ip_input.c revision 36707
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 1.86 1998/06/05 22:39:55 julian 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 83int 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_second & 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_cookie, &m); 366 if (port) { /* Divert packet */ 367 frag_divert_port = port; 368 goto ours; 369 } 370#else 371 int dummy; 372 /* If ipfw says divert, we have to just drop packet */ 373 if ((*ip_fw_chk_ptr)(&ip, hlen, NULL, &dummy, &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 ip_divert_cookie = 0; 500#endif 501 return; 502 } 503 ip = mtod(m, struct ip *); 504 } 505 sum = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id); 506 /* 507 * Look for queue of fragments 508 * of this datagram. 509 */ 510 for (fp = ipq[sum].next; fp != &ipq[sum]; fp = fp->next) 511 if (ip->ip_id == fp->ipq_id && 512 ip->ip_src.s_addr == fp->ipq_src.s_addr && 513 ip->ip_dst.s_addr == fp->ipq_dst.s_addr && 514 ip->ip_p == fp->ipq_p) 515 goto found; 516 517 fp = 0; 518 519 /* check if there's a place for the new queue */ 520 if (nipq > maxnipq) { 521 /* 522 * drop something from the tail of the current queue 523 * before proceeding further 524 */ 525 if (ipq[sum].prev == &ipq[sum]) { /* gak */ 526 for (i = 0; i < IPREASS_NHASH; i++) { 527 if (ipq[i].prev != &ipq[i]) { 528 ip_freef(ipq[i].prev); 529 break; 530 } 531 } 532 } else 533 ip_freef(ipq[sum].prev); 534 } 535found: 536 /* 537 * Adjust ip_len to not reflect header, 538 * set ip_mff if more fragments are expected, 539 * convert offset of this to bytes. 540 */ 541 ip->ip_len -= hlen; 542 ((struct ipasfrag *)ip)->ipf_mff &= ~1; 543 if (ip->ip_off & IP_MF) 544 ((struct ipasfrag *)ip)->ipf_mff |= 1; 545 ip->ip_off <<= 3; 546 547 /* 548 * If datagram marked as having more fragments 549 * or if this is not the first fragment, 550 * attempt reassembly; if it succeeds, proceed. 551 */ 552 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) { 553 ipstat.ips_fragments++; 554 ip = ip_reass((struct ipasfrag *)ip, fp, &ipq[sum]); 555 if (ip == 0) 556 return; 557 /* Get the length of the reassembled packets header */ 558 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 559 ipstat.ips_reassembled++; 560 m = dtom(ip); 561#ifdef IPDIVERT 562 if (frag_divert_port) { 563 ip->ip_len += hlen; 564 HTONS(ip->ip_len); 565 HTONS(ip->ip_off); 566 HTONS(ip->ip_id); 567 ip->ip_sum = 0; 568 ip->ip_sum = in_cksum_hdr(ip); 569 NTOHS(ip->ip_id); 570 NTOHS(ip->ip_off); 571 NTOHS(ip->ip_len); 572 ip->ip_len -= hlen; 573 } 574#endif 575 } else 576 if (fp) 577 ip_freef(fp); 578 } else 579 ip->ip_len -= hlen; 580 581#ifdef IPDIVERT 582 /* 583 * Divert reassembled packets to the divert protocol if required 584 */ 585 if (frag_divert_port) { 586 ipstat.ips_delivered++; 587 ip_divert_port = frag_divert_port; 588 frag_divert_port = 0; 589 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, hlen); 590 return; 591 } 592 593 /* Don't let packets divert themselves */ 594 if (ip->ip_p == IPPROTO_DIVERT) { 595 ipstat.ips_noproto++; 596 goto bad; 597 } 598 599#endif 600 601 /* 602 * Switch out to protocol's input routine. 603 */ 604 ipstat.ips_delivered++; 605 (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); 606 return; 607bad: 608 m_freem(m); 609} 610 611/* 612 * IP software interrupt routine - to go away sometime soon 613 */ 614static void 615ipintr(void) 616{ 617 int s; 618 struct mbuf *m; 619 620 while(1) { 621 s = splimp(); 622 IF_DEQUEUE(&ipintrq, m); 623 splx(s); 624 if (m == 0) 625 return; 626 ip_input(m); 627 } 628} 629 630NETISR_SET(NETISR_IP, ipintr); 631 632/* 633 * Take incoming datagram fragment and try to 634 * reassemble it into whole datagram. If a chain for 635 * reassembly of this datagram already exists, then it 636 * is given as fp; otherwise have to make a chain. 637 */ 638static struct ip * 639ip_reass(ip, fp, where) 640 register struct ipasfrag *ip; 641 register struct ipq *fp; 642 struct ipq *where; 643{ 644 register struct mbuf *m = dtom(ip); 645 register struct ipasfrag *q; 646 struct mbuf *t; 647 int hlen = ip->ip_hl << 2; 648 int i, next; 649 650 /* 651 * Presence of header sizes in mbufs 652 * would confuse code below. 653 */ 654 m->m_data += hlen; 655 m->m_len -= hlen; 656 657 /* 658 * If first fragment to arrive, create a reassembly queue. 659 */ 660 if (fp == 0) { 661 if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) 662 goto dropfrag; 663 fp = mtod(t, struct ipq *); 664 insque(fp, where); 665 nipq++; 666 fp->ipq_ttl = IPFRAGTTL; 667 fp->ipq_p = ip->ip_p; 668 fp->ipq_id = ip->ip_id; 669 fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; 670 fp->ipq_src = ((struct ip *)ip)->ip_src; 671 fp->ipq_dst = ((struct ip *)ip)->ip_dst; 672#ifdef IPDIVERT 673 fp->ipq_divert = 0; 674 fp->ipq_div_cookie = 0; 675#endif 676 q = (struct ipasfrag *)fp; 677 goto insert; 678 } 679 680 /* 681 * Find a segment which begins after this one does. 682 */ 683 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) 684 if (q->ip_off > ip->ip_off) 685 break; 686 687 /* 688 * If there is a preceding segment, it may provide some of 689 * our data already. If so, drop the data from the incoming 690 * segment. If it provides all of our data, drop us. 691 */ 692 if (q->ipf_prev != (struct ipasfrag *)fp) { 693 i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; 694 if (i > 0) { 695 if (i >= ip->ip_len) 696 goto dropfrag; 697 m_adj(dtom(ip), i); 698 ip->ip_off += i; 699 ip->ip_len -= i; 700 } 701 } 702 703 /* 704 * While we overlap succeeding segments trim them or, 705 * if they are completely covered, dequeue them. 706 */ 707 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { 708 struct mbuf *m0; 709 710 i = (ip->ip_off + ip->ip_len) - q->ip_off; 711 if (i < q->ip_len) { 712 q->ip_len -= i; 713 q->ip_off += i; 714 m_adj(dtom(q), i); 715 break; 716 } 717 m0 = dtom(q); 718 q = q->ipf_next; 719 ip_deq(q->ipf_prev); 720 m_freem(m0); 721 } 722 723insert: 724 725#ifdef IPDIVERT 726 /* 727 * Any fragment diverting causes the whole packet to divert 728 */ 729 if (frag_divert_port != 0) { 730 fp->ipq_divert = frag_divert_port; 731 fp->ipq_div_cookie = ip_divert_cookie; 732 } 733 frag_divert_port = 0; 734 ip_divert_cookie = 0; 735#endif 736 737 /* 738 * Stick new segment in its place; 739 * check for complete reassembly. 740 */ 741 ip_enq(ip, q->ipf_prev); 742 next = 0; 743 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { 744 if (q->ip_off != next) 745 return (0); 746 next += q->ip_len; 747 } 748 if (q->ipf_prev->ipf_mff & 1) 749 return (0); 750 751 /* 752 * Reassembly is complete. Make sure the packet is a sane size. 753 */ 754 if (next + (IP_VHL_HL(((struct ip *)fp->ipq_next)->ip_vhl) << 2) 755 > IP_MAXPACKET) { 756 ipstat.ips_toolong++; 757 ip_freef(fp); 758 return (0); 759 } 760 761 /* 762 * Concatenate fragments. 763 */ 764 q = fp->ipq_next; 765 m = dtom(q); 766 t = m->m_next; 767 m->m_next = 0; 768 m_cat(m, t); 769 q = q->ipf_next; 770 while (q != (struct ipasfrag *)fp) { 771 t = dtom(q); 772 q = q->ipf_next; 773 m_cat(m, t); 774 } 775 776#ifdef IPDIVERT 777 /* 778 * extract divert port for packet, if any 779 */ 780 frag_divert_port = fp->ipq_divert; 781 ip_divert_cookie = fp->ipq_div_cookie; 782#endif 783 784 /* 785 * Create header for new ip packet by 786 * modifying header of first packet; 787 * dequeue and discard fragment reassembly header. 788 * Make header visible. 789 */ 790 ip = fp->ipq_next; 791 ip->ip_len = next; 792 ip->ipf_mff &= ~1; 793 ((struct ip *)ip)->ip_src = fp->ipq_src; 794 ((struct ip *)ip)->ip_dst = fp->ipq_dst; 795 remque(fp); 796 nipq--; 797 (void) m_free(dtom(fp)); 798 m = dtom(ip); 799 m->m_len += (ip->ip_hl << 2); 800 m->m_data -= (ip->ip_hl << 2); 801 /* some debugging cruft by sklower, below, will go away soon */ 802 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 803 register int plen = 0; 804 for (t = m; m; m = m->m_next) 805 plen += m->m_len; 806 t->m_pkthdr.len = plen; 807 } 808 return ((struct ip *)ip); 809 810dropfrag: 811 ipstat.ips_fragdropped++; 812 m_freem(m); 813 return (0); 814} 815 816/* 817 * Free a fragment reassembly header and all 818 * associated datagrams. 819 */ 820static void 821ip_freef(fp) 822 struct ipq *fp; 823{ 824 register struct ipasfrag *q, *p; 825 826 for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { 827 p = q->ipf_next; 828 ip_deq(q); 829 m_freem(dtom(q)); 830 } 831 remque(fp); 832 (void) m_free(dtom(fp)); 833 nipq--; 834} 835 836/* 837 * Put an ip fragment on a reassembly chain. 838 * Like insque, but pointers in middle of structure. 839 */ 840static void 841ip_enq(p, prev) 842 register struct ipasfrag *p, *prev; 843{ 844 845 p->ipf_prev = prev; 846 p->ipf_next = prev->ipf_next; 847 prev->ipf_next->ipf_prev = p; 848 prev->ipf_next = p; 849} 850 851/* 852 * To ip_enq as remque is to insque. 853 */ 854static void 855ip_deq(p) 856 register struct ipasfrag *p; 857{ 858 859 p->ipf_prev->ipf_next = p->ipf_next; 860 p->ipf_next->ipf_prev = p->ipf_prev; 861} 862 863/* 864 * IP timer processing; 865 * if a timer expires on a reassembly 866 * queue, discard it. 867 */ 868void 869ip_slowtimo() 870{ 871 register struct ipq *fp; 872 int s = splnet(); 873 int i; 874 875 for (i = 0; i < IPREASS_NHASH; i++) { 876 fp = ipq[i].next; 877 if (fp == 0) 878 continue; 879 while (fp != &ipq[i]) { 880 --fp->ipq_ttl; 881 fp = fp->next; 882 if (fp->prev->ipq_ttl == 0) { 883 ipstat.ips_fragtimeout++; 884 ip_freef(fp->prev); 885 } 886 } 887 } 888 ipflow_slowtimo(); 889 splx(s); 890} 891 892/* 893 * Drain off all datagram fragments. 894 */ 895void 896ip_drain() 897{ 898 int i; 899 900 for (i = 0; i < IPREASS_NHASH; i++) { 901 while (ipq[i].next != &ipq[i]) { 902 ipstat.ips_fragdropped++; 903 ip_freef(ipq[i].next); 904 } 905 } 906 in_rtqdrain(); 907} 908 909/* 910 * Do option processing on a datagram, 911 * possibly discarding it if bad options are encountered, 912 * or forwarding it if source-routed. 913 * Returns 1 if packet has been forwarded/freed, 914 * 0 if the packet should be processed further. 915 */ 916static int 917ip_dooptions(m) 918 struct mbuf *m; 919{ 920 register struct ip *ip = mtod(m, struct ip *); 921 register u_char *cp; 922 register struct ip_timestamp *ipt; 923 register struct in_ifaddr *ia; 924 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 925 struct in_addr *sin, dst; 926 n_time ntime; 927 928 dst = ip->ip_dst; 929 cp = (u_char *)(ip + 1); 930 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 931 for (; cnt > 0; cnt -= optlen, cp += optlen) { 932 opt = cp[IPOPT_OPTVAL]; 933 if (opt == IPOPT_EOL) 934 break; 935 if (opt == IPOPT_NOP) 936 optlen = 1; 937 else { 938 optlen = cp[IPOPT_OLEN]; 939 if (optlen <= 0 || optlen > cnt) { 940 code = &cp[IPOPT_OLEN] - (u_char *)ip; 941 goto bad; 942 } 943 } 944 switch (opt) { 945 946 default: 947 break; 948 949 /* 950 * Source routing with record. 951 * Find interface with current destination address. 952 * If none on this machine then drop if strictly routed, 953 * or do nothing if loosely routed. 954 * Record interface address and bring up next address 955 * component. If strictly routed make sure next 956 * address is on directly accessible net. 957 */ 958 case IPOPT_LSRR: 959 case IPOPT_SSRR: 960 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 961 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 962 goto bad; 963 } 964 ipaddr.sin_addr = ip->ip_dst; 965 ia = (struct in_ifaddr *) 966 ifa_ifwithaddr((struct sockaddr *)&ipaddr); 967 if (ia == 0) { 968 if (opt == IPOPT_SSRR) { 969 type = ICMP_UNREACH; 970 code = ICMP_UNREACH_SRCFAIL; 971 goto bad; 972 } 973 if (!ip_dosourceroute) 974 goto nosourcerouting; 975 /* 976 * Loose routing, and not at next destination 977 * yet; nothing to do except forward. 978 */ 979 break; 980 } 981 off--; /* 0 origin */ 982 if (off > optlen - sizeof(struct in_addr)) { 983 /* 984 * End of source route. Should be for us. 985 */ 986 if (!ip_acceptsourceroute) 987 goto nosourcerouting; 988 save_rte(cp, ip->ip_src); 989 break; 990 } 991 992 if (!ip_dosourceroute) { 993 char buf[4*sizeof "123"]; 994 995nosourcerouting: 996 strcpy(buf, inet_ntoa(ip->ip_dst)); 997 log(LOG_WARNING, 998 "attempted source route from %s to %s\n", 999 inet_ntoa(ip->ip_src), buf); 1000 type = ICMP_UNREACH; 1001 code = ICMP_UNREACH_SRCFAIL; 1002 goto bad; 1003 } 1004 1005 /* 1006 * locate outgoing interface 1007 */ 1008 (void)memcpy(&ipaddr.sin_addr, cp + off, 1009 sizeof(ipaddr.sin_addr)); 1010 1011 if (opt == IPOPT_SSRR) { 1012#define INA struct in_ifaddr * 1013#define SA struct sockaddr * 1014 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) 1015 ia = (INA)ifa_ifwithnet((SA)&ipaddr); 1016 } else 1017 ia = ip_rtaddr(ipaddr.sin_addr); 1018 if (ia == 0) { 1019 type = ICMP_UNREACH; 1020 code = ICMP_UNREACH_SRCFAIL; 1021 goto bad; 1022 } 1023 ip->ip_dst = ipaddr.sin_addr; 1024 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 1025 sizeof(struct in_addr)); 1026 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1027 /* 1028 * Let ip_intr's mcast routing check handle mcast pkts 1029 */ 1030 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr)); 1031 break; 1032 1033 case IPOPT_RR: 1034 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1035 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1036 goto bad; 1037 } 1038 /* 1039 * If no space remains, ignore. 1040 */ 1041 off--; /* 0 origin */ 1042 if (off > optlen - sizeof(struct in_addr)) 1043 break; 1044 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst, 1045 sizeof(ipaddr.sin_addr)); 1046 /* 1047 * locate outgoing interface; if we're the destination, 1048 * use the incoming interface (should be same). 1049 */ 1050 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 1051 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 1052 type = ICMP_UNREACH; 1053 code = ICMP_UNREACH_HOST; 1054 goto bad; 1055 } 1056 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr), 1057 sizeof(struct in_addr)); 1058 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1059 break; 1060 1061 case IPOPT_TS: 1062 code = cp - (u_char *)ip; 1063 ipt = (struct ip_timestamp *)cp; 1064 if (ipt->ipt_len < 5) 1065 goto bad; 1066 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { 1067 if (++ipt->ipt_oflw == 0) 1068 goto bad; 1069 break; 1070 } 1071 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 1072 switch (ipt->ipt_flg) { 1073 1074 case IPOPT_TS_TSONLY: 1075 break; 1076 1077 case IPOPT_TS_TSANDADDR: 1078 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1079 sizeof(struct in_addr) > ipt->ipt_len) 1080 goto bad; 1081 ipaddr.sin_addr = dst; 1082 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 1083 m->m_pkthdr.rcvif); 1084 if (ia == 0) 1085 continue; 1086 (void)memcpy(sin, &IA_SIN(ia)->sin_addr, 1087 sizeof(struct in_addr)); 1088 ipt->ipt_ptr += sizeof(struct in_addr); 1089 break; 1090 1091 case IPOPT_TS_PRESPEC: 1092 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1093 sizeof(struct in_addr) > ipt->ipt_len) 1094 goto bad; 1095 (void)memcpy(&ipaddr.sin_addr, sin, 1096 sizeof(struct in_addr)); 1097 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 1098 continue; 1099 ipt->ipt_ptr += sizeof(struct in_addr); 1100 break; 1101 1102 default: 1103 goto bad; 1104 } 1105 ntime = iptime(); 1106 (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime, 1107 sizeof(n_time)); 1108 ipt->ipt_ptr += sizeof(n_time); 1109 } 1110 } 1111 if (forward && ipforwarding) { 1112 ip_forward(m, 1); 1113 return (1); 1114 } 1115 return (0); 1116bad: 1117 ip->ip_len -= IP_VHL_HL(ip->ip_vhl) << 2; /* XXX icmp_error adds in hdr length */ 1118 icmp_error(m, type, code, 0, 0); 1119 ipstat.ips_badoptions++; 1120 return (1); 1121} 1122 1123/* 1124 * Given address of next destination (final or next hop), 1125 * return internet address info of interface to be used to get there. 1126 */ 1127static struct in_ifaddr * 1128ip_rtaddr(dst) 1129 struct in_addr dst; 1130{ 1131 register struct sockaddr_in *sin; 1132 1133 sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; 1134 1135 if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { 1136 if (ipforward_rt.ro_rt) { 1137 RTFREE(ipforward_rt.ro_rt); 1138 ipforward_rt.ro_rt = 0; 1139 } 1140 sin->sin_family = AF_INET; 1141 sin->sin_len = sizeof(*sin); 1142 sin->sin_addr = dst; 1143 1144 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1145 } 1146 if (ipforward_rt.ro_rt == 0) 1147 return ((struct in_ifaddr *)0); 1148 return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); 1149} 1150 1151/* 1152 * Save incoming source route for use in replies, 1153 * to be picked up later by ip_srcroute if the receiver is interested. 1154 */ 1155void 1156save_rte(option, dst) 1157 u_char *option; 1158 struct in_addr dst; 1159{ 1160 unsigned olen; 1161 1162 olen = option[IPOPT_OLEN]; 1163#ifdef DIAGNOSTIC 1164 if (ipprintfs) 1165 printf("save_rte: olen %d\n", olen); 1166#endif 1167 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1168 return; 1169 bcopy(option, ip_srcrt.srcopt, olen); 1170 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1171 ip_srcrt.dst = dst; 1172} 1173 1174/* 1175 * Retrieve incoming source route for use in replies, 1176 * in the same form used by setsockopt. 1177 * The first hop is placed before the options, will be removed later. 1178 */ 1179struct mbuf * 1180ip_srcroute() 1181{ 1182 register struct in_addr *p, *q; 1183 register struct mbuf *m; 1184 1185 if (ip_nhops == 0) 1186 return ((struct mbuf *)0); 1187 m = m_get(M_DONTWAIT, MT_SOOPTS); 1188 if (m == 0) 1189 return ((struct mbuf *)0); 1190 1191#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1192 1193 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1194 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1195 OPTSIZ; 1196#ifdef DIAGNOSTIC 1197 if (ipprintfs) 1198 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1199#endif 1200 1201 /* 1202 * First save first hop for return route 1203 */ 1204 p = &ip_srcrt.route[ip_nhops - 1]; 1205 *(mtod(m, struct in_addr *)) = *p--; 1206#ifdef DIAGNOSTIC 1207 if (ipprintfs) 1208 printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); 1209#endif 1210 1211 /* 1212 * Copy option fields and padding (nop) to mbuf. 1213 */ 1214 ip_srcrt.nop = IPOPT_NOP; 1215 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1216 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), 1217 &ip_srcrt.nop, OPTSIZ); 1218 q = (struct in_addr *)(mtod(m, caddr_t) + 1219 sizeof(struct in_addr) + OPTSIZ); 1220#undef OPTSIZ 1221 /* 1222 * Record return path as an IP source route, 1223 * reversing the path (pointers are now aligned). 1224 */ 1225 while (p >= ip_srcrt.route) { 1226#ifdef DIAGNOSTIC 1227 if (ipprintfs) 1228 printf(" %lx", ntohl(q->s_addr)); 1229#endif 1230 *q++ = *p--; 1231 } 1232 /* 1233 * Last hop goes to final destination. 1234 */ 1235 *q = ip_srcrt.dst; 1236#ifdef DIAGNOSTIC 1237 if (ipprintfs) 1238 printf(" %lx\n", ntohl(q->s_addr)); 1239#endif 1240 return (m); 1241} 1242 1243/* 1244 * Strip out IP options, at higher 1245 * level protocol in the kernel. 1246 * Second argument is buffer to which options 1247 * will be moved, and return value is their length. 1248 * XXX should be deleted; last arg currently ignored. 1249 */ 1250void 1251ip_stripoptions(m, mopt) 1252 register struct mbuf *m; 1253 struct mbuf *mopt; 1254{ 1255 register int i; 1256 struct ip *ip = mtod(m, struct ip *); 1257 register caddr_t opts; 1258 int olen; 1259 1260 olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 1261 opts = (caddr_t)(ip + 1); 1262 i = m->m_len - (sizeof (struct ip) + olen); 1263 bcopy(opts + olen, opts, (unsigned)i); 1264 m->m_len -= olen; 1265 if (m->m_flags & M_PKTHDR) 1266 m->m_pkthdr.len -= olen; 1267 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2); 1268} 1269 1270u_char inetctlerrmap[PRC_NCMDS] = { 1271 0, 0, 0, 0, 1272 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1273 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1274 EMSGSIZE, EHOSTUNREACH, 0, 0, 1275 0, 0, 0, 0, 1276 ENOPROTOOPT 1277}; 1278 1279/* 1280 * Forward a packet. If some error occurs return the sender 1281 * an icmp packet. Note we can't always generate a meaningful 1282 * icmp message because icmp doesn't have a large enough repertoire 1283 * of codes and types. 1284 * 1285 * If not forwarding, just drop the packet. This could be confusing 1286 * if ipforwarding was zero but some routing protocol was advancing 1287 * us as a gateway to somewhere. However, we must let the routing 1288 * protocol deal with that. 1289 * 1290 * The srcrt parameter indicates whether the packet is being forwarded 1291 * via a source route. 1292 */ 1293static void 1294ip_forward(m, srcrt) 1295 struct mbuf *m; 1296 int srcrt; 1297{ 1298 register struct ip *ip = mtod(m, struct ip *); 1299 register struct sockaddr_in *sin; 1300 register struct rtentry *rt; 1301 int error, type = 0, code = 0; 1302 struct mbuf *mcopy; 1303 n_long dest; 1304 struct ifnet *destifp; 1305 1306 dest = 0; 1307#ifdef DIAGNOSTIC 1308 if (ipprintfs) 1309 printf("forward: src %lx dst %lx ttl %x\n", 1310 ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl); 1311#endif 1312 1313 1314 if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { 1315 ipstat.ips_cantforward++; 1316 m_freem(m); 1317 return; 1318 } 1319 HTONS(ip->ip_id); 1320 if (ip->ip_ttl <= IPTTLDEC) { 1321 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1322 return; 1323 } 1324 ip->ip_ttl -= IPTTLDEC; 1325 1326 sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; 1327 if ((rt = ipforward_rt.ro_rt) == 0 || 1328 ip->ip_dst.s_addr != sin->sin_addr.s_addr) { 1329 if (ipforward_rt.ro_rt) { 1330 RTFREE(ipforward_rt.ro_rt); 1331 ipforward_rt.ro_rt = 0; 1332 } 1333 sin->sin_family = AF_INET; 1334 sin->sin_len = sizeof(*sin); 1335 sin->sin_addr = ip->ip_dst; 1336 1337 rtalloc_ign(&ipforward_rt, RTF_PRCLONING); 1338 if (ipforward_rt.ro_rt == 0) { 1339 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1340 return; 1341 } 1342 rt = ipforward_rt.ro_rt; 1343 } 1344 1345 /* 1346 * Save at most 64 bytes of the packet in case 1347 * we need to generate an ICMP message to the src. 1348 */ 1349 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); 1350 1351 /* 1352 * If forwarding packet using same interface that it came in on, 1353 * perhaps should send a redirect to sender to shortcut a hop. 1354 * Only send redirect if source is sending directly to us, 1355 * and if packet was not source routed (or has any options). 1356 * Also, don't send redirect if forwarding using a default route 1357 * or a route modified by a redirect. 1358 */ 1359#define satosin(sa) ((struct sockaddr_in *)(sa)) 1360 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1361 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1362 satosin(rt_key(rt))->sin_addr.s_addr != 0 && 1363 ipsendredirects && !srcrt) { 1364#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) 1365 u_long src = ntohl(ip->ip_src.s_addr); 1366 1367 if (RTA(rt) && 1368 (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { 1369 if (rt->rt_flags & RTF_GATEWAY) 1370 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1371 else 1372 dest = ip->ip_dst.s_addr; 1373 /* Router requirements says to only send host redirects */ 1374 type = ICMP_REDIRECT; 1375 code = ICMP_REDIRECT_HOST; 1376#ifdef DIAGNOSTIC 1377 if (ipprintfs) 1378 printf("redirect (%d) to %lx\n", code, (u_long)dest); 1379#endif 1380 } 1381 } 1382 1383 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1384 IP_FORWARDING, 0); 1385 if (error) 1386 ipstat.ips_cantforward++; 1387 else { 1388 ipstat.ips_forward++; 1389 if (type) 1390 ipstat.ips_redirectsent++; 1391 else { 1392 if (mcopy) { 1393 ipflow_create(&ipforward_rt, mcopy); 1394 m_freem(mcopy); 1395 } 1396 return; 1397 } 1398 } 1399 if (mcopy == NULL) 1400 return; 1401 destifp = NULL; 1402 1403 switch (error) { 1404 1405 case 0: /* forwarded, but need redirect */ 1406 /* type, code set above */ 1407 break; 1408 1409 case ENETUNREACH: /* shouldn't happen, checked above */ 1410 case EHOSTUNREACH: 1411 case ENETDOWN: 1412 case EHOSTDOWN: 1413 default: 1414 type = ICMP_UNREACH; 1415 code = ICMP_UNREACH_HOST; 1416 break; 1417 1418 case EMSGSIZE: 1419 type = ICMP_UNREACH; 1420 code = ICMP_UNREACH_NEEDFRAG; 1421 if (ipforward_rt.ro_rt) 1422 destifp = ipforward_rt.ro_rt->rt_ifp; 1423 ipstat.ips_cantfrag++; 1424 break; 1425 1426 case ENOBUFS: 1427 type = ICMP_SOURCEQUENCH; 1428 code = 0; 1429 break; 1430 } 1431 icmp_error(mcopy, type, code, dest, destifp); 1432} 1433 1434void 1435ip_savecontrol(inp, mp, ip, m) 1436 register struct inpcb *inp; 1437 register struct mbuf **mp; 1438 register struct ip *ip; 1439 register struct mbuf *m; 1440{ 1441 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1442 struct timeval tv; 1443 1444 microtime(&tv); 1445 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1446 SCM_TIMESTAMP, SOL_SOCKET); 1447 if (*mp) 1448 mp = &(*mp)->m_next; 1449 } 1450 if (inp->inp_flags & INP_RECVDSTADDR) { 1451 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1452 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1453 if (*mp) 1454 mp = &(*mp)->m_next; 1455 } 1456#ifdef notyet 1457 /* XXX 1458 * Moving these out of udp_input() made them even more broken 1459 * than they already were. 1460 */ 1461 /* options were tossed already */ 1462 if (inp->inp_flags & INP_RECVOPTS) { 1463 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1464 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1465 if (*mp) 1466 mp = &(*mp)->m_next; 1467 } 1468 /* ip_srcroute doesn't do what we want here, need to fix */ 1469 if (inp->inp_flags & INP_RECVRETOPTS) { 1470 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1471 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1472 if (*mp) 1473 mp = &(*mp)->m_next; 1474 } 1475#endif 1476 if (inp->inp_flags & INP_RECVIF) { 1477 struct ifnet *ifp; 1478 struct sdlbuf { 1479 struct sockaddr_dl sdl; 1480 u_char pad[32]; 1481 } sdlbuf; 1482 struct sockaddr_dl *sdp; 1483 struct sockaddr_dl *sdl2 = &sdlbuf.sdl; 1484 1485 if (((ifp = m->m_pkthdr.rcvif)) 1486 && ( ifp->if_index && (ifp->if_index <= if_index))) { 1487 sdp = (struct sockaddr_dl *)(ifnet_addrs 1488 [ifp->if_index - 1]->ifa_addr); 1489 /* 1490 * Change our mind and don't try copy. 1491 */ 1492 if ((sdp->sdl_family != AF_LINK) 1493 || (sdp->sdl_len > sizeof(sdlbuf))) { 1494 goto makedummy; 1495 } 1496 bcopy(sdp, sdl2, sdp->sdl_len); 1497 } else { 1498makedummy: 1499 sdl2->sdl_len 1500 = offsetof(struct sockaddr_dl, sdl_data[0]); 1501 sdl2->sdl_family = AF_LINK; 1502 sdl2->sdl_index = 0; 1503 sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0; 1504 } 1505 *mp = sbcreatecontrol((caddr_t) sdl2, sdl2->sdl_len, 1506 IP_RECVIF, IPPROTO_IP); 1507 if (*mp) 1508 mp = &(*mp)->m_next; 1509 } 1510} 1511 1512int 1513ip_rsvp_init(struct socket *so) 1514{ 1515 if (so->so_type != SOCK_RAW || 1516 so->so_proto->pr_protocol != IPPROTO_RSVP) 1517 return EOPNOTSUPP; 1518 1519 if (ip_rsvpd != NULL) 1520 return EADDRINUSE; 1521 1522 ip_rsvpd = so; 1523 /* 1524 * This may seem silly, but we need to be sure we don't over-increment 1525 * the RSVP counter, in case something slips up. 1526 */ 1527 if (!ip_rsvp_on) { 1528 ip_rsvp_on = 1; 1529 rsvp_on++; 1530 } 1531 1532 return 0; 1533} 1534 1535int 1536ip_rsvp_done(void) 1537{ 1538 ip_rsvpd = NULL; 1539 /* 1540 * This may seem silly, but we need to be sure we don't over-decrement 1541 * the RSVP counter, in case something slips up. 1542 */ 1543 if (ip_rsvp_on) { 1544 ip_rsvp_on = 0; 1545 rsvp_on--; 1546 } 1547 return 0; 1548} 1549