1 /* 2 * Copyright (c) 1982, 1986, 1988, 1990, 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_output.c 8.3 (Berkeley) 1/21/94 34 * $FreeBSD$ 35 */ 36 37 #define _IP_VHL 38 39 #include "opt_ipfw.h" 40 #include "opt_ipdn.h" 41 #include "opt_ipdivert.h" 42 #include "opt_ipfilter.h" 43 #include "opt_ipsec.h" 44 #include "opt_pfil_hooks.h" 45 #include "opt_random_ip_id.h" 46 47 #include <sys/param.h> 48 #include <sys/systm.h> 49 #include <sys/kernel.h> 50 #include <sys/malloc.h> 51 #include <sys/mbuf.h> 52 #include <sys/protosw.h> 53 #include <sys/socket.h> 54 #include <sys/socketvar.h> 55 56 #include <net/if.h> 57 #include <net/route.h> 58 59 #include <netinet/in.h> 60 #include <netinet/in_systm.h> 61 #include <netinet/ip.h> 62 #include <netinet/in_pcb.h> 63 #include <netinet/in_var.h> 64 #include <netinet/ip_var.h> 65 66 #include "faith.h" 67 68 #include <machine/in_cksum.h> 69 70 static MALLOC_DEFINE(M_IPMOPTS, "ip_moptions", "internet multicast options"); 71 72 #ifdef IPSEC 73 #include <netinet6/ipsec.h> 74 #include <netkey/key.h> 75 #ifdef IPSEC_DEBUG 76 #include <netkey/key_debug.h> 77 #else 78 #define KEYDEBUG(lev,arg) 79 #endif 80 #endif /*IPSEC*/ 81 82 #include <netinet/ip_fw.h> 83 84 #ifdef DUMMYNET 85 #include <netinet/ip_dummynet.h> 86 #endif 87 88 #ifdef IPFIREWALL_FORWARD_DEBUG 89 #define print_ip(a) printf("%ld.%ld.%ld.%ld",(ntohl(a.s_addr)>>24)&0xFF,\ 90 (ntohl(a.s_addr)>>16)&0xFF,\ 91 (ntohl(a.s_addr)>>8)&0xFF,\ 92 (ntohl(a.s_addr))&0xFF); 93 #endif 94 95 u_short ip_id; 96 97 static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 98 static struct ifnet *ip_multicast_if __P((struct in_addr *, int *)); 99 static void ip_mloopback 100 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *, int)); 101 static int ip_getmoptions 102 __P((struct sockopt *, struct ip_moptions *)); 103 static int ip_pcbopts __P((int, struct mbuf **, struct mbuf *)); 104 static int ip_setmoptions 105 __P((struct sockopt *, struct ip_moptions **)); 106 107 int ip_optcopy __P((struct ip *, struct ip *)); 108 109 110 extern struct protosw inetsw[]; 111 112 /* 113 * IP output. The packet in mbuf chain m contains a skeletal IP 114 * header (with len, off, ttl, proto, tos, src, dst). 115 * The mbuf chain containing the packet will be freed. 116 * The mbuf opt, if present, will not be freed. 117 */ 118 int 119 ip_output(m0, opt, ro, flags, imo) 120 struct mbuf *m0; 121 struct mbuf *opt; 122 struct route *ro; 123 int flags; 124 struct ip_moptions *imo; 125 { 126 struct ip *ip, *mhip; 127 struct ifnet *ifp; 128 struct mbuf *m = m0; 129 int hlen = sizeof (struct ip); 130 int len, off, error = 0; 131 struct sockaddr_in *dst; 132 struct in_ifaddr *ia; 133 int isbroadcast, sw_csum; 134 #ifdef IPSEC 135 struct route iproute; 136 struct socket *so = NULL; 137 struct secpolicy *sp = NULL; 138 #endif 139 u_int16_t divert_cookie; /* firewall cookie */ 140 #ifdef PFIL_HOOKS 141 struct packet_filter_hook *pfh; 142 struct mbuf *m1; 143 int rv; 144 #endif /* PFIL_HOOKS */ 145 #ifdef IPFIREWALL_FORWARD 146 int fwd_rewrite_src = 0; 147 #endif 148 struct ip_fw_chain *rule = NULL; 149 150 #ifdef IPDIVERT 151 /* Get and reset firewall cookie */ 152 divert_cookie = ip_divert_cookie; 153 ip_divert_cookie = 0; 154 #else 155 divert_cookie = 0; 156 #endif 157 158 #if defined(IPFIREWALL) && defined(DUMMYNET) 159 /* 160 * dummynet packet are prepended a vestigial mbuf with 161 * m_type = MT_DUMMYNET and m_data pointing to the matching 162 * rule. 163 */ 164 if (m->m_type == MT_DUMMYNET) { 165 /* 166 * the packet was already tagged, so part of the 167 * processing was already done, and we need to go down. 168 * Get parameters from the header. 169 */ 170 rule = (struct ip_fw_chain *)(m->m_data) ; 171 opt = NULL ; 172 ro = & ( ((struct dn_pkt *)m)->ro ) ; 173 imo = NULL ; 174 dst = ((struct dn_pkt *)m)->dn_dst ; 175 ifp = ((struct dn_pkt *)m)->ifp ; 176 flags = ((struct dn_pkt *)m)->flags ; 177 178 m0 = m = m->m_next ; 179 #ifdef IPSEC 180 so = ipsec_getsocket(m); 181 (void)ipsec_setsocket(m, NULL); 182 #endif 183 ip = mtod(m, struct ip *); 184 hlen = IP_VHL_HL(ip->ip_vhl) << 2 ; 185 ia = (struct in_ifaddr *)ro->ro_rt->rt_ifa; 186 goto sendit; 187 } else 188 rule = NULL ; 189 #endif 190 #ifdef IPSEC 191 so = ipsec_getsocket(m); 192 (void)ipsec_setsocket(m, NULL); 193 #endif 194 195 #ifdef DIAGNOSTIC 196 if ((m->m_flags & M_PKTHDR) == 0) 197 panic("ip_output no HDR"); 198 if (!ro) 199 panic("ip_output no route, proto = %d", 200 mtod(m, struct ip *)->ip_p); 201 #endif 202 if (opt) { 203 m = ip_insertoptions(m, opt, &len); 204 hlen = len; 205 } 206 ip = mtod(m, struct ip *); 207 /* 208 * Fill in IP header. 209 */ 210 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 211 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 212 ip->ip_off &= IP_DF; 213 #ifdef RANDOM_IP_ID 214 ip->ip_id = ip_randomid(); 215 #else 216 ip->ip_id = htons(ip_id++); 217 #endif 218 ipstat.ips_localout++; 219 } else { 220 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 221 } 222 223 dst = (struct sockaddr_in *)&ro->ro_dst; 224 /* 225 * If there is a cached route, 226 * check that it is to the same destination 227 * and is still up. If not, free it and try again. 228 */ 229 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 230 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 231 RTFREE(ro->ro_rt); 232 ro->ro_rt = (struct rtentry *)0; 233 } 234 if (ro->ro_rt == 0) { 235 dst->sin_family = AF_INET; 236 dst->sin_len = sizeof(*dst); 237 dst->sin_addr = ip->ip_dst; 238 } 239 /* 240 * If routing to interface only, 241 * short circuit routing lookup. 242 */ 243 #define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 244 #define sintosa(sin) ((struct sockaddr *)(sin)) 245 if (flags & IP_ROUTETOIF) { 246 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 247 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 248 ipstat.ips_noroute++; 249 error = ENETUNREACH; 250 goto bad; 251 } 252 ifp = ia->ia_ifp; 253 ip->ip_ttl = 1; 254 isbroadcast = in_broadcast(dst->sin_addr, ifp); 255 } else { 256 /* 257 * If this is the case, we probably don't want to allocate 258 * a protocol-cloned route since we didn't get one from the 259 * ULP. This lets TCP do its thing, while not burdening 260 * forwarding or ICMP with the overhead of cloning a route. 261 * Of course, we still want to do any cloning requested by 262 * the link layer, as this is probably required in all cases 263 * for correct operation (as it is for ARP). 264 */ 265 if (ro->ro_rt == 0) 266 rtalloc_ign(ro, RTF_PRCLONING); 267 if (ro->ro_rt == 0) { 268 ipstat.ips_noroute++; 269 error = EHOSTUNREACH; 270 goto bad; 271 } 272 ia = ifatoia(ro->ro_rt->rt_ifa); 273 ifp = ro->ro_rt->rt_ifp; 274 ro->ro_rt->rt_use++; 275 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 276 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 277 if (ro->ro_rt->rt_flags & RTF_HOST) 278 isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST); 279 else 280 isbroadcast = in_broadcast(dst->sin_addr, ifp); 281 } 282 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 283 struct in_multi *inm; 284 285 m->m_flags |= M_MCAST; 286 /* 287 * IP destination address is multicast. Make sure "dst" 288 * still points to the address in "ro". (It may have been 289 * changed to point to a gateway address, above.) 290 */ 291 dst = (struct sockaddr_in *)&ro->ro_dst; 292 /* 293 * See if the caller provided any multicast options 294 */ 295 if (imo != NULL) { 296 ip->ip_ttl = imo->imo_multicast_ttl; 297 if (imo->imo_multicast_ifp != NULL) 298 ifp = imo->imo_multicast_ifp; 299 if (imo->imo_multicast_vif != -1) 300 ip->ip_src.s_addr = 301 ip_mcast_src(imo->imo_multicast_vif); 302 } else 303 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 304 /* 305 * Confirm that the outgoing interface supports multicast. 306 */ 307 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 308 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 309 ipstat.ips_noroute++; 310 error = ENETUNREACH; 311 goto bad; 312 } 313 } 314 /* 315 * If source address not specified yet, use address 316 * of outgoing interface. 317 */ 318 if (ip->ip_src.s_addr == INADDR_ANY) { 319 register struct in_ifaddr *ia1; 320 321 TAILQ_FOREACH(ia1, &in_ifaddrhead, ia_link) 322 if (ia1->ia_ifp == ifp) { 323 ip->ip_src = IA_SIN(ia1)->sin_addr; 324 break; 325 } 326 } 327 328 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 329 if (inm != NULL && 330 (imo == NULL || imo->imo_multicast_loop)) { 331 /* 332 * If we belong to the destination multicast group 333 * on the outgoing interface, and the caller did not 334 * forbid loopback, loop back a copy. 335 */ 336 ip_mloopback(ifp, m, dst, hlen); 337 } 338 else { 339 /* 340 * If we are acting as a multicast router, perform 341 * multicast forwarding as if the packet had just 342 * arrived on the interface to which we are about 343 * to send. The multicast forwarding function 344 * recursively calls this function, using the 345 * IP_FORWARDING flag to prevent infinite recursion. 346 * 347 * Multicasts that are looped back by ip_mloopback(), 348 * above, will be forwarded by the ip_input() routine, 349 * if necessary. 350 */ 351 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 352 /* 353 * Check if rsvp daemon is running. If not, don't 354 * set ip_moptions. This ensures that the packet 355 * is multicast and not just sent down one link 356 * as prescribed by rsvpd. 357 */ 358 if (!rsvp_on) 359 imo = NULL; 360 if (ip_mforward(ip, ifp, m, imo) != 0) { 361 m_freem(m); 362 goto done; 363 } 364 } 365 } 366 367 /* 368 * Multicasts with a time-to-live of zero may be looped- 369 * back, above, but must not be transmitted on a network. 370 * Also, multicasts addressed to the loopback interface 371 * are not sent -- the above call to ip_mloopback() will 372 * loop back a copy if this host actually belongs to the 373 * destination group on the loopback interface. 374 */ 375 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 376 m_freem(m); 377 goto done; 378 } 379 380 goto sendit; 381 } 382 #ifndef notdef 383 /* 384 * If source address not specified yet, use address 385 * of outgoing interface. 386 */ 387 if (ip->ip_src.s_addr == INADDR_ANY) { 388 ip->ip_src = IA_SIN(ia)->sin_addr; 389 #ifdef IPFIREWALL_FORWARD 390 /* Keep note that we did this - if the firewall changes 391 * the next-hop, our interface may change, changing the 392 * default source IP. It's a shame so much effort happens 393 * twice. Oh well. 394 */ 395 fwd_rewrite_src++; 396 #endif /* IPFIREWALL_FORWARD */ 397 } 398 #endif /* notdef */ 399 /* 400 * Verify that we have any chance at all of being able to queue 401 * the packet or packet fragments 402 */ 403 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 404 ifp->if_snd.ifq_maxlen) { 405 error = ENOBUFS; 406 goto bad; 407 } 408 409 /* 410 * Look for broadcast address and 411 * and verify user is allowed to send 412 * such a packet. 413 */ 414 if (isbroadcast) { 415 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 416 error = EADDRNOTAVAIL; 417 goto bad; 418 } 419 if ((flags & IP_ALLOWBROADCAST) == 0) { 420 error = EACCES; 421 goto bad; 422 } 423 /* don't allow broadcast messages to be fragmented */ 424 if ((u_short)ip->ip_len > ifp->if_mtu) { 425 error = EMSGSIZE; 426 goto bad; 427 } 428 m->m_flags |= M_BCAST; 429 } else { 430 m->m_flags &= ~M_BCAST; 431 } 432 433 sendit: 434 #ifdef IPSEC 435 /* get SP for this packet */ 436 if (so == NULL) 437 sp = ipsec4_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, flags, &error); 438 else 439 sp = ipsec4_getpolicybysock(m, IPSEC_DIR_OUTBOUND, so, &error); 440 441 if (sp == NULL) { 442 ipsecstat.out_inval++; 443 goto bad; 444 } 445 446 error = 0; 447 448 /* check policy */ 449 switch (sp->policy) { 450 case IPSEC_POLICY_DISCARD: 451 /* 452 * This packet is just discarded. 453 */ 454 ipsecstat.out_polvio++; 455 goto bad; 456 457 case IPSEC_POLICY_BYPASS: 458 case IPSEC_POLICY_NONE: 459 /* no need to do IPsec. */ 460 goto skip_ipsec; 461 462 case IPSEC_POLICY_IPSEC: 463 if (sp->req == NULL) { 464 /* acquire a policy */ 465 error = key_spdacquire(sp); 466 goto bad; 467 } 468 break; 469 470 case IPSEC_POLICY_ENTRUST: 471 default: 472 printf("ip_output: Invalid policy found. %d\n", sp->policy); 473 } 474 { 475 struct ipsec_output_state state; 476 bzero(&state, sizeof(state)); 477 state.m = m; 478 if (flags & IP_ROUTETOIF) { 479 state.ro = &iproute; 480 bzero(&iproute, sizeof(iproute)); 481 } else 482 state.ro = ro; 483 state.dst = (struct sockaddr *)dst; 484 485 ip->ip_sum = 0; 486 487 /* 488 * XXX 489 * delayed checksums are not currently compatible with IPsec 490 */ 491 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 492 in_delayed_cksum(m); 493 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 494 } 495 496 HTONS(ip->ip_len); 497 HTONS(ip->ip_off); 498 499 error = ipsec4_output(&state, sp, flags); 500 501 m = state.m; 502 if (flags & IP_ROUTETOIF) { 503 /* 504 * if we have tunnel mode SA, we may need to ignore 505 * IP_ROUTETOIF. 506 */ 507 if (state.ro != &iproute || state.ro->ro_rt != NULL) { 508 flags &= ~IP_ROUTETOIF; 509 ro = state.ro; 510 } 511 } else 512 ro = state.ro; 513 dst = (struct sockaddr_in *)state.dst; 514 if (error) { 515 /* mbuf is already reclaimed in ipsec4_output. */ 516 m0 = NULL; 517 switch (error) { 518 case EHOSTUNREACH: 519 case ENETUNREACH: 520 case EMSGSIZE: 521 case ENOBUFS: 522 case ENOMEM: 523 break; 524 default: 525 printf("ip4_output (ipsec): error code %d\n", error); 526 /*fall through*/ 527 case ENOENT: 528 /* don't show these error codes to the user */ 529 error = 0; 530 break; 531 } 532 goto bad; 533 } 534 } 535 536 /* be sure to update variables that are affected by ipsec4_output() */ 537 ip = mtod(m, struct ip *); 538 #ifdef _IP_VHL 539 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 540 #else 541 hlen = ip->ip_hl << 2; 542 #endif 543 if (ro->ro_rt == NULL) { 544 if ((flags & IP_ROUTETOIF) == 0) { 545 printf("ip_output: " 546 "can't update route after IPsec processing\n"); 547 error = EHOSTUNREACH; /*XXX*/ 548 goto bad; 549 } 550 } else { 551 ia = ifatoia(ro->ro_rt->rt_ifa); 552 ifp = ro->ro_rt->rt_ifp; 553 } 554 555 /* make it flipped, again. */ 556 NTOHS(ip->ip_len); 557 NTOHS(ip->ip_off); 558 skip_ipsec: 559 #endif /*IPSEC*/ 560 561 /* 562 * IpHack's section. 563 * - Xlate: translate packet's addr/port (NAT). 564 * - Firewall: deny/allow/etc. 565 * - Wrap: fake packet's addr/port <unimpl.> 566 * - Encapsulate: put it in another IP and send out. <unimp.> 567 */ 568 #ifdef PFIL_HOOKS 569 /* 570 * Run through list of hooks for output packets. 571 */ 572 m1 = m; 573 pfh = pfil_hook_get(PFIL_OUT, &inetsw[ip_protox[IPPROTO_IP]].pr_pfh); 574 for (; pfh; pfh = TAILQ_NEXT(pfh, pfil_link)) 575 if (pfh->pfil_func) { 576 rv = pfh->pfil_func(ip, hlen, ifp, 1, &m1); 577 if (rv) { 578 error = EHOSTUNREACH; 579 goto done; 580 } 581 m = m1; 582 if (m == NULL) 583 goto done; 584 ip = mtod(m, struct ip *); 585 } 586 #endif /* PFIL_HOOKS */ 587 588 /* 589 * Check with the firewall... 590 */ 591 if (fw_enable && ip_fw_chk_ptr) { 592 struct sockaddr_in *old = dst; 593 594 off = (*ip_fw_chk_ptr)(&ip, 595 hlen, ifp, &divert_cookie, &m, &rule, &dst); 596 /* 597 * On return we must do the following: 598 * m == NULL -> drop the pkt (old interface, deprecated) 599 * (off & 0x40000) -> drop the pkt (new interface) 600 * 1<=off<= 0xffff -> DIVERT 601 * (off & 0x10000) -> send to a DUMMYNET pipe 602 * (off & 0x20000) -> TEE the packet 603 * dst != old -> IPFIREWALL_FORWARD 604 * off==0, dst==old -> accept 605 * If some of the above modules is not compiled in, then 606 * we should't have to check the corresponding condition 607 * (because the ipfw control socket should not accept 608 * unsupported rules), but better play safe and drop 609 * packets in case of doubt. 610 */ 611 if (off & IP_FW_PORT_DENY_FLAG) { /* XXX new interface-denied */ 612 if (m) 613 m_freem(m); 614 error = EACCES ; 615 goto done; 616 } 617 if (!m) { /* firewall said to reject */ 618 static int __debug=10; 619 if (__debug >0) { 620 printf("firewall returns NULL, please update!\n"); 621 __debug-- ; 622 } 623 error = EACCES; 624 goto done; 625 } 626 if (off == 0 && dst == old) /* common case */ 627 goto pass ; 628 #ifdef DUMMYNET 629 if ((off & IP_FW_PORT_DYNT_FLAG) != 0) { 630 /* 631 * pass the pkt to dummynet. Need to include 632 * pipe number, m, ifp, ro, dst because these are 633 * not recomputed in the next pass. 634 * All other parameters have been already used and 635 * so they are not needed anymore. 636 * XXX note: if the ifp or ro entry are deleted 637 * while a pkt is in dummynet, we are in trouble! 638 */ 639 error = dummynet_io(off & 0xffff, DN_TO_IP_OUT, m, 640 ifp,ro,dst,rule, flags); 641 goto done; 642 } 643 #endif 644 #ifdef IPDIVERT 645 if (off != 0 && (off & IP_FW_PORT_DYNT_FLAG) == 0) { 646 struct mbuf *clone = NULL; 647 648 /* Clone packet if we're doing a 'tee' */ 649 if ((off & IP_FW_PORT_TEE_FLAG) != 0) 650 clone = m_dup(m, M_DONTWAIT); 651 652 /* 653 * XXX 654 * delayed checksums are not currently compatible 655 * with divert sockets. 656 */ 657 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 658 in_delayed_cksum(m); 659 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 660 } 661 662 /* Restore packet header fields to original values */ 663 HTONS(ip->ip_len); 664 HTONS(ip->ip_off); 665 666 /* Deliver packet to divert input routine */ 667 ip_divert_cookie = divert_cookie; 668 divert_packet(m, 0, off & 0xffff); 669 670 /* If 'tee', continue with original packet */ 671 if (clone != NULL) { 672 m = clone; 673 ip = mtod(m, struct ip *); 674 goto pass; 675 } 676 goto done; 677 } 678 #endif 679 680 #ifdef IPFIREWALL_FORWARD 681 /* Here we check dst to make sure it's directly reachable on the 682 * interface we previously thought it was. 683 * If it isn't (which may be likely in some situations) we have 684 * to re-route it (ie, find a route for the next-hop and the 685 * associated interface) and set them here. This is nested 686 * forwarding which in most cases is undesirable, except where 687 * such control is nigh impossible. So we do it here. 688 * And I'm babbling. 689 */ 690 if (off == 0 && old != dst) { 691 struct in_ifaddr *ia; 692 693 /* It's changed... */ 694 /* There must be a better way to do this next line... */ 695 static struct route sro_fwd, *ro_fwd = &sro_fwd; 696 #ifdef IPFIREWALL_FORWARD_DEBUG 697 printf("IPFIREWALL_FORWARD: New dst ip: "); 698 print_ip(dst->sin_addr); 699 printf("\n"); 700 #endif 701 /* 702 * We need to figure out if we have been forwarded 703 * to a local socket. If so then we should somehow 704 * "loop back" to ip_input, and get directed to the 705 * PCB as if we had received this packet. This is 706 * because it may be dificult to identify the packets 707 * you want to forward until they are being output 708 * and have selected an interface. (e.g. locally 709 * initiated packets) If we used the loopback inteface, 710 * we would not be able to control what happens 711 * as the packet runs through ip_input() as 712 * it is done through a ISR. 713 */ 714 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 715 /* 716 * If the addr to forward to is one 717 * of ours, we pretend to 718 * be the destination for this packet. 719 */ 720 if (IA_SIN(ia)->sin_addr.s_addr == 721 dst->sin_addr.s_addr) 722 break; 723 } 724 if (ia) { 725 /* tell ip_input "dont filter" */ 726 ip_fw_fwd_addr = dst; 727 if (m->m_pkthdr.rcvif == NULL) 728 m->m_pkthdr.rcvif = ifunit("lo0"); 729 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 730 m->m_pkthdr.csum_flags |= 731 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 732 m0->m_pkthdr.csum_data = 0xffff; 733 } 734 m->m_pkthdr.csum_flags |= 735 CSUM_IP_CHECKED | CSUM_IP_VALID; 736 HTONS(ip->ip_len); 737 HTONS(ip->ip_off); 738 ip_input(m); 739 goto done; 740 } 741 /* Some of the logic for this was 742 * nicked from above. 743 * 744 * This rewrites the cached route in a local PCB. 745 * Is this what we want to do? 746 */ 747 bcopy(dst, &ro_fwd->ro_dst, sizeof(*dst)); 748 749 ro_fwd->ro_rt = 0; 750 rtalloc_ign(ro_fwd, RTF_PRCLONING); 751 752 if (ro_fwd->ro_rt == 0) { 753 ipstat.ips_noroute++; 754 error = EHOSTUNREACH; 755 goto bad; 756 } 757 758 ia = ifatoia(ro_fwd->ro_rt->rt_ifa); 759 ifp = ro_fwd->ro_rt->rt_ifp; 760 ro_fwd->ro_rt->rt_use++; 761 if (ro_fwd->ro_rt->rt_flags & RTF_GATEWAY) 762 dst = (struct sockaddr_in *)ro_fwd->ro_rt->rt_gateway; 763 if (ro_fwd->ro_rt->rt_flags & RTF_HOST) 764 isbroadcast = 765 (ro_fwd->ro_rt->rt_flags & RTF_BROADCAST); 766 else 767 isbroadcast = in_broadcast(dst->sin_addr, ifp); 768 RTFREE(ro->ro_rt); 769 ro->ro_rt = ro_fwd->ro_rt; 770 dst = (struct sockaddr_in *)&ro_fwd->ro_dst; 771 772 /* 773 * If we added a default src ip earlier, 774 * which would have been gotten from the-then 775 * interface, do it again, from the new one. 776 */ 777 if (fwd_rewrite_src) 778 ip->ip_src = IA_SIN(ia)->sin_addr; 779 goto pass ; 780 } 781 #endif /* IPFIREWALL_FORWARD */ 782 /* 783 * if we get here, none of the above matches, and 784 * we have to drop the pkt 785 */ 786 m_freem(m); 787 error = EACCES; /* not sure this is the right error msg */ 788 goto done; 789 } 790 791 pass: 792 m->m_pkthdr.csum_flags |= CSUM_IP; 793 sw_csum = m->m_pkthdr.csum_flags & ~ifp->if_hwassist; 794 if (sw_csum & CSUM_DELAY_DATA) { 795 in_delayed_cksum(m); 796 sw_csum &= ~CSUM_DELAY_DATA; 797 } 798 m->m_pkthdr.csum_flags &= ifp->if_hwassist; 799 800 /* 801 * If small enough for interface, or the interface will take 802 * care of the fragmentation for us, can just send directly. 803 */ 804 if ((u_short)ip->ip_len <= ifp->if_mtu || 805 ifp->if_hwassist & CSUM_FRAGMENT) { 806 HTONS(ip->ip_len); 807 HTONS(ip->ip_off); 808 ip->ip_sum = 0; 809 if (sw_csum & CSUM_DELAY_IP) { 810 if (ip->ip_vhl == IP_VHL_BORING) { 811 ip->ip_sum = in_cksum_hdr(ip); 812 } else { 813 ip->ip_sum = in_cksum(m, hlen); 814 } 815 } 816 817 /* Record statistics for this interface address. */ 818 if (!(flags & IP_FORWARDING)) { 819 ia->ia_ifa.if_opackets++; 820 ia->ia_ifa.if_obytes += m->m_pkthdr.len; 821 } 822 823 #ifdef IPSEC 824 /* clean ipsec history once it goes out of the node */ 825 ipsec_delaux(m); 826 #endif 827 828 error = (*ifp->if_output)(ifp, m, 829 (struct sockaddr *)dst, ro->ro_rt); 830 goto done; 831 } 832 /* 833 * Too large for interface; fragment if possible. 834 * Must be able to put at least 8 bytes per fragment. 835 */ 836 if (ip->ip_off & IP_DF) { 837 error = EMSGSIZE; 838 /* 839 * This case can happen if the user changed the MTU 840 * of an interface after enabling IP on it. Because 841 * most netifs don't keep track of routes pointing to 842 * them, there is no way for one to update all its 843 * routes when the MTU is changed. 844 */ 845 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 846 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 847 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 848 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 849 } 850 ipstat.ips_cantfrag++; 851 goto bad; 852 } 853 len = (ifp->if_mtu - hlen) &~ 7; 854 if (len < 8) { 855 error = EMSGSIZE; 856 goto bad; 857 } 858 859 /* 860 * if the interface will not calculate checksums on 861 * fragmented packets, then do it here. 862 */ 863 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA && 864 (ifp->if_hwassist & CSUM_IP_FRAGS) == 0) { 865 in_delayed_cksum(m); 866 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 867 } 868 869 { 870 int mhlen, firstlen = len; 871 struct mbuf **mnext = &m->m_nextpkt; 872 int nfrags = 1; 873 874 /* 875 * Loop through length of segment after first fragment, 876 * make new header and copy data of each part and link onto chain. 877 */ 878 m0 = m; 879 mhlen = sizeof (struct ip); 880 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 881 MGETHDR(m, M_DONTWAIT, MT_HEADER); 882 if (m == 0) { 883 error = ENOBUFS; 884 ipstat.ips_odropped++; 885 goto sendorfree; 886 } 887 m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG; 888 m->m_data += max_linkhdr; 889 mhip = mtod(m, struct ip *); 890 *mhip = *ip; 891 if (hlen > sizeof (struct ip)) { 892 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 893 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 894 } 895 m->m_len = mhlen; 896 mhip->ip_off = ((off - hlen) >> 3) + ip->ip_off; 897 if (off + len >= (u_short)ip->ip_len) 898 len = (u_short)ip->ip_len - off; 899 else 900 mhip->ip_off |= IP_MF; 901 mhip->ip_len = htons((u_short)(len + mhlen)); 902 m->m_next = m_copy(m0, off, len); 903 if (m->m_next == 0) { 904 (void) m_free(m); 905 error = ENOBUFS; /* ??? */ 906 ipstat.ips_odropped++; 907 goto sendorfree; 908 } 909 m->m_pkthdr.len = mhlen + len; 910 m->m_pkthdr.rcvif = (struct ifnet *)0; 911 m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags; 912 HTONS(mhip->ip_off); 913 mhip->ip_sum = 0; 914 if (sw_csum & CSUM_DELAY_IP) { 915 if (mhip->ip_vhl == IP_VHL_BORING) { 916 mhip->ip_sum = in_cksum_hdr(mhip); 917 } else { 918 mhip->ip_sum = in_cksum(m, mhlen); 919 } 920 } 921 *mnext = m; 922 mnext = &m->m_nextpkt; 923 nfrags++; 924 } 925 ipstat.ips_ofragments += nfrags; 926 927 /* set first/last markers for fragment chain */ 928 m->m_flags |= M_LASTFRAG; 929 m0->m_flags |= M_FIRSTFRAG | M_FRAG; 930 m0->m_pkthdr.csum_data = nfrags; 931 932 /* 933 * Update first fragment by trimming what's been copied out 934 * and updating header, then send each fragment (in order). 935 */ 936 m = m0; 937 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 938 m->m_pkthdr.len = hlen + firstlen; 939 ip->ip_len = htons((u_short)m->m_pkthdr.len); 940 ip->ip_off |= IP_MF; 941 HTONS(ip->ip_off); 942 ip->ip_sum = 0; 943 if (sw_csum & CSUM_DELAY_IP) { 944 if (ip->ip_vhl == IP_VHL_BORING) { 945 ip->ip_sum = in_cksum_hdr(ip); 946 } else { 947 ip->ip_sum = in_cksum(m, hlen); 948 } 949 } 950 sendorfree: 951 for (m = m0; m; m = m0) { 952 m0 = m->m_nextpkt; 953 m->m_nextpkt = 0; 954 #ifdef IPSEC 955 /* clean ipsec history once it goes out of the node */ 956 ipsec_delaux(m); 957 #endif 958 if (error == 0) { 959 /* Record statistics for this interface address. */ 960 ia->ia_ifa.if_opackets++; 961 ia->ia_ifa.if_obytes += m->m_pkthdr.len; 962 963 error = (*ifp->if_output)(ifp, m, 964 (struct sockaddr *)dst, ro->ro_rt); 965 } else 966 m_freem(m); 967 } 968 969 if (error == 0) 970 ipstat.ips_fragmented++; 971 } 972 done: 973 #ifdef IPSEC 974 if (ro == &iproute && ro->ro_rt) { 975 RTFREE(ro->ro_rt); 976 ro->ro_rt = NULL; 977 } 978 if (sp != NULL) { 979 KEYDEBUG(KEYDEBUG_IPSEC_STAMP, 980 printf("DP ip_output call free SP:%p\n", sp)); 981 key_freesp(sp); 982 } 983 #endif /* IPSEC */ 984 return (error); 985 bad: 986 m_freem(m0); 987 goto done; 988 } 989 990 void 991 in_delayed_cksum(struct mbuf *m) 992 { 993 struct ip *ip; 994 u_short csum, offset; 995 996 ip = mtod(m, struct ip *); 997 offset = IP_VHL_HL(ip->ip_vhl) << 2 ; 998 csum = in_cksum_skip(m, ip->ip_len, offset); 999 if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0) 1000 csum = 0xffff; 1001 offset += m->m_pkthdr.csum_data; /* checksum offset */ 1002 1003 if (offset + sizeof(u_short) > m->m_len) { 1004 printf("delayed m_pullup, m->len: %d off: %d p: %d\n", 1005 m->m_len, offset, ip->ip_p); 1006 /* 1007 * XXX 1008 * this shouldn't happen, but if it does, the 1009 * correct behavior may be to insert the checksum 1010 * in the existing chain instead of rearranging it. 1011 */ 1012 m = m_pullup(m, offset + sizeof(u_short)); 1013 } 1014 *(u_short *)(m->m_data + offset) = csum; 1015 } 1016 1017 /* 1018 * Insert IP options into preformed packet. 1019 * Adjust IP destination as required for IP source routing, 1020 * as indicated by a non-zero in_addr at the start of the options. 1021 * 1022 * XXX This routine assumes that the packet has no options in place. 1023 */ 1024 static struct mbuf * 1025 ip_insertoptions(m, opt, phlen) 1026 register struct mbuf *m; 1027 struct mbuf *opt; 1028 int *phlen; 1029 { 1030 register struct ipoption *p = mtod(opt, struct ipoption *); 1031 struct mbuf *n; 1032 register struct ip *ip = mtod(m, struct ip *); 1033 unsigned optlen; 1034 1035 optlen = opt->m_len - sizeof(p->ipopt_dst); 1036 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 1037 return (m); /* XXX should fail */ 1038 if (p->ipopt_dst.s_addr) 1039 ip->ip_dst = p->ipopt_dst; 1040 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 1041 MGETHDR(n, M_DONTWAIT, MT_HEADER); 1042 if (n == 0) 1043 return (m); 1044 n->m_pkthdr.rcvif = (struct ifnet *)0; 1045 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 1046 m->m_len -= sizeof(struct ip); 1047 m->m_data += sizeof(struct ip); 1048 n->m_next = m; 1049 m = n; 1050 m->m_len = optlen + sizeof(struct ip); 1051 m->m_data += max_linkhdr; 1052 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 1053 } else { 1054 m->m_data -= optlen; 1055 m->m_len += optlen; 1056 m->m_pkthdr.len += optlen; 1057 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 1058 } 1059 ip = mtod(m, struct ip *); 1060 bcopy(p->ipopt_list, ip + 1, optlen); 1061 *phlen = sizeof(struct ip) + optlen; 1062 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 1063 ip->ip_len += optlen; 1064 return (m); 1065 } 1066 1067 /* 1068 * Copy options from ip to jp, 1069 * omitting those not copied during fragmentation. 1070 */ 1071 int 1072 ip_optcopy(ip, jp) 1073 struct ip *ip, *jp; 1074 { 1075 register u_char *cp, *dp; 1076 int opt, optlen, cnt; 1077 1078 cp = (u_char *)(ip + 1); 1079 dp = (u_char *)(jp + 1); 1080 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 1081 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1082 opt = cp[0]; 1083 if (opt == IPOPT_EOL) 1084 break; 1085 if (opt == IPOPT_NOP) { 1086 /* Preserve for IP mcast tunnel's LSRR alignment. */ 1087 *dp++ = IPOPT_NOP; 1088 optlen = 1; 1089 continue; 1090 } 1091 #ifdef DIAGNOSTIC 1092 if (cnt < IPOPT_OLEN + sizeof(*cp)) 1093 panic("malformed IPv4 option passed to ip_optcopy"); 1094 #endif 1095 optlen = cp[IPOPT_OLEN]; 1096 #ifdef DIAGNOSTIC 1097 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) 1098 panic("malformed IPv4 option passed to ip_optcopy"); 1099 #endif 1100 /* bogus lengths should have been caught by ip_dooptions */ 1101 if (optlen > cnt) 1102 optlen = cnt; 1103 if (IPOPT_COPIED(opt)) { 1104 bcopy(cp, dp, optlen); 1105 dp += optlen; 1106 } 1107 } 1108 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 1109 *dp++ = IPOPT_EOL; 1110 return (optlen); 1111 } 1112 1113 /* 1114 * IP socket option processing. 1115 */ 1116 int 1117 ip_ctloutput(so, sopt) 1118 struct socket *so; 1119 struct sockopt *sopt; 1120 { 1121 struct inpcb *inp = sotoinpcb(so); 1122 int error, optval; 1123 1124 error = optval = 0; 1125 if (sopt->sopt_level != IPPROTO_IP) { 1126 return (EINVAL); 1127 } 1128 1129 switch (sopt->sopt_dir) { 1130 case SOPT_SET: 1131 switch (sopt->sopt_name) { 1132 case IP_OPTIONS: 1133 #ifdef notyet 1134 case IP_RETOPTS: 1135 #endif 1136 { 1137 struct mbuf *m; 1138 if (sopt->sopt_valsize > MLEN) { 1139 error = EMSGSIZE; 1140 break; 1141 } 1142 MGET(m, sopt->sopt_p ? M_TRYWAIT : M_DONTWAIT, MT_HEADER); 1143 if (m == 0) { 1144 error = ENOBUFS; 1145 break; 1146 } 1147 m->m_len = sopt->sopt_valsize; 1148 error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 1149 m->m_len); 1150 1151 return (ip_pcbopts(sopt->sopt_name, &inp->inp_options, 1152 m)); 1153 } 1154 1155 case IP_TOS: 1156 case IP_TTL: 1157 case IP_RECVOPTS: 1158 case IP_RECVRETOPTS: 1159 case IP_RECVDSTADDR: 1160 case IP_RECVIF: 1161 #if defined(NFAITH) && NFAITH > 0 1162 case IP_FAITH: 1163 #endif 1164 error = sooptcopyin(sopt, &optval, sizeof optval, 1165 sizeof optval); 1166 if (error) 1167 break; 1168 1169 switch (sopt->sopt_name) { 1170 case IP_TOS: 1171 inp->inp_ip_tos = optval; 1172 break; 1173 1174 case IP_TTL: 1175 inp->inp_ip_ttl = optval; 1176 break; 1177 #define OPTSET(bit) \ 1178 if (optval) \ 1179 inp->inp_flags |= bit; \ 1180 else \ 1181 inp->inp_flags &= ~bit; 1182 1183 case IP_RECVOPTS: 1184 OPTSET(INP_RECVOPTS); 1185 break; 1186 1187 case IP_RECVRETOPTS: 1188 OPTSET(INP_RECVRETOPTS); 1189 break; 1190 1191 case IP_RECVDSTADDR: 1192 OPTSET(INP_RECVDSTADDR); 1193 break; 1194 1195 case IP_RECVIF: 1196 OPTSET(INP_RECVIF); 1197 break; 1198 1199 #if defined(NFAITH) && NFAITH > 0 1200 case IP_FAITH: 1201 OPTSET(INP_FAITH); 1202 break; 1203 #endif 1204 } 1205 break; 1206 #undef OPTSET 1207 1208 case IP_MULTICAST_IF: 1209 case IP_MULTICAST_VIF: 1210 case IP_MULTICAST_TTL: 1211 case IP_MULTICAST_LOOP: 1212 case IP_ADD_MEMBERSHIP: 1213 case IP_DROP_MEMBERSHIP: 1214 error = ip_setmoptions(sopt, &inp->inp_moptions); 1215 break; 1216 1217 case IP_PORTRANGE: 1218 error = sooptcopyin(sopt, &optval, sizeof optval, 1219 sizeof optval); 1220 if (error) 1221 break; 1222 1223 switch (optval) { 1224 case IP_PORTRANGE_DEFAULT: 1225 inp->inp_flags &= ~(INP_LOWPORT); 1226 inp->inp_flags &= ~(INP_HIGHPORT); 1227 break; 1228 1229 case IP_PORTRANGE_HIGH: 1230 inp->inp_flags &= ~(INP_LOWPORT); 1231 inp->inp_flags |= INP_HIGHPORT; 1232 break; 1233 1234 case IP_PORTRANGE_LOW: 1235 inp->inp_flags &= ~(INP_HIGHPORT); 1236 inp->inp_flags |= INP_LOWPORT; 1237 break; 1238 1239 default: 1240 error = EINVAL; 1241 break; 1242 } 1243 break; 1244 1245 #ifdef IPSEC 1246 case IP_IPSEC_POLICY: 1247 { 1248 caddr_t req; 1249 size_t len = 0; 1250 int priv; 1251 struct mbuf *m; 1252 int optname; 1253 1254 if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */ 1255 break; 1256 if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */ 1257 break; 1258 priv = (sopt->sopt_p != NULL && 1259 suser(sopt->sopt_p) != 0) ? 0 : 1; 1260 req = mtod(m, caddr_t); 1261 len = m->m_len; 1262 optname = sopt->sopt_name; 1263 error = ipsec4_set_policy(inp, optname, req, len, priv); 1264 m_freem(m); 1265 break; 1266 } 1267 #endif /*IPSEC*/ 1268 1269 default: 1270 error = ENOPROTOOPT; 1271 break; 1272 } 1273 break; 1274 1275 case SOPT_GET: 1276 switch (sopt->sopt_name) { 1277 case IP_OPTIONS: 1278 case IP_RETOPTS: 1279 if (inp->inp_options) 1280 error = sooptcopyout(sopt, 1281 mtod(inp->inp_options, 1282 char *), 1283 inp->inp_options->m_len); 1284 else 1285 sopt->sopt_valsize = 0; 1286 break; 1287 1288 case IP_TOS: 1289 case IP_TTL: 1290 case IP_RECVOPTS: 1291 case IP_RECVRETOPTS: 1292 case IP_RECVDSTADDR: 1293 case IP_RECVIF: 1294 case IP_PORTRANGE: 1295 #if defined(NFAITH) && NFAITH > 0 1296 case IP_FAITH: 1297 #endif 1298 switch (sopt->sopt_name) { 1299 1300 case IP_TOS: 1301 optval = inp->inp_ip_tos; 1302 break; 1303 1304 case IP_TTL: 1305 optval = inp->inp_ip_ttl; 1306 break; 1307 1308 #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 1309 1310 case IP_RECVOPTS: 1311 optval = OPTBIT(INP_RECVOPTS); 1312 break; 1313 1314 case IP_RECVRETOPTS: 1315 optval = OPTBIT(INP_RECVRETOPTS); 1316 break; 1317 1318 case IP_RECVDSTADDR: 1319 optval = OPTBIT(INP_RECVDSTADDR); 1320 break; 1321 1322 case IP_RECVIF: 1323 optval = OPTBIT(INP_RECVIF); 1324 break; 1325 1326 case IP_PORTRANGE: 1327 if (inp->inp_flags & INP_HIGHPORT) 1328 optval = IP_PORTRANGE_HIGH; 1329 else if (inp->inp_flags & INP_LOWPORT) 1330 optval = IP_PORTRANGE_LOW; 1331 else 1332 optval = 0; 1333 break; 1334 1335 #if defined(NFAITH) && NFAITH > 0 1336 case IP_FAITH: 1337 optval = OPTBIT(INP_FAITH); 1338 break; 1339 #endif 1340 } 1341 error = sooptcopyout(sopt, &optval, sizeof optval); 1342 break; 1343 1344 case IP_MULTICAST_IF: 1345 case IP_MULTICAST_VIF: 1346 case IP_MULTICAST_TTL: 1347 case IP_MULTICAST_LOOP: 1348 case IP_ADD_MEMBERSHIP: 1349 case IP_DROP_MEMBERSHIP: 1350 error = ip_getmoptions(sopt, inp->inp_moptions); 1351 break; 1352 1353 #ifdef IPSEC 1354 case IP_IPSEC_POLICY: 1355 { 1356 struct mbuf *m = NULL; 1357 caddr_t req = NULL; 1358 size_t len = 0; 1359 1360 if (m != 0) { 1361 req = mtod(m, caddr_t); 1362 len = m->m_len; 1363 } 1364 error = ipsec4_get_policy(sotoinpcb(so), req, len, &m); 1365 if (error == 0) 1366 error = soopt_mcopyout(sopt, m); /* XXX */ 1367 if (error == 0) 1368 m_freem(m); 1369 break; 1370 } 1371 #endif /*IPSEC*/ 1372 1373 default: 1374 error = ENOPROTOOPT; 1375 break; 1376 } 1377 break; 1378 } 1379 return (error); 1380 } 1381 1382 /* 1383 * Set up IP options in pcb for insertion in output packets. 1384 * Store in mbuf with pointer in pcbopt, adding pseudo-option 1385 * with destination address if source routed. 1386 */ 1387 static int 1388 ip_pcbopts(optname, pcbopt, m) 1389 int optname; 1390 struct mbuf **pcbopt; 1391 register struct mbuf *m; 1392 { 1393 register int cnt, optlen; 1394 register u_char *cp; 1395 u_char opt; 1396 1397 /* turn off any old options */ 1398 if (*pcbopt) 1399 (void)m_free(*pcbopt); 1400 *pcbopt = 0; 1401 if (m == (struct mbuf *)0 || m->m_len == 0) { 1402 /* 1403 * Only turning off any previous options. 1404 */ 1405 if (m) 1406 (void)m_free(m); 1407 return (0); 1408 } 1409 1410 if (m->m_len % sizeof(int32_t)) 1411 goto bad; 1412 /* 1413 * IP first-hop destination address will be stored before 1414 * actual options; move other options back 1415 * and clear it when none present. 1416 */ 1417 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 1418 goto bad; 1419 cnt = m->m_len; 1420 m->m_len += sizeof(struct in_addr); 1421 cp = mtod(m, u_char *) + sizeof(struct in_addr); 1422 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 1423 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 1424 1425 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1426 opt = cp[IPOPT_OPTVAL]; 1427 if (opt == IPOPT_EOL) 1428 break; 1429 if (opt == IPOPT_NOP) 1430 optlen = 1; 1431 else { 1432 if (cnt < IPOPT_OLEN + sizeof(*cp)) 1433 goto bad; 1434 optlen = cp[IPOPT_OLEN]; 1435 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) 1436 goto bad; 1437 } 1438 switch (opt) { 1439 1440 default: 1441 break; 1442 1443 case IPOPT_LSRR: 1444 case IPOPT_SSRR: 1445 /* 1446 * user process specifies route as: 1447 * ->A->B->C->D 1448 * D must be our final destination (but we can't 1449 * check that since we may not have connected yet). 1450 * A is first hop destination, which doesn't appear in 1451 * actual IP option, but is stored before the options. 1452 */ 1453 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 1454 goto bad; 1455 m->m_len -= sizeof(struct in_addr); 1456 cnt -= sizeof(struct in_addr); 1457 optlen -= sizeof(struct in_addr); 1458 cp[IPOPT_OLEN] = optlen; 1459 /* 1460 * Move first hop before start of options. 1461 */ 1462 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 1463 sizeof(struct in_addr)); 1464 /* 1465 * Then copy rest of options back 1466 * to close up the deleted entry. 1467 */ 1468 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 1469 sizeof(struct in_addr)), 1470 (caddr_t)&cp[IPOPT_OFFSET+1], 1471 (unsigned)cnt + sizeof(struct in_addr)); 1472 break; 1473 } 1474 } 1475 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 1476 goto bad; 1477 *pcbopt = m; 1478 return (0); 1479 1480 bad: 1481 (void)m_free(m); 1482 return (EINVAL); 1483 } 1484 1485 /* 1486 * XXX 1487 * The whole multicast option thing needs to be re-thought. 1488 * Several of these options are equally applicable to non-multicast 1489 * transmission, and one (IP_MULTICAST_TTL) totally duplicates a 1490 * standard option (IP_TTL). 1491 */ 1492 1493 /* 1494 * following RFC1724 section 3.3, 0.0.0.0/8 is interpreted as interface index. 1495 */ 1496 static struct ifnet * 1497 ip_multicast_if(a, ifindexp) 1498 struct in_addr *a; 1499 int *ifindexp; 1500 { 1501 int ifindex; 1502 struct ifnet *ifp; 1503 1504 if (ifindexp) 1505 *ifindexp = 0; 1506 if (ntohl(a->s_addr) >> 24 == 0) { 1507 ifindex = ntohl(a->s_addr) & 0xffffff; 1508 if (ifindex < 0 || if_index < ifindex) 1509 return NULL; 1510 ifp = ifindex2ifnet[ifindex]; 1511 if (ifindexp) 1512 *ifindexp = ifindex; 1513 } else { 1514 INADDR_TO_IFP(*a, ifp); 1515 } 1516 return ifp; 1517 } 1518 1519 /* 1520 * Set the IP multicast options in response to user setsockopt(). 1521 */ 1522 static int 1523 ip_setmoptions(sopt, imop) 1524 struct sockopt *sopt; 1525 struct ip_moptions **imop; 1526 { 1527 int error = 0; 1528 int i; 1529 struct in_addr addr; 1530 struct ip_mreq mreq; 1531 struct ifnet *ifp; 1532 struct ip_moptions *imo = *imop; 1533 struct route ro; 1534 struct sockaddr_in *dst; 1535 int ifindex; 1536 int s; 1537 1538 if (imo == NULL) { 1539 /* 1540 * No multicast option buffer attached to the pcb; 1541 * allocate one and initialize to default values. 1542 */ 1543 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 1544 M_WAITOK); 1545 1546 if (imo == NULL) 1547 return (ENOBUFS); 1548 *imop = imo; 1549 imo->imo_multicast_ifp = NULL; 1550 imo->imo_multicast_addr.s_addr = INADDR_ANY; 1551 imo->imo_multicast_vif = -1; 1552 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 1553 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 1554 imo->imo_num_memberships = 0; 1555 } 1556 1557 switch (sopt->sopt_name) { 1558 /* store an index number for the vif you wanna use in the send */ 1559 case IP_MULTICAST_VIF: 1560 if (legal_vif_num == 0) { 1561 error = EOPNOTSUPP; 1562 break; 1563 } 1564 error = sooptcopyin(sopt, &i, sizeof i, sizeof i); 1565 if (error) 1566 break; 1567 if (!legal_vif_num(i) && (i != -1)) { 1568 error = EINVAL; 1569 break; 1570 } 1571 imo->imo_multicast_vif = i; 1572 break; 1573 1574 case IP_MULTICAST_IF: 1575 /* 1576 * Select the interface for outgoing multicast packets. 1577 */ 1578 error = sooptcopyin(sopt, &addr, sizeof addr, sizeof addr); 1579 if (error) 1580 break; 1581 /* 1582 * INADDR_ANY is used to remove a previous selection. 1583 * When no interface is selected, a default one is 1584 * chosen every time a multicast packet is sent. 1585 */ 1586 if (addr.s_addr == INADDR_ANY) { 1587 imo->imo_multicast_ifp = NULL; 1588 break; 1589 } 1590 /* 1591 * The selected interface is identified by its local 1592 * IP address. Find the interface and confirm that 1593 * it supports multicasting. 1594 */ 1595 s = splimp(); 1596 ifp = ip_multicast_if(&addr, &ifindex); 1597 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1598 splx(s); 1599 error = EADDRNOTAVAIL; 1600 break; 1601 } 1602 imo->imo_multicast_ifp = ifp; 1603 if (ifindex) 1604 imo->imo_multicast_addr = addr; 1605 else 1606 imo->imo_multicast_addr.s_addr = INADDR_ANY; 1607 splx(s); 1608 break; 1609 1610 case IP_MULTICAST_TTL: 1611 /* 1612 * Set the IP time-to-live for outgoing multicast packets. 1613 * The original multicast API required a char argument, 1614 * which is inconsistent with the rest of the socket API. 1615 * We allow either a char or an int. 1616 */ 1617 if (sopt->sopt_valsize == 1) { 1618 u_char ttl; 1619 error = sooptcopyin(sopt, &ttl, 1, 1); 1620 if (error) 1621 break; 1622 imo->imo_multicast_ttl = ttl; 1623 } else { 1624 u_int ttl; 1625 error = sooptcopyin(sopt, &ttl, sizeof ttl, 1626 sizeof ttl); 1627 if (error) 1628 break; 1629 if (ttl > 255) 1630 error = EINVAL; 1631 else 1632 imo->imo_multicast_ttl = ttl; 1633 } 1634 break; 1635 1636 case IP_MULTICAST_LOOP: 1637 /* 1638 * Set the loopback flag for outgoing multicast packets. 1639 * Must be zero or one. The original multicast API required a 1640 * char argument, which is inconsistent with the rest 1641 * of the socket API. We allow either a char or an int. 1642 */ 1643 if (sopt->sopt_valsize == 1) { 1644 u_char loop; 1645 error = sooptcopyin(sopt, &loop, 1, 1); 1646 if (error) 1647 break; 1648 imo->imo_multicast_loop = !!loop; 1649 } else { 1650 u_int loop; 1651 error = sooptcopyin(sopt, &loop, sizeof loop, 1652 sizeof loop); 1653 if (error) 1654 break; 1655 imo->imo_multicast_loop = !!loop; 1656 } 1657 break; 1658 1659 case IP_ADD_MEMBERSHIP: 1660 /* 1661 * Add a multicast group membership. 1662 * Group must be a valid IP multicast address. 1663 */ 1664 error = sooptcopyin(sopt, &mreq, sizeof mreq, sizeof mreq); 1665 if (error) 1666 break; 1667 1668 if (!IN_MULTICAST(ntohl(mreq.imr_multiaddr.s_addr))) { 1669 error = EINVAL; 1670 break; 1671 } 1672 s = splimp(); 1673 /* 1674 * If no interface address was provided, use the interface of 1675 * the route to the given multicast address. 1676 */ 1677 if (mreq.imr_interface.s_addr == INADDR_ANY) { 1678 bzero((caddr_t)&ro, sizeof(ro)); 1679 dst = (struct sockaddr_in *)&ro.ro_dst; 1680 dst->sin_len = sizeof(*dst); 1681 dst->sin_family = AF_INET; 1682 dst->sin_addr = mreq.imr_multiaddr; 1683 rtalloc(&ro); 1684 if (ro.ro_rt == NULL) { 1685 error = EADDRNOTAVAIL; 1686 splx(s); 1687 break; 1688 } 1689 ifp = ro.ro_rt->rt_ifp; 1690 rtfree(ro.ro_rt); 1691 } 1692 else { 1693 ifp = ip_multicast_if(&mreq.imr_interface, NULL); 1694 } 1695 1696 /* 1697 * See if we found an interface, and confirm that it 1698 * supports multicast. 1699 */ 1700 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1701 error = EADDRNOTAVAIL; 1702 splx(s); 1703 break; 1704 } 1705 /* 1706 * See if the membership already exists or if all the 1707 * membership slots are full. 1708 */ 1709 for (i = 0; i < imo->imo_num_memberships; ++i) { 1710 if (imo->imo_membership[i]->inm_ifp == ifp && 1711 imo->imo_membership[i]->inm_addr.s_addr 1712 == mreq.imr_multiaddr.s_addr) 1713 break; 1714 } 1715 if (i < imo->imo_num_memberships) { 1716 error = EADDRINUSE; 1717 splx(s); 1718 break; 1719 } 1720 if (i == IP_MAX_MEMBERSHIPS) { 1721 error = ETOOMANYREFS; 1722 splx(s); 1723 break; 1724 } 1725 /* 1726 * Everything looks good; add a new record to the multicast 1727 * address list for the given interface. 1728 */ 1729 if ((imo->imo_membership[i] = 1730 in_addmulti(&mreq.imr_multiaddr, ifp)) == NULL) { 1731 error = ENOBUFS; 1732 splx(s); 1733 break; 1734 } 1735 ++imo->imo_num_memberships; 1736 splx(s); 1737 break; 1738 1739 case IP_DROP_MEMBERSHIP: 1740 /* 1741 * Drop a multicast group membership. 1742 * Group must be a valid IP multicast address. 1743 */ 1744 error = sooptcopyin(sopt, &mreq, sizeof mreq, sizeof mreq); 1745 if (error) 1746 break; 1747 1748 if (!IN_MULTICAST(ntohl(mreq.imr_multiaddr.s_addr))) { 1749 error = EINVAL; 1750 break; 1751 } 1752 1753 s = splimp(); 1754 /* 1755 * If an interface address was specified, get a pointer 1756 * to its ifnet structure. 1757 */ 1758 if (mreq.imr_interface.s_addr == INADDR_ANY) 1759 ifp = NULL; 1760 else { 1761 ifp = ip_multicast_if(&mreq.imr_interface, NULL); 1762 if (ifp == NULL) { 1763 error = EADDRNOTAVAIL; 1764 splx(s); 1765 break; 1766 } 1767 } 1768 /* 1769 * Find the membership in the membership array. 1770 */ 1771 for (i = 0; i < imo->imo_num_memberships; ++i) { 1772 if ((ifp == NULL || 1773 imo->imo_membership[i]->inm_ifp == ifp) && 1774 imo->imo_membership[i]->inm_addr.s_addr == 1775 mreq.imr_multiaddr.s_addr) 1776 break; 1777 } 1778 if (i == imo->imo_num_memberships) { 1779 error = EADDRNOTAVAIL; 1780 splx(s); 1781 break; 1782 } 1783 /* 1784 * Give up the multicast address record to which the 1785 * membership points. 1786 */ 1787 in_delmulti(imo->imo_membership[i]); 1788 /* 1789 * Remove the gap in the membership array. 1790 */ 1791 for (++i; i < imo->imo_num_memberships; ++i) 1792 imo->imo_membership[i-1] = imo->imo_membership[i]; 1793 --imo->imo_num_memberships; 1794 splx(s); 1795 break; 1796 1797 default: 1798 error = EOPNOTSUPP; 1799 break; 1800 } 1801 1802 /* 1803 * If all options have default values, no need to keep the mbuf. 1804 */ 1805 if (imo->imo_multicast_ifp == NULL && 1806 imo->imo_multicast_vif == -1 && 1807 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1808 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1809 imo->imo_num_memberships == 0) { 1810 free(*imop, M_IPMOPTS); 1811 *imop = NULL; 1812 } 1813 1814 return (error); 1815 } 1816 1817 /* 1818 * Return the IP multicast options in response to user getsockopt(). 1819 */ 1820 static int 1821 ip_getmoptions(sopt, imo) 1822 struct sockopt *sopt; 1823 register struct ip_moptions *imo; 1824 { 1825 struct in_addr addr; 1826 struct in_ifaddr *ia; 1827 int error, optval; 1828 u_char coptval; 1829 1830 error = 0; 1831 switch (sopt->sopt_name) { 1832 case IP_MULTICAST_VIF: 1833 if (imo != NULL) 1834 optval = imo->imo_multicast_vif; 1835 else 1836 optval = -1; 1837 error = sooptcopyout(sopt, &optval, sizeof optval); 1838 break; 1839 1840 case IP_MULTICAST_IF: 1841 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1842 addr.s_addr = INADDR_ANY; 1843 else if (imo->imo_multicast_addr.s_addr) { 1844 /* return the value user has set */ 1845 addr = imo->imo_multicast_addr; 1846 } else { 1847 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1848 addr.s_addr = (ia == NULL) ? INADDR_ANY 1849 : IA_SIN(ia)->sin_addr.s_addr; 1850 } 1851 error = sooptcopyout(sopt, &addr, sizeof addr); 1852 break; 1853 1854 case IP_MULTICAST_TTL: 1855 if (imo == 0) 1856 optval = coptval = IP_DEFAULT_MULTICAST_TTL; 1857 else 1858 optval = coptval = imo->imo_multicast_ttl; 1859 if (sopt->sopt_valsize == 1) 1860 error = sooptcopyout(sopt, &coptval, 1); 1861 else 1862 error = sooptcopyout(sopt, &optval, sizeof optval); 1863 break; 1864 1865 case IP_MULTICAST_LOOP: 1866 if (imo == 0) 1867 optval = coptval = IP_DEFAULT_MULTICAST_LOOP; 1868 else 1869 optval = coptval = imo->imo_multicast_loop; 1870 if (sopt->sopt_valsize == 1) 1871 error = sooptcopyout(sopt, &coptval, 1); 1872 else 1873 error = sooptcopyout(sopt, &optval, sizeof optval); 1874 break; 1875 1876 default: 1877 error = ENOPROTOOPT; 1878 break; 1879 } 1880 return (error); 1881 } 1882 1883 /* 1884 * Discard the IP multicast options. 1885 */ 1886 void 1887 ip_freemoptions(imo) 1888 register struct ip_moptions *imo; 1889 { 1890 register int i; 1891 1892 if (imo != NULL) { 1893 for (i = 0; i < imo->imo_num_memberships; ++i) 1894 in_delmulti(imo->imo_membership[i]); 1895 free(imo, M_IPMOPTS); 1896 } 1897 } 1898 1899 /* 1900 * Routine called from ip_output() to loop back a copy of an IP multicast 1901 * packet to the input queue of a specified interface. Note that this 1902 * calls the output routine of the loopback "driver", but with an interface 1903 * pointer that might NOT be a loopback interface -- evil, but easier than 1904 * replicating that code here. 1905 */ 1906 static void 1907 ip_mloopback(ifp, m, dst, hlen) 1908 struct ifnet *ifp; 1909 register struct mbuf *m; 1910 register struct sockaddr_in *dst; 1911 int hlen; 1912 { 1913 register struct ip *ip; 1914 struct mbuf *copym; 1915 1916 copym = m_copy(m, 0, M_COPYALL); 1917 if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen)) 1918 copym = m_pullup(copym, hlen); 1919 if (copym != NULL) { 1920 /* 1921 * We don't bother to fragment if the IP length is greater 1922 * than the interface's MTU. Can this possibly matter? 1923 */ 1924 ip = mtod(copym, struct ip *); 1925 HTONS(ip->ip_len); 1926 HTONS(ip->ip_off); 1927 ip->ip_sum = 0; 1928 if (ip->ip_vhl == IP_VHL_BORING) { 1929 ip->ip_sum = in_cksum_hdr(ip); 1930 } else { 1931 ip->ip_sum = in_cksum(copym, hlen); 1932 } 1933 /* 1934 * NB: 1935 * It's not clear whether there are any lingering 1936 * reentrancy problems in other areas which might 1937 * be exposed by using ip_input directly (in 1938 * particular, everything which modifies the packet 1939 * in-place). Yet another option is using the 1940 * protosw directly to deliver the looped back 1941 * packet. For the moment, we'll err on the side 1942 * of safety by using if_simloop(). 1943 */ 1944 #if 1 /* XXX */ 1945 if (dst->sin_family != AF_INET) { 1946 printf("ip_mloopback: bad address family %d\n", 1947 dst->sin_family); 1948 dst->sin_family = AF_INET; 1949 } 1950 #endif 1951 1952 #ifdef notdef 1953 copym->m_pkthdr.rcvif = ifp; 1954 ip_input(copym); 1955 #else 1956 /* if the checksum hasn't been computed, mark it as valid */ 1957 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1958 copym->m_pkthdr.csum_flags |= 1959 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1960 copym->m_pkthdr.csum_data = 0xffff; 1961 } 1962 if_simloop(ifp, copym, dst->sin_family, 0); 1963 #endif 1964 } 1965 } 1966