m_pkthdr.rcvif); 858 if (ia == 0) 859 continue; 860 (void)memcpy(sin, &IA_SIN(ia)->sin_addr, 861 sizeof(struct in_addr)); 862 ipt->ipt_ptr += sizeof(struct in_addr); 863 break; 864 865 case IPOPT_TS_PRESPEC: 866 if (ipt->ipt_ptr + sizeof(n_time) + 867 sizeof(struct in_addr) > ipt->ipt_len) 868 goto bad; 869 (void)memcpy(&ipaddr.sin_addr, sin, 870 sizeof(struct in_addr)); 871 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 872 continue; 873 ipt->ipt_ptr += sizeof(struct in_addr); 874 break; 875 876 default: 877 goto bad; 878 } 879 ntime = iptime(); 880 (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime, 881 sizeof(n_time)); 882 ipt->ipt_ptr += sizeof(n_time); 883 } 884 } 885 if (forward) { 886 ip_forward(m, 1); 887 return (1); 888 } 889 return (0); 890bad: 891 ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */ 892 icmp_error(m, type, code, 0, 0); 893 ipstat.ips_badoptions++; 894 return (1); 895} 896 897/* 898 * Given address of next destination (final or next hop), 899 * return internet address info of interface to be used to get there. 900 */ 901static struct in_ifaddr * 902ip_rtaddr(dst) 903 struct in_addr dst; 904{ 905 register struct sockaddr_in *sin; 906 907 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 908 909 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 910 if (ipforward_rt.ro_rt) { 911 RTFREE(ipforward_rt.ro_rt); 912 ipforward_rt.ro_rt = 0; 913 } 914 sin->sin_family = AF_INET; 915 sin->sin_len = sizeof(*sin); 916 sin->sin_addr = dst; 917 918 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 919 } 920 if (ipforward_rt.ro_rt == 0) 921 return ((struct in_ifaddr *)0); 922 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 923} 924 925/* 926 * Save incoming source route for use in replies, 927 * to be picked up later by ip_srcroute if the receiver is interested. 928 */ 929void 930save_rte(option, dst) 931 u_char *option; 932 struct in_addr dst; 933{ 934 unsigned olen; 935 936 olen = option[IPOPT_OLEN]; 937#ifdef DIAGNOSTIC 938 if (ipprintfs) 939 printf("save_rte: olen %d\n", olen); 940#endif 941 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 942 return; 943 (void)memcpy(ip_srcrt.srcopt, option, olen); 944 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 945 ip_srcrt.dst = dst; 946} 947 948/* 949 * Retrieve incoming source route for use in replies, 950 * in the same form used by setsockopt. 951 * The first hop is placed before the options, will be removed later. 952 */ 953struct mbuf * 954ip_srcroute() 955{ 956 register struct in_addr *p, *q; 957 register struct mbuf *m; 958 959 if (ip_nhops == 0) 960 return ((struct mbuf *)0); 961 m = m_get(M_DONTWAIT, MT_SOOPTS); 962 if (m == 0) 963 return ((struct mbuf *)0); 964 965#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 966 967 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 968 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 969 OPTSIZ; 970#ifdef DIAGNOSTIC 971 if (ipprintfs) 972 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 973#endif 974 975 /* 976 * First save first hop for return route 977 */ 978 p = &ip_srcrt.route[ip_nhops - 1]; 979 *(mtod(m, struct in_addr *)) = *p--; 980#ifdef DIAGNOSTIC 981 if (ipprintfs) 982 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 983#endif 984 985 /* 986 * Copy option fields and padding (nop) to mbuf. 987 */ 988 ip_srcrt.nop = IPOPT_NOP; 989 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 990 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), 991 &ip_srcrt.nop, OPTSIZ); 992 q = (struct in_addr *)(mtod(m, caddr_t) + 993 sizeof(struct in_addr) + OPTSIZ); 994#undef OPTSIZ 995 /* 996 * Record return path as an IP source route, 997 * reversing the path (pointers are now aligned). 998 */ 999 while (p >= ip_srcrt.route) { 1000#ifdef DIAGNOSTIC 1001 if (ipprintfs) 1002 printf(" %lx", ntohl(q->s_addr)); 1003#endif 1004 *q++ = *p--; 1005 } 1006 /* 1007 * Last hop goes to final destination. 1008 */ 1009 *q = ip_srcrt.dst; 1010#ifdef DIAGNOSTIC 1011 if (ipprintfs) 1012 printf(" %lx\n", ntohl(q->s_addr)); 1013#endif 1014 return (m); 1015} 1016 1017/* 1018 * Strip out IP options, at higher 1019 * level protocol in the kernel. 1020 * Second argument is buffer to which options 1021 * will be moved, and return value is their length. 1022 * XXX should be deleted; last arg currently ignored. 1023 */ 1024void 1025ip_stripoptions(m, mopt) 1026 register struct mbuf *m; 1027 struct mbuf *mopt; 1028{ 1029 register int i; 1030 struct ip *ip = mtod(m, struct ip *); 1031 register caddr_t opts; 1032 int olen; 1033 1034 olen = (ip->ip_hl<<2) - sizeof (struct ip); 1035 opts = (caddr_t)(ip + 1); 1036 i = m->m_len - (sizeof (struct ip) + olen); 1037 bcopy(opts + olen, opts, (unsigned)i); 1038 m->m_len -= olen; 1039 if (m->m_flags & M_PKTHDR) 1040 m->m_pkthdr.len -= olen; 1041 ip->ip_hl = sizeof(struct ip) >> 2; 1042} 1043 1044u_char inetctlerrmap[PRC_NCMDS] = { 1045 0, 0, 0, 0, 1046 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1047 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1048 EMSGSIZE, EHOSTUNREACH, 0, 0, 1049 0, 0, 0, 0, 1050 ENOPROTOOPT 1051}; 1052 1053/* 1054 * Forward a packet. If some error occurs return the sender 1055 * an icmp packet. Note we can't always generate a meaningful 1056 * icmp message because icmp doesn't have a large enough repertoire 1057 * of codes and types. 1058 * 1059 * If not forwarding, just drop the packet. This could be confusing 1060 * if ipforwarding was zero but some routing protocol was advancing 1061 * us as a gateway to somewhere. However, we must let the routing 1062 * protocol deal with that. 1063 * 1064 * The srcrt parameter indicates whether the packet is being forwarded 1065 * via a source route. 1066 */ 1067static void 1068ip_forward(m, srcrt) 1069 struct mbuf *m; 1070 int srcrt; 1071{ 1072 register struct ip *ip = mtod(m, struct ip *); 1073 register struct sockaddr_in *sin; 1074 register struct rtentry *rt; 1075 int error, type = 0, code = 0; 1076 struct mbuf *mcopy; 1077 n_long dest; 1078 struct ifnet *destifp; 1079 1080 dest = 0; 1081#ifdef DIAGNOSTIC 1082 if (ipprintfs) 1083 printf("forward: src %lx dst %lx ttl %x\n", 1084 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1085#endif 1086 1087 1088 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1089 ipstat.ips_cantforward++; 1090 m_freem(m); 1091 return; 1092 } 1093 HTONS(ip->ip_id); 1094 if (ip->ip_ttl <= IPTTLDEC) { 1095 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1096 return; 1097 } 1098 ip->ip_ttl -= IPTTLDEC; 1099 1100 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1101 if ((rt = ipforward_rt.ro_rt) == 0 || 1102 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1103 if (ipforward_rt.ro_rt) { 1104 RTFREE(ipforward_rt.ro_rt); 1105 ipforward_rt.ro_rt = 0; 1106 } 1107 sin->sin_family = AF_INET; 1108 sin->sin_len = sizeof(*sin); 1109 sin->sin_addr = ip->ip_dst; 1110 1111 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1112 if (ipforward_rt.ro_rt == 0) { 1113 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1114 return; 1115 } 1116 rt = ipforward_rt.ro_rt; 1117 } 1118 1119 /* 1120 * Save at most 64 bytes of the packet in case 1121 * we need to generate an ICMP message to the src. 1122 */ 1123 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1124 1125 /* 1126 * If forwarding packet using same interface that it came in on, 1127 * perhaps should send a redirect to sender to shortcut a hop. 1128 * Only send redirect if source is sending directly to us, 1129 * and if packet was not source routed (or has any options). 1130 * Also, don't send redirect if forwarding using a default route 1131 * or a route modified by a redirect. 1132 */ 1133#define satosin(sa) ((struct sockaddr_in *)(sa)) 1134 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1135 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1136 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1137 ipsendredirects && !srcrt) { 1138#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1139 u_long src = ntohl(ip->ip_src.s_addr); 1140 1141 if (RTA(rt) && 1142 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1143 if (rt->rt_flags & RTF_GATEWAY) 1144 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1145 else 1146 dest = ip->ip_dst.s_addr; 1147 /* Router requirements says to only send host redirects */ 1148 type = ICMP_REDIRECT; 1149 code = ICMP_REDIRECT_HOST; 1150#ifdef DIAGNOSTIC 1151 if (ipprintfs) 1152 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1153#endif 1154 } 1155 } 1156 1157 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1158 IP_FORWARDING, 0); 1159 if (error) 1160 ipstat.ips_cantforward++; 1161 else { 1162 ipstat.ips_forward++; 1163 if (type) 1164 ipstat.ips_redirectsent++; 1165 else { 1166 if (mcopy) 1167 m_freem(mcopy); 1168 return; 1169 } 1170 } 1171 if (mcopy == NULL) 1172 return; 1173 destifp = NULL; 1174 1175 switch (error) { 1176 1177 case 0: /* forwarded, but need redirect */ 1178 /* type, code set above */ 1179 break; 1180 1181 case ENETUNREACH: /* shouldn't happen, checked above */ 1182 case EHOSTUNREACH: 1183 case ENETDOWN: 1184 case EHOSTDOWN: 1185 default: 1186 type = ICMP_UNREACH; 1187 code = ICMP_UNREACH_HOST; 1188 break; 1189 1190 case EMSGSIZE: 1191 type = ICMP_UNREACH; 1192 code = ICMP_UNREACH_NEEDFRAG; 1193 if (ipforward_rt.ro_rt) 1194 destifp = ipforward_rt.ro_rt->rt_ifp; 1195 ipstat.ips_cantfrag++; 1196 break; 1197 1198 case ENOBUFS: 1199 type = ICMP_SOURCEQUENCH; 1200 code = 0; 1201 break; 1202 } 1203 icmp_error(mcopy, type, code, dest, destifp); 1204} 1205 1206int 1207ip_rsvp_init(struct socket *so) 1208{ 1209 if (so->so_type != SOCK_RAW || 1210 so->so_proto->pr_protocol != IPPROTO_RSVP) 1211 return EOPNOTSUPP; 1212 1213 if (ip_rsvpd != NULL) 1214 return EADDRINUSE; 1215 1216 ip_rsvpd = so; 1217 /* 1218 * This may seem silly, but we need to be sure we don't over-increment 1219 * the RSVP counter, in case something slips up. 1220 */ 1221 if (!ip_rsvp_on) { 1222 ip_rsvp_on = 1; 1223 rsvp_on++; 1224 } 1225 1226 return 0; 1227} 1228 1229int 1230ip_rsvp_done(void) 1231{ 1232 ip_rsvpd = NULL; 1233 /* 1234 * This may seem silly, but we need to be sure we don't over-decrement 1235 * the RSVP counter, in case something slips up. 1236 */ 1237 if (ip_rsvp_on) { 1238 ip_rsvp_on = 0; 1239 rsvp_on--; 1240 } 1241 return 0; 1242} 1243