ip_reass.c revision 34746
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 1.79 1998/02/26 08:31:53 dima 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_bootp.h" 41#include "opt_ipfw.h" 42#include "opt_ipdivert.h" 43#include "opt_ipfilter.h" 44 45#include <stddef.h> 46 47#include <sys/param.h> 48#include <sys/systm.h> 49#include <sys/mbuf.h> 50#include <sys/domain.h> 51#include <sys/protosw.h> 52#include <sys/socket.h> 53#include <sys/time.h> 54#include <sys/kernel.h> 55#include <sys/syslog.h> 56#include <sys/sysctl.h> 57 58#include <net/if.h> 59#include <net/if_var.h> 60#include <net/if_dl.h> 61#include <net/route.h> 62#include <net/netisr.h> 63 64#include <netinet/in.h> 65#include <netinet/in_systm.h> 66#include <netinet/in_var.h> 67#include <netinet/ip.h> 68#include <netinet/in_pcb.h> 69#include <netinet/ip_var.h> 70#include <netinet/ip_icmp.h> 71#include <machine/in_cksum.h> 72 73#include <sys/socketvar.h> 74 75#ifdef IPFIREWALL 76#include <netinet/ip_fw.h> 77#endif 78 79int rsvp_on = 0; 80static int ip_rsvp_on; 81struct socket *ip_rsvpd; 82 83static int ipforwarding = 0; 84SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW, 85 &ipforwarding, 0, ""); 86 87static int ipsendredirects = 1; /* XXX */ 88SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW, 89 &ipsendredirects, 0, ""); 90 91int ip_defttl = IPDEFTTL; 92SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW, 93 &ip_defttl, 0, ""); 94 95static int ip_dosourceroute = 0; 96SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW, 97 &ip_dosourceroute, 0, ""); 98 99static int ip_acceptsourceroute = 0; 100SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute, 101 CTLFLAG_RW, &ip_acceptsourceroute, 0, ""); 102#ifdef DIAGNOSTIC 103static int ipprintfs = 0; 104#endif 105 106extern struct domain inetdomain; 107extern struct protosw inetsw[]; 108u_char ip_protox[IPPROTO_MAX]; 109static int ipqmaxlen = IFQ_MAXLEN; 110struct in_ifaddrhead in_ifaddrhead; /* first inet address */ 111struct ifqueue ipintrq; 112SYSCTL_INT(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLFLAG_RD, 113 &ipintrq.ifq_maxlen, 0, ""); 114SYSCTL_INT(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLFLAG_RD, 115 &ipintrq.ifq_drops, 0, ""); 116 117struct ipstat ipstat; 118SYSCTL_STRUCT(_net_inet_ip, IPCTL_STATS, stats, CTLFLAG_RD, 119 &ipstat, ipstat, ""); 120 121/* Packet reassembly stuff */ 122#define IPREASS_NHASH_LOG2 6 123#define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2) 124#define IPREASS_HMASK (IPREASS_NHASH - 1) 125#define IPREASS_HASH(x,y) \ 126 ((((x) & 0xF | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK) 127 128static struct ipq ipq[IPREASS_NHASH]; 129static int nipq = 0; /* total # of reass queues */ 130static int maxnipq; 131 132#ifdef IPCTL_DEFMTU 133SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW, 134 &ip_mtu, 0, ""); 135#endif 136 137#if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1 138#undef COMPAT_IPFW 139#define COMPAT_IPFW 1 140#else 141#undef COMPAT_IPFW 142#endif 143 144#ifdef COMPAT_IPFW 145/* Firewall hooks */ 146ip_fw_chk_t *ip_fw_chk_ptr; 147ip_fw_ctl_t *ip_fw_ctl_ptr; 148 149/* IP Network Address Translation (NAT) hooks */ 150ip_nat_t *ip_nat_ptr; 151ip_nat_ctl_t *ip_nat_ctl_ptr; 152#endif 153 154#if defined(IPFILTER_LKM) || defined(IPFILTER) 155int iplattach __P((void)); 156int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **)) = NULL; 157#endif 158 159 160/* 161 * We need to save the IP options in case a protocol wants to respond 162 * to an incoming packet over the same route if the packet got here 163 * using IP source routing. This allows connection establishment and 164 * maintenance when the remote end is on a network that is not known 165 * to us. 166 */ 167static int ip_nhops = 0; 168static struct ip_srcrt { 169 struct in_addr dst; /* final destination */ 170 char nop; /* one NOP to align */ 171 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 172 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 173} ip_srcrt; 174 175#ifdef IPDIVERT 176/* 177 * Shared variable between ip_input() and ip_reass() to communicate 178 * about which packets, once assembled from fragments, get diverted, 179 * and to which port. 180 */ 181static u_short frag_divert_port; 182#endif 183 184static void save_rte __P((u_char *, struct in_addr)); 185static void ip_deq __P((struct ipasfrag *)); 186static int ip_dooptions __P((struct mbuf *)); 187static void ip_enq __P((struct ipasfrag *, struct ipasfrag *)); 188static void ip_forward __P((struct mbuf *, int)); 189static void ip_freef __P((struct ipq *)); 190static struct ip * 191 ip_reass __P((struct ipasfrag *, struct ipq *, struct ipq *)); 192static struct in_ifaddr * 193 ip_rtaddr __P((struct in_addr)); 194static void ipintr __P((void)); 195/* 196 * IP initialization: fill in IP protocol switch table. 197 * All protocols not implemented in kernel go to raw IP protocol handler. 198 */ 199void 200ip_init() 201{ 202 register struct protosw *pr; 203 register int i; 204 205 TAILQ_INIT(&in_ifaddrhead); 206 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 207 if (pr == 0) 208 panic("ip_init"); 209 for (i = 0; i < IPPROTO_MAX; i++) 210 ip_protox[i] = pr - inetsw; 211 for (pr = inetdomain.dom_protosw; 212 pr < inetdomain.dom_protoswNPROTOSW; pr++) 213 if (pr->pr_domain->dom_family == PF_INET && 214 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 215 ip_protox[pr->pr_protocol] = pr - inetsw; 216 217 for (i = 0; i < IPREASS_NHASH; i++) 218 ipq[i].next = ipq[i].prev = &ipq[i]; 219 220 maxnipq = nmbclusters/4; 221 222 ip_id = time.tv_sec & 0xffff; 223 ipintrq.ifq_maxlen = ipqmaxlen; 224#ifdef IPFIREWALL 225 ip_fw_init(); 226#endif 227#ifdef IPNAT 228 ip_nat_init(); 229#endif 230#ifdef IPFILTER 231 iplattach(); 232#endif 233 234} 235 236static struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 237static struct route ipforward_rt; 238 239/* 240 * Ip input routine. Checksum and byte swap header. If fragmented 241 * try to reassemble. Process options. Pass to next level. 242 */ 243void 244ip_input(struct mbuf *m) 245{ 246 struct ip *ip; 247 struct ipq *fp; 248 struct in_ifaddr *ia; 249 int i, hlen; 250 u_short sum; 251 252#ifdef DIAGNOSTIC 253 if ((m->m_flags & M_PKTHDR) == 0) 254 panic("ip_input no HDR"); 255#endif 256 /* 257 * If no IP addresses have been set yet but the interfaces 258 * are receiving, can't do anything with incoming packets yet. 259 * XXX This is broken! We should be able to receive broadcasts 260 * and multicasts even without any local addresses configured. 261 */ 262 if (TAILQ_EMPTY(&in_ifaddrhead)) 263 goto bad; 264 ipstat.ips_total++; 265 266 if (m->m_pkthdr.len < sizeof(struct ip)) 267 goto tooshort; 268 269#ifdef DIAGNOSTIC 270 if (m->m_len < sizeof(struct ip)) 271 panic("ipintr mbuf too short"); 272#endif 273 274 if (m->m_len < sizeof (struct ip) && 275 (m = m_pullup(m, sizeof (struct ip))) == 0) { 276 ipstat.ips_toosmall++; 277 return; 278 } 279 ip = mtod(m, struct ip *); 280 281 if (IP_VHL_V(ip->ip_vhl) != IPVERSION) { 282 ipstat.ips_badvers++; 283 goto bad; 284 } 285 286 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 287 if (hlen < sizeof(struct ip)) { /* minimum header length */ 288 ipstat.ips_badhlen++; 289 goto bad; 290 } 291 if (hlen > m->m_len) { 292 if ((m = m_pullup(m, hlen)) == 0) { 293 ipstat.ips_badhlen++; 294 return; 295 } 296 ip = mtod(m, struct ip *); 297 } 298 if (hlen == sizeof(struct ip)) { 299 sum = in_cksum_hdr(ip); 300 } else { 301 sum = in_cksum(m, hlen); 302 } 303 if (sum) { 304 ipstat.ips_badsum++; 305 goto bad; 306 } 307 308 /* 309 * Convert fields to host representation. 310 */ 311 NTOHS(ip->ip_len); 312 if (ip->ip_len < hlen) { 313 ipstat.ips_badlen++; 314 goto bad; 315 } 316 NTOHS(ip->ip_id); 317 NTOHS(ip->ip_off); 318 319 /* 320 * Check that the amount of data in the buffers 321 * is as at least much as the IP header would have us expect. 322 * Trim mbufs if longer than we expect. 323 * Drop packet if shorter than we expect. 324 */ 325 if (m->m_pkthdr.len < ip->ip_len) { 326tooshort: 327 ipstat.ips_tooshort++; 328 goto bad; 329 } 330 if (m->m_pkthdr.len > ip->ip_len) { 331 if (m->m_len == m->m_pkthdr.len) { 332 m->m_len = ip->ip_len; 333 m->m_pkthdr.len = ip->ip_len; 334 } else 335 m_adj(m, ip->ip_len - m->m_pkthdr.len); 336 } 337 /* 338 * IpHack's section. 339 * Right now when no processing on packet has done 340 * and it is still fresh out of network we do our black 341 * deals with it. 342 * - Firewall: deny/allow/divert 343 * - Xlate: translate packet's addr/port (NAT). 344 * - Wrap: fake packet's addr/port <unimpl.> 345 * - Encapsulate: put it in another IP and send out. <unimp.> 346 */ 347#if defined(IPFILTER) || defined(IPFILTER_LKM) 348 /* 349 * Check if we want to allow this packet to be processed. 350 * Consider it to be bad if not. 351 */ 352 if (fr_checkp) { 353 struct mbuf *m1 = m; 354 355 if ((*fr_checkp)(ip, hlen, m->m_pkthdr.rcvif, 0, &m1) || !m1) 356 return; 357 ip = mtod(m = m1, struct ip *); 358 } 359#endif 360#ifdef COMPAT_IPFW 361 if (ip_fw_chk_ptr) { 362#ifdef IPDIVERT 363 u_short port; 364 365 port = (*ip_fw_chk_ptr)(&ip, hlen, NULL, ip_divert_ignore, &m); 366 ip_divert_ignore = 0; 367 if (port) { /* Divert packet */ 368 frag_divert_port = port; 369 goto ours; 370 } 371#else 372 /* If ipfw says divert, we have to just drop packet */ 373 if ((*ip_fw_chk_ptr)(&ip, hlen, NULL, 0, &m)) { 374 m_freem(m); 375 m = NULL; 376 } 377#endif 378 if (!m) 379 return; 380 } 381 382 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, m->m_pkthdr.rcvif, IP_NAT_IN)) 383 return; 384#endif 385 386 /* 387 * Process options and, if not destined for us, 388 * ship it on. ip_dooptions returns 1 when an 389 * error was detected (causing an icmp message 390 * to be sent and the original packet to be freed). 391 */ 392 ip_nhops = 0; /* for source routed packets */ 393 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 394 return; 395 396 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no 397 * matter if it is destined to another node, or whether it is 398 * a multicast one, RSVP wants it! and prevents it from being forwarded 399 * anywhere else. Also checks if the rsvp daemon is running before 400 * grabbing the packet. 401 */ 402 if (rsvp_on && ip->ip_p==IPPROTO_RSVP) 403 goto ours; 404 405 /* 406 * Check our list of addresses, to see if the packet is for us. 407 */ 408 for (ia = in_ifaddrhead.tqh_first; ia; ia = ia->ia_link.tqe_next) { 409#define satosin(sa) ((struct sockaddr_in *)(sa)) 410 411 if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) 412 goto ours; 413#ifdef BOOTP_COMPAT 414 if (IA_SIN(ia)->sin_addr.s_addr == INADDR_ANY) 415 goto ours; 416#endif 417 if (ia->ia_ifp && ia->ia_ifp->if_flags & IFF_BROADCAST) { 418 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 419 ip->ip_dst.s_addr) 420 goto ours; 421 if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) 422 goto ours; 423 } 424 } 425 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 426 struct in_multi *inm; 427 if (ip_mrouter) { 428 /* 429 * If we are acting as a multicast router, all 430 * incoming multicast packets are passed to the 431 * kernel-level multicast forwarding function. 432 * The packet is returned (relatively) intact; if 433 * ip_mforward() returns a non-zero value, the packet 434 * must be discarded, else it may be accepted below. 435 * 436 * (The IP ident field is put in the same byte order 437 * as expected when ip_mforward() is called from 438 * ip_output().) 439 */ 440 ip->ip_id = htons(ip->ip_id); 441 if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) { 442 ipstat.ips_cantforward++; 443 m_freem(m); 444 return; 445 } 446 ip->ip_id = ntohs(ip->ip_id); 447 448 /* 449 * The process-level routing demon needs to receive 450 * all multicast IGMP packets, whether or not this 451 * host belongs to their destination groups. 452 */ 453 if (ip->ip_p == IPPROTO_IGMP) 454 goto ours; 455 ipstat.ips_forward++; 456 } 457 /* 458 * See if we belong to the destination multicast group on the 459 * arrival interface. 460 */ 461 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 462 if (inm == NULL) { 463 ipstat.ips_notmember++; 464 m_freem(m); 465 return; 466 } 467 goto ours; 468 } 469 if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) 470 goto ours; 471 if (ip->ip_dst.s_addr == INADDR_ANY) 472 goto ours; 473 474 /* 475 * Not for us; forward if possible and desirable. 476 */ 477 if (ipforwarding == 0) { 478 ipstat.ips_cantforward++; 479 m_freem(m); 480 } else 481 ip_forward(m, 0); 482 return; 483 484ours: 485 486 /* 487 * If offset or IP_MF are set, must reassemble. 488 * Otherwise, nothing need be done. 489 * (We could look in the reassembly queue to see 490 * if the packet was previously fragmented, 491 * but it's not worth the time; just let them time out.) 492 */ 493 if (ip->ip_off & (IP_MF | IP_OFFMASK | IP_RF)) { 494 if (m->m_flags & M_EXT) { /* XXX */ 495 if ((m = m_pullup(m, sizeof (struct ip))) == 0) { 496 ipstat.ips_toosmall++; 497#ifdef IPDIVERT 498 frag_divert_port = 0; 499#endif 500 return; 501 } 502 ip = mtod(m, struct ip *); 503 } 504 sum = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id); 505 /* 506 * Look for queue of fragments 507 * of this datagram. 508 */ 509 for (fp = ipq[sum].next; fp != &ipq[sum]; fp = fp->next) 510 if (ip->ip_id == fp->ipq_id && 511 ip->ip_src.s_addr == fp->ipq_src.s_addr && 512 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 513 ip->ip_p == fp->ipq_p) 514 goto found; 515 516 fp = 0; 517 518 /* check if there's a place for the new queue */ 519 if (nipq > maxnipq) { 520 /* 521 * drop something from the tail of the current queue 522 * before proceeding further 523 */ 524 if (ipq[sum].prev == &ipq[sum]) { /* gak */ 525 for (i = 0; i < IPREASS_NHASH; i++) { 526 if (ipq[i].prev != &ipq[i]) { 527 ip_freef(ipq[i].prev); 528 break; 529 } 530 } 531 } else 532 ip_freef(ipq[sum].prev); 533 } 534found: 535 /* 536 * Adjust ip_len to not reflect header, 537 * set ip_mff if more fragments are expected, 538 * convert offset of this to bytes. 539 */ 540 ip->ip_len -= hlen; 541 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 542 if (ip->ip_off & IP_MF) 543 ((struct ipasfrag *)ip)->ipf_mff |= 1; 544 ip->ip_off <<= 3; 545 546 /* 547 * If datagram marked as having more fragments 548 * or if this is not the first fragment, 549 * attempt reassembly; if it succeeds, proceed. 550 */ 551 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 552 ipstat.ips_fragments++; 553 ip = ip_reass((struct ipasfrag *)ip, fp, &ipq[sum]); 554 if (ip == 0) 555 return; 556 ipstat.ips_reassembled++; 557 m = dtom(ip); 558#ifdef IPDIVERT 559 if (frag_divert_port) { 560 ip->ip_len += hlen; 561 HTONS(ip->ip_len); 562 HTONS(ip->ip_off); 563 HTONS(ip->ip_id); 564 ip->ip_sum = 0; 565 ip->ip_sum = in_cksum_hdr(ip); 566 NTOHS(ip->ip_id); 567 NTOHS(ip->ip_off); 568 NTOHS(ip->ip_len); 569 ip->ip_len -= hlen; 570 } 571#endif 572 } else 573 if (fp) 574 ip_freef(fp); 575 } else 576 ip->ip_len -= hlen; 577 578#ifdef IPDIVERT 579 /* 580 * Divert reassembled packets to the divert protocol if required 581 */ 582 if (frag_divert_port) { 583 ipstat.ips_delivered++; 584 ip_divert_port = frag_divert_port; 585 frag_divert_port = 0; 586 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, hlen); 587 return; 588 } 589 590 /* Don't let packets divert themselves */ 591 if (ip->ip_p == IPPROTO_DIVERT) { 592 ipstat.ips_noproto++; 593 goto bad; 594 } 595#endif 596 597 /* 598 * Switch out to protocol's input routine. 599 */ 600 ipstat.ips_delivered++; 601 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 602 return; 603bad: 604 m_freem(m); 605} 606 607/* 608 * IP software interrupt routine - to go away sometime soon 609 */ 610static void 611ipintr(void) 612{ 613 int s; 614 struct mbuf *m; 615 616 while(1) { 617 s = splimp(); 618 IF_DEQUEUE(&ipintrq, m); 619 splx(s); 620 if (m == 0) 621 return; 622 ip_input(m); 623 } 624} 625 626NETISR_SET(NETISR_IP, ipintr); 627 628/* 629 * Take incoming datagram fragment and try to 630 * reassemble it into whole datagram. If a chain for 631 * reassembly of this datagram already exists, then it 632 * is given as fp; otherwise have to make a chain. 633 */ 634static struct ip * 635ip_reass(ip, fp, where) 636 register struct ipasfrag *ip; 637 register struct ipq *fp; 638 struct ipq *where; 639{ 640 register struct mbuf *m = dtom(ip); 641 register struct ipasfrag *q; 642 struct mbuf *t; 643 int hlen = ip->ip_hl << 2; 644 int i, next; 645 646 /* 647 * Presence of header sizes in mbufs 648 * would confuse code below. 649 */ 650 m->m_data += hlen; 651 m->m_len -= hlen; 652 653 /* 654 * If first fragment to arrive, create a reassembly queue. 655 */ 656 if (fp == 0) { 657 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 658 goto dropfrag; 659 fp = mtod(t, struct ipq *); 660 insque(fp, where); 661 nipq++; 662 fp->ipq_ttl = IPFRAGTTL; 663 fp->ipq_p = ip->ip_p; 664 fp->ipq_id = ip->ip_id; 665 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 666 fp->ipq_src = ((struct ip *)ip)->ip_src; 667 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 668#ifdef IPDIVERT 669 fp->ipq_divert = 0; 670#endif 671 q = (struct ipasfrag *)fp; 672 goto insert; 673 } 674 675 /* 676 * Find a segment which begins after this one does. 677 */ 678 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 679 if (q->ip_off > ip->ip_off) 680 break; 681 682 /* 683 * If there is a preceding segment, it may provide some of 684 * our data already. If so, drop the data from the incoming 685 * segment. If it provides all of our data, drop us. 686 */ 687 if (q->ipf_prev != (struct ipasfrag *)fp) { 688 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 689 if (i > 0) { 690 if (i >= ip->ip_len) 691 goto dropfrag; 692 m_adj(dtom(ip), i); 693 ip->ip_off += i; 694 ip->ip_len -= i; 695 } 696 } 697 698 /* 699 * While we overlap succeeding segments trim them or, 700 * if they are completely covered, dequeue them. 701 */ 702 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 703 struct mbuf *m0; 704 705 i = (ip->ip_off + ip->ip_len) - q->ip_off; 706 if (i < q->ip_len) { 707 q->ip_len -= i; 708 q->ip_off += i; 709 m_adj(dtom(q), i); 710 break; 711 } 712 m0 = dtom(q); 713 q = q->ipf_next; 714 ip_deq(q->ipf_prev); 715 m_freem(m0); 716 } 717 718insert: 719 720#ifdef IPDIVERT 721 /* 722 * Any fragment diverting causes the whole packet to divert 723 */ 724 if (frag_divert_port != 0) 725 fp->ipq_divert = frag_divert_port; 726 frag_divert_port = 0; 727#endif 728 729 /* 730 * Stick new segment in its place; 731 * check for complete reassembly. 732 */ 733 ip_enq(ip, q->ipf_prev); 734 next = 0; 735 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 736 if (q->ip_off != next) 737 return (0); 738 next += q->ip_len; 739 } 740 if (q->ipf_prev->ipf_mff & 1) 741 return (0); 742 743 /* 744 * Reassembly is complete. Make sure the packet is a sane size. 745 */ 746 if (next + (IP_VHL_HL(((struct ip *)fp->ipq_next)->ip_vhl) << 2) 747 > IP_MAXPACKET) { 748 ipstat.ips_toolong++; 749 ip_freef(fp); 750 return (0); 751 } 752 753 /* 754 * Concatenate fragments. 755 */ 756 q = fp->ipq_next; 757 m = dtom(q); 758 t = m->m_next; 759 m->m_next = 0; 760 m_cat(m, t); 761 q = q->ipf_next; 762 while (q != (struct ipasfrag *)fp) { 763 t = dtom(q); 764 q = q->ipf_next; 765 m_cat(m, t); 766 } 767 768#ifdef IPDIVERT 769 /* 770 * Record divert port for packet, if any 771 */ 772 frag_divert_port = fp->ipq_divert; 773#endif 774 775 /* 776 * Create header for new ip packet by 777 * modifying header of first packet; 778 * dequeue and discard fragment reassembly header. 779 * Make header visible. 780 */ 781 ip = fp->ipq_next; 782 ip->ip_len = next; 783 ip->ipf_mff &= ~1; 784 ((struct ip *)ip)->ip_src = fp->ipq_src; 785 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 786 remque(fp); 787 nipq--; 788 (void) m_free(dtom(fp)); 789 m = dtom(ip); 790 m->m_len += (ip->ip_hl << 2); 791 m->m_data -= (ip->ip_hl << 2); 792 /* some debugging cruft by sklower, below, will go away soon */ 793 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 794 register int plen = 0; 795 for (t = m; m; m = m->m_next) 796 plen += m->m_len; 797 t->m_pkthdr.len = plen; 798 } 799 return ((struct ip *)ip); 800 801dropfrag: 802 ipstat.ips_fragdropped++; 803 m_freem(m); 804 return (0); 805} 806 807/* 808 * Free a fragment reassembly header and all 809 * associated datagrams. 810 */ 811static void 812ip_freef(fp) 813 struct ipq *fp; 814{ 815 register struct ipasfrag *q, *p; 816 817 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 818 p = q->ipf_next; 819 ip_deq(q); 820 m_freem(dtom(q)); 821 } 822 remque(fp); 823 (void) m_free(dtom(fp)); 824 nipq--; 825} 826 827/* 828 * Put an ip fragment on a reassembly chain. 829 * Like insque, but pointers in middle of structure. 830 */ 831static void 832ip_enq(p, prev) 833 register struct ipasfrag *p, *prev; 834{ 835 836 p->ipf_prev = prev; 837 p->ipf_next = prev->ipf_next; 838 prev->ipf_next->ipf_prev = p; 839 prev->ipf_next = p; 840} 841 842/* 843 * To ip_enq as remque is to insque. 844 */ 845static void 846ip_deq(p) 847 register struct ipasfrag *p; 848{ 849 850 p->ipf_prev->ipf_next = p->ipf_next; 851 p->ipf_next->ipf_prev = p->ipf_prev; 852} 853 854/* 855 * IP timer processing; 856 * if a timer expires on a reassembly 857 * queue, discard it. 858 */ 859void 860ip_slowtimo() 861{ 862 register struct ipq *fp; 863 int s = splnet(); 864 int i; 865 866 for (i = 0; i < IPREASS_NHASH; i++) { 867 fp = ipq[i].next; 868 if (fp == 0) 869 continue; 870 while (fp != &ipq[i]) { 871 --fp->ipq_ttl; 872 fp = fp->next; 873 if (fp->prev->ipq_ttl == 0) { 874 ipstat.ips_fragtimeout++; 875 ip_freef(fp->prev); 876 } 877 } 878 } 879 splx(s); 880} 881 882/* 883 * Drain off all datagram fragments. 884 */ 885void 886ip_drain() 887{ 888 int i; 889 890 for (i = 0; i < IPREASS_NHASH; i++) { 891 while (ipq[i].next != &ipq[i]) { 892 ipstat.ips_fragdropped++; 893 ip_freef(ipq[i].next); 894 } 895 } 896 in_rtqdrain(); 897} 898 899/* 900 * Do option processing on a datagram, 901 * possibly discarding it if bad options are encountered, 902 * or forwarding it if source-routed. 903 * Returns 1 if packet has been forwarded/freed, 904 * 0 if the packet should be processed further. 905 */ 906static int 907ip_dooptions(m) 908 struct mbuf *m; 909{ 910 register struct ip *ip = mtod(m, struct ip *); 911 register u_char *cp; 912 register struct ip_timestamp *ipt; 913 register struct in_ifaddr *ia; 914 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 915 struct in_addr *sin, dst; 916 n_time ntime; 917 918 dst = ip->ip_dst; 919 cp = (u_char *)(ip + 1); 920 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 921 for (; cnt > 0; cnt -= optlen, cp += optlen) { 922 opt = cp[IPOPT_OPTVAL]; 923 if (opt == IPOPT_EOL) 924 break; 925 if (opt == IPOPT_NOP) 926 optlen = 1; 927 else { 928 optlen = cp[IPOPT_OLEN]; 929 if (optlen <= 0 || optlen > cnt) { 930 code = &cp[IPOPT_OLEN] - (u_char *)ip; 931 goto bad; 932 } 933 } 934 switch (opt) { 935 936 default: 937 break; 938 939 /* 940 * Source routing with record. 941 * Find interface with current destination address. 942 * If none on this machine then drop if strictly routed, 943 * or do nothing if loosely routed. 944 * Record interface address and bring up next address 945 * component. If strictly routed make sure next 946 * address is on directly accessible net. 947 */ 948 case IPOPT_LSRR: 949 case IPOPT_SSRR: 950 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 951 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 952 goto bad; 953 } 954 ipaddr.sin_addr = ip->ip_dst; 955 ia = (struct in_ifaddr *) 956 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 957 if (ia == 0) { 958 if (opt == IPOPT_SSRR) { 959 type = ICMP_UNREACH; 960 code = ICMP_UNREACH_SRCFAIL; 961 goto bad; 962 } 963 if (!ip_dosourceroute) 964 goto nosourcerouting; 965 /* 966 * Loose routing, and not at next destination 967 * yet; nothing to do except forward. 968 */ 969 break; 970 } 971 off--; /* 0 origin */ 972 if (off > optlen - sizeof(struct in_addr)) { 973 /* 974 * End of source route. Should be for us. 975 */ 976 if (!ip_acceptsourceroute) 977 goto nosourcerouting; 978 save_rte(cp, ip->ip_src); 979 break; 980 } 981 982 if (!ip_dosourceroute) { 983 char buf[4*sizeof "123"]; 984 985nosourcerouting: 986 strcpy(buf, inet_ntoa(ip->ip_dst)); 987 log(LOG_WARNING, 988 "attempted source route from %s to %s\n", 989 inet_ntoa(ip->ip_src), buf); 990 type = ICMP_UNREACH; 991 code = ICMP_UNREACH_SRCFAIL; 992 goto bad; 993 } 994 995 /* 996 * locate outgoing interface 997 */ 998 (void)memcpy(&ipaddr.sin_addr, cp + off, 999 sizeof(ipaddr.sin_addr)); 1000 1001 if (opt == IPOPT_SSRR) { 1002#define INA struct in_ifaddr * 1003#define SA struct sockaddr * 1004 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 1005 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 1006 } else 1007 ia = ip_rtaddr(ipaddr.sin_addr); 1008 if (ia == 0) { 1009 type = ICMP_UNREACH; 1010 code = ICMP_UNREACH_SRCFAIL; 1011 goto bad; 1012 } 1013 ip->ip_dst = ipaddr.sin_addr; 1014 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 1015 sizeof(struct in_addr)); 1016 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1017 /* 1018 * Let ip_intr's mcast routing check handle mcast pkts 1019 */ 1020 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 1021 break; 1022 1023 case IPOPT_RR: 1024 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1025 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1026 goto bad; 1027 } 1028 /* 1029 * If no space remains, ignore. 1030 */ 1031 off--; /* 0 origin */ 1032 if (off > optlen - sizeof(struct in_addr)) 1033 break; 1034 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst, 1035 sizeof(ipaddr.sin_addr)); 1036 /* 1037 * locate outgoing interface; if we're the destination, 1038 * use the incoming interface (should be same). 1039 */ 1040 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 1041 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 1042 type = ICMP_UNREACH; 1043 code = ICMP_UNREACH_HOST; 1044 goto bad; 1045 } 1046 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 1047 sizeof(struct in_addr)); 1048 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1049 break; 1050 1051 case IPOPT_TS: 1052 code = cp - (u_char *)ip; 1053 ipt = (struct ip_timestamp *)cp; 1054 if (ipt->ipt_len < 5) 1055 goto bad; 1056 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 1057 if (++ipt->ipt_oflw == 0) 1058 goto bad; 1059 break; 1060 } 1061 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 1062 switch (ipt->ipt_flg) { 1063 1064 case IPOPT_TS_TSONLY: 1065 break; 1066 1067 case IPOPT_TS_TSANDADDR: 1068 if (ipt->ipt_ptr + sizeof(n_time) + 1069 sizeof(struct in_addr) > ipt->ipt_len) 1070 goto bad; 1071 ipaddr.sin_addr = dst; 1072 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 1073 m->m_pkthdr.rcvif); 1074 if (ia == 0) 1075 continue; 1076 (void)memcpy(sin, &IA_SIN(ia)->sin_addr, 1077 sizeof(struct in_addr)); 1078 ipt->ipt_ptr += sizeof(struct in_addr); 1079 break; 1080 1081 case IPOPT_TS_PRESPEC: 1082 if (ipt->ipt_ptr + sizeof(n_time) + 1083 sizeof(struct in_addr) > ipt->ipt_len) 1084 goto bad; 1085 (void)memcpy(&ipaddr.sin_addr, sin, 1086 sizeof(struct in_addr)); 1087 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 1088 continue; 1089 ipt->ipt_ptr += sizeof(struct in_addr); 1090 break; 1091 1092 default: 1093 goto bad; 1094 } 1095 ntime = iptime(); 1096 (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime, 1097 sizeof(n_time)); 1098 ipt->ipt_ptr += sizeof(n_time); 1099 } 1100 } 1101 if (forward && ipforwarding) { 1102 ip_forward(m, 1); 1103 return (1); 1104 } 1105 return (0); 1106bad: 1107 ip->ip_len -= IP_VHL_HL(ip->ip_vhl) << 2; /* XXX icmp_error adds in hdr length */ 1108 icmp_error(m, type, code, 0, 0); 1109 ipstat.ips_badoptions++; 1110 return (1); 1111} 1112 1113/* 1114 * Given address of next destination (final or next hop), 1115 * return internet address info of interface to be used to get there. 1116 */ 1117static struct in_ifaddr * 1118ip_rtaddr(dst) 1119 struct in_addr dst; 1120{ 1121 register struct sockaddr_in *sin; 1122 1123 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 1124 1125 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 1126 if (ipforward_rt.ro_rt) { 1127 RTFREE(ipforward_rt.ro_rt); 1128 ipforward_rt.ro_rt = 0; 1129 } 1130 sin->sin_family = AF_INET; 1131 sin->sin_len = sizeof(*sin); 1132 sin->sin_addr = dst; 1133 1134 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1135 } 1136 if (ipforward_rt.ro_rt == 0) 1137 return ((struct in_ifaddr *)0); 1138 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 1139} 1140 1141/* 1142 * Save incoming source route for use in replies, 1143 * to be picked up later by ip_srcroute if the receiver is interested. 1144 */ 1145void 1146save_rte(option, dst) 1147 u_char *option; 1148 struct in_addr dst; 1149{ 1150 unsigned olen; 1151 1152 olen = option[IPOPT_OLEN]; 1153#ifdef DIAGNOSTIC 1154 if (ipprintfs) 1155 printf("save_rte: olen %d\n", olen); 1156#endif 1157 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1158 return; 1159 bcopy(option, ip_srcrt.srcopt, olen); 1160 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1161 ip_srcrt.dst = dst; 1162} 1163 1164/* 1165 * Retrieve incoming source route for use in replies, 1166 * in the same form used by setsockopt. 1167 * The first hop is placed before the options, will be removed later. 1168 */ 1169struct mbuf * 1170ip_srcroute() 1171{ 1172 register struct in_addr *p, *q; 1173 register struct mbuf *m; 1174 1175 if (ip_nhops == 0) 1176 return ((struct mbuf *)0); 1177 m = m_get(M_DONTWAIT, MT_SOOPTS); 1178 if (m == 0) 1179 return ((struct mbuf *)0); 1180 1181#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1182 1183 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1184 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1185 OPTSIZ; 1186#ifdef DIAGNOSTIC 1187 if (ipprintfs) 1188 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1189#endif 1190 1191 /* 1192 * First save first hop for return route 1193 */ 1194 p = &ip_srcrt.route[ip_nhops - 1]; 1195 *(mtod(m, struct in_addr *)) = *p--; 1196#ifdef DIAGNOSTIC 1197 if (ipprintfs) 1198 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 1199#endif 1200 1201 /* 1202 * Copy option fields and padding (nop) to mbuf. 1203 */ 1204 ip_srcrt.nop = IPOPT_NOP; 1205 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1206 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), 1207 &ip_srcrt.nop, OPTSIZ); 1208 q = (struct in_addr *)(mtod(m, caddr_t) + 1209 sizeof(struct in_addr) + OPTSIZ); 1210#undef OPTSIZ 1211 /* 1212 * Record return path as an IP source route, 1213 * reversing the path (pointers are now aligned). 1214 */ 1215 while (p >= ip_srcrt.route) { 1216#ifdef DIAGNOSTIC 1217 if (ipprintfs) 1218 printf(" %lx", ntohl(q->s_addr)); 1219#endif 1220 *q++ = *p--; 1221 } 1222 /* 1223 * Last hop goes to final destination. 1224 */ 1225 *q = ip_srcrt.dst; 1226#ifdef DIAGNOSTIC 1227 if (ipprintfs) 1228 printf(" %lx\n", ntohl(q->s_addr)); 1229#endif 1230 return (m); 1231} 1232 1233/* 1234 * Strip out IP options, at higher 1235 * level protocol in the kernel. 1236 * Second argument is buffer to which options 1237 * will be moved, and return value is their length. 1238 * XXX should be deleted; last arg currently ignored. 1239 */ 1240void 1241ip_stripoptions(m, mopt) 1242 register struct mbuf *m; 1243 struct mbuf *mopt; 1244{ 1245 register int i; 1246 struct ip *ip = mtod(m, struct ip *); 1247 register caddr_t opts; 1248 int olen; 1249 1250 olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 1251 opts = (caddr_t)(ip + 1); 1252 i = m->m_len - (sizeof (struct ip) + olen); 1253 bcopy(opts + olen, opts, (unsigned)i); 1254 m->m_len -= olen; 1255 if (m->m_flags & M_PKTHDR) 1256 m->m_pkthdr.len -= olen; 1257 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2); 1258} 1259 1260u_char inetctlerrmap[PRC_NCMDS] = { 1261 0, 0, 0, 0, 1262 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1263 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1264 EMSGSIZE, EHOSTUNREACH, 0, 0, 1265 0, 0, 0, 0, 1266 ENOPROTOOPT 1267}; 1268 1269/* 1270 * Forward a packet. If some error occurs return the sender 1271 * an icmp packet. Note we can't always generate a meaningful 1272 * icmp message because icmp doesn't have a large enough repertoire 1273 * of codes and types. 1274 * 1275 * If not forwarding, just drop the packet. This could be confusing 1276 * if ipforwarding was zero but some routing protocol was advancing 1277 * us as a gateway to somewhere. However, we must let the routing 1278 * protocol deal with that. 1279 * 1280 * The srcrt parameter indicates whether the packet is being forwarded 1281 * via a source route. 1282 */ 1283static void 1284ip_forward(m, srcrt) 1285 struct mbuf *m; 1286 int srcrt; 1287{ 1288 register struct ip *ip = mtod(m, struct ip *); 1289 register struct sockaddr_in *sin; 1290 register struct rtentry *rt; 1291 int error, type = 0, code = 0; 1292 struct mbuf *mcopy; 1293 n_long dest; 1294 struct ifnet *destifp; 1295 1296 dest = 0; 1297#ifdef DIAGNOSTIC 1298 if (ipprintfs) 1299 printf("forward: src %lx dst %lx ttl %x\n", 1300 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1301#endif 1302 1303 1304 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1305 ipstat.ips_cantforward++; 1306 m_freem(m); 1307 return; 1308 } 1309 HTONS(ip->ip_id); 1310 if (ip->ip_ttl <= IPTTLDEC) { 1311 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1312 return; 1313 } 1314 ip->ip_ttl -= IPTTLDEC; 1315 1316 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1317 if ((rt = ipforward_rt.ro_rt) == 0 || 1318 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1319 if (ipforward_rt.ro_rt) { 1320 RTFREE(ipforward_rt.ro_rt); 1321 ipforward_rt.ro_rt = 0; 1322 } 1323 sin->sin_family = AF_INET; 1324 sin->sin_len = sizeof(*sin); 1325 sin->sin_addr = ip->ip_dst; 1326 1327 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1328 if (ipforward_rt.ro_rt == 0) { 1329 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1330 return; 1331 } 1332 rt = ipforward_rt.ro_rt; 1333 } 1334 1335 /* 1336 * Save at most 64 bytes of the packet in case 1337 * we need to generate an ICMP message to the src. 1338 */ 1339 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1340 1341 /* 1342 * If forwarding packet using same interface that it came in on, 1343 * perhaps should send a redirect to sender to shortcut a hop. 1344 * Only send redirect if source is sending directly to us, 1345 * and if packet was not source routed (or has any options). 1346 * Also, don't send redirect if forwarding using a default route 1347 * or a route modified by a redirect. 1348 */ 1349#define satosin(sa) ((struct sockaddr_in *)(sa)) 1350 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1351 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1352 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1353 ipsendredirects && !srcrt) { 1354#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1355 u_long src = ntohl(ip->ip_src.s_addr); 1356 1357 if (RTA(rt) && 1358 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1359 if (rt->rt_flags & RTF_GATEWAY) 1360 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1361 else 1362 dest = ip->ip_dst.s_addr; 1363 /* Router requirements says to only send host redirects */ 1364 type = ICMP_REDIRECT; 1365 code = ICMP_REDIRECT_HOST; 1366#ifdef DIAGNOSTIC 1367 if (ipprintfs) 1368 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1369#endif 1370 } 1371 } 1372 1373 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1374 IP_FORWARDING, 0); 1375 if (error) 1376 ipstat.ips_cantforward++; 1377 else { 1378 ipstat.ips_forward++; 1379 if (type) 1380 ipstat.ips_redirectsent++; 1381 else { 1382 if (mcopy) 1383 m_freem(mcopy); 1384 return; 1385 } 1386 } 1387 if (mcopy == NULL) 1388 return; 1389 destifp = NULL; 1390 1391 switch (error) { 1392 1393 case 0: /* forwarded, but need redirect */ 1394 /* type, code set above */ 1395 break; 1396 1397 case ENETUNREACH: /* shouldn't happen, checked above */ 1398 case EHOSTUNREACH: 1399 case ENETDOWN: 1400 case EHOSTDOWN: 1401 default: 1402 type = ICMP_UNREACH; 1403 code = ICMP_UNREACH_HOST; 1404 break; 1405 1406 case EMSGSIZE: 1407 type = ICMP_UNREACH; 1408 code = ICMP_UNREACH_NEEDFRAG; 1409 if (ipforward_rt.ro_rt) 1410 destifp = ipforward_rt.ro_rt->rt_ifp; 1411 ipstat.ips_cantfrag++; 1412 break; 1413 1414 case ENOBUFS: 1415 type = ICMP_SOURCEQUENCH; 1416 code = 0; 1417 break; 1418 } 1419 icmp_error(mcopy, type, code, dest, destifp); 1420} 1421 1422void 1423ip_savecontrol(inp, mp, ip, m) 1424 register struct inpcb *inp; 1425 register struct mbuf **mp; 1426 register struct ip *ip; 1427 register struct mbuf *m; 1428{ 1429 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1430 struct timeval tv; 1431 1432 microtime(&tv); 1433 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1434 SCM_TIMESTAMP, SOL_SOCKET); 1435 if (*mp) 1436 mp = &(*mp)->m_next; 1437 } 1438 if (inp->inp_flags & INP_RECVDSTADDR) { 1439 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1440 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1441 if (*mp) 1442 mp = &(*mp)->m_next; 1443 } 1444#ifdef notyet 1445 /* XXX 1446 * Moving these out of udp_input() made them even more broken 1447 * than they already were. 1448 */ 1449 /* options were tossed already */ 1450 if (inp->inp_flags & INP_RECVOPTS) { 1451 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1452 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1453 if (*mp) 1454 mp = &(*mp)->m_next; 1455 } 1456 /* ip_srcroute doesn't do what we want here, need to fix */ 1457 if (inp->inp_flags & INP_RECVRETOPTS) { 1458 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1459 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1460 if (*mp) 1461 mp = &(*mp)->m_next; 1462 } 1463#endif 1464 if (inp->inp_flags & INP_RECVIF) { 1465 struct ifnet *ifp; 1466 struct sdlbuf { 1467 struct sockaddr_dl sdl; 1468 u_char pad[32]; 1469 } sdlbuf; 1470 struct sockaddr_dl *sdp; 1471 struct sockaddr_dl *sdl2 = &sdlbuf.sdl; 1472 1473 if (((ifp = m->m_pkthdr.rcvif)) 1474 && ( ifp->if_index && (ifp->if_index <= if_index))) { 1475 sdp = (struct sockaddr_dl *)(ifnet_addrs 1476 [ifp->if_index - 1]->ifa_addr); 1477 /* 1478 * Change our mind and don't try copy. 1479 */ 1480 if ((sdp->sdl_family != AF_LINK) 1481 || (sdp->sdl_len > sizeof(sdlbuf))) { 1482 goto makedummy; 1483 } 1484 bcopy(sdp, sdl2, sdp->sdl_len); 1485 } else { 1486makedummy: 1487 sdl2->sdl_len 1488 = offsetof(struct sockaddr_dl, sdl_data[0]); 1489 sdl2->sdl_family = AF_LINK; 1490 sdl2->sdl_index = 0; 1491 sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0; 1492 } 1493 *mp = sbcreatecontrol((caddr_t) sdl2, sdl2->sdl_len, 1494 IP_RECVIF, IPPROTO_IP); 1495 if (*mp) 1496 mp = &(*mp)->m_next; 1497 } 1498} 1499 1500int 1501ip_rsvp_init(struct socket *so) 1502{ 1503 if (so->so_type != SOCK_RAW || 1504 so->so_proto->pr_protocol != IPPROTO_RSVP) 1505 return EOPNOTSUPP; 1506 1507 if (ip_rsvpd != NULL) 1508 return EADDRINUSE; 1509 1510 ip_rsvpd = so; 1511 /* 1512 * This may seem silly, but we need to be sure we don't over-increment 1513 * the RSVP counter, in case something slips up. 1514 */ 1515 if (!ip_rsvp_on) { 1516 ip_rsvp_on = 1; 1517 rsvp_on++; 1518 } 1519 1520 return 0; 1521} 1522 1523int 1524ip_rsvp_done(void) 1525{ 1526 ip_rsvpd = NULL; 1527 /* 1528 * This may seem silly, but we need to be sure we don't over-decrement 1529 * the RSVP counter, in case something slips up. 1530 */ 1531 if (ip_rsvp_on) { 1532 ip_rsvp_on = 0; 1533 rsvp_on--; 1534 } 1535 return 0; 1536} 1537