ip_output.c revision 12296
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 * $Id: ip_output.c,v 1.24 1995/10/16 18:21:09 wollman Exp $ 35 */ 36 37#include <sys/param.h> 38#include <sys/systm.h> 39#include <sys/malloc.h> 40#include <sys/mbuf.h> 41#include <sys/errno.h> 42#include <sys/protosw.h> 43#include <sys/socket.h> 44#include <sys/socketvar.h> 45#include <sys/queue.h> 46 47#include <net/if.h> 48#include <net/route.h> 49 50#include <netinet/in.h> 51#include <netinet/in_systm.h> 52#include <netinet/ip.h> 53#include <netinet/in_pcb.h> 54#include <netinet/in_var.h> 55#include <netinet/ip_var.h> 56 57#include <netinet/ip_fw.h> 58 59#ifdef vax 60#include <machine/mtpr.h> 61#endif 62 63u_short ip_id; 64 65static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 66static void ip_mloopback 67 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *)); 68static int ip_getmoptions 69 __P((int, struct ip_moptions *, struct mbuf **)); 70static int ip_optcopy __P((struct ip *, struct ip *)); 71static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 72static int ip_setmoptions 73 __P((int, struct ip_moptions **, struct mbuf *)); 74 75/* 76 * IP output. The packet in mbuf chain m contains a skeletal IP 77 * header (with len, off, ttl, proto, tos, src, dst). 78 * The mbuf chain containing the packet will be freed. 79 * The mbuf opt, if present, will not be freed. 80 */ 81int 82ip_output(m0, opt, ro, flags, imo) 83 struct mbuf *m0; 84 struct mbuf *opt; 85 struct route *ro; 86 int flags; 87 struct ip_moptions *imo; 88{ 89 register struct ip *ip, *mhip; 90 register struct ifnet *ifp; 91 register struct mbuf *m = m0; 92 register int hlen = sizeof (struct ip); 93 int len, off, error = 0; 94 struct route iproute; 95 struct sockaddr_in *dst; 96 struct in_ifaddr *ia; 97 98#ifdef DIAGNOSTIC 99 if ((m->m_flags & M_PKTHDR) == 0) 100 panic("ip_output no HDR"); 101#endif 102 if (opt) { 103 m = ip_insertoptions(m, opt, &len); 104 hlen = len; 105 } 106 ip = mtod(m, struct ip *); 107 /* 108 * Fill in IP header. 109 */ 110 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 111 ip->ip_v = IPVERSION; 112 ip->ip_off &= IP_DF; 113 ip->ip_id = htons(ip_id++); 114 ip->ip_hl = hlen >> 2; 115 ipstat.ips_localout++; 116 } else { 117 hlen = ip->ip_hl << 2; 118 } 119 /* 120 * Route packet. 121 */ 122 if (ro == 0) { 123 ro = &iproute; 124 bzero((caddr_t)ro, sizeof (*ro)); 125 } 126 dst = (struct sockaddr_in *)&ro->ro_dst; 127 /* 128 * If there is a cached route, 129 * check that it is to the same destination 130 * and is still up. If not, free it and try again. 131 */ 132 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 133 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 134 RTFREE(ro->ro_rt); 135 ro->ro_rt = (struct rtentry *)0; 136 } 137 if (ro->ro_rt == 0) { 138 dst->sin_family = AF_INET; 139 dst->sin_len = sizeof(*dst); 140 dst->sin_addr = ip->ip_dst; 141 } 142 /* 143 * If routing to interface only, 144 * short circuit routing lookup. 145 */ 146#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 147#define sintosa(sin) ((struct sockaddr *)(sin)) 148 if (flags & IP_ROUTETOIF) { 149 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 150 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 151 ipstat.ips_noroute++; 152 error = ENETUNREACH; 153 goto bad; 154 } 155 ifp = ia->ia_ifp; 156 ip->ip_ttl = 1; 157 } else { 158 /* 159 * If this is the case, we probably don't want to allocate 160 * a protocol-cloned route since we didn't get one from the 161 * ULP. This lets TCP do its thing, while not burdening 162 * forwarding or ICMP with the overhead of cloning a route. 163 * Of course, we still want to do any cloning requested by 164 * the link layer, as this is probably required in all cases 165 * for correct operation (as it is for ARP). 166 */ 167 if (ro->ro_rt == 0) 168 rtalloc_ign(ro, RTF_PRCLONING); 169 if (ro->ro_rt == 0) { 170 ipstat.ips_noroute++; 171 error = EHOSTUNREACH; 172 goto bad; 173 } 174 ia = ifatoia(ro->ro_rt->rt_ifa); 175 ifp = ro->ro_rt->rt_ifp; 176 ro->ro_rt->rt_use++; 177 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 178 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 179 } 180 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 181 struct in_multi *inm; 182 183 m->m_flags |= M_MCAST; 184 /* 185 * IP destination address is multicast. Make sure "dst" 186 * still points to the address in "ro". (It may have been 187 * changed to point to a gateway address, above.) 188 */ 189 dst = (struct sockaddr_in *)&ro->ro_dst; 190 /* 191 * See if the caller provided any multicast options 192 */ 193 if (imo != NULL) { 194 ip->ip_ttl = imo->imo_multicast_ttl; 195 if (imo->imo_multicast_ifp != NULL) 196 ifp = imo->imo_multicast_ifp; 197 if (imo->imo_multicast_vif != -1) 198 ip->ip_src.s_addr = 199 ip_mcast_src(imo->imo_multicast_vif); 200 } else 201 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 202 /* 203 * Confirm that the outgoing interface supports multicast. 204 */ 205 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 206 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 207 ipstat.ips_noroute++; 208 error = ENETUNREACH; 209 goto bad; 210 } 211 } 212 /* 213 * If source address not specified yet, use address 214 * of outgoing interface. 215 */ 216 if (ip->ip_src.s_addr == INADDR_ANY) { 217 register struct in_ifaddr *ia; 218 219 for (ia = in_ifaddr; ia; ia = ia->ia_next) 220 if (ia->ia_ifp == ifp) { 221 ip->ip_src = IA_SIN(ia)->sin_addr; 222 break; 223 } 224 } 225 226 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 227 if (inm != NULL && 228 (imo == NULL || imo->imo_multicast_loop)) { 229 /* 230 * If we belong to the destination multicast group 231 * on the outgoing interface, and the caller did not 232 * forbid loopback, loop back a copy. 233 */ 234 ip_mloopback(ifp, m, dst); 235 } 236 else { 237 /* 238 * If we are acting as a multicast router, perform 239 * multicast forwarding as if the packet had just 240 * arrived on the interface to which we are about 241 * to send. The multicast forwarding function 242 * recursively calls this function, using the 243 * IP_FORWARDING flag to prevent infinite recursion. 244 * 245 * Multicasts that are looped back by ip_mloopback(), 246 * above, will be forwarded by the ip_input() routine, 247 * if necessary. 248 */ 249 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 250 /* 251 * Check if rsvp daemon is running. If not, don't 252 * set ip_moptions. This ensures that the packet 253 * is multicast and not just sent down one link 254 * as prescribed by rsvpd. 255 */ 256 if (!rsvp_on) 257 imo = NULL; 258 if (ip_mforward(ip, ifp, m, imo) != 0) { 259 m_freem(m); 260 goto done; 261 } 262 } 263 } 264 265 /* 266 * Multicasts with a time-to-live of zero may be looped- 267 * back, above, but must not be transmitted on a network. 268 * Also, multicasts addressed to the loopback interface 269 * are not sent -- the above call to ip_mloopback() will 270 * loop back a copy if this host actually belongs to the 271 * destination group on the loopback interface. 272 */ 273 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 274 m_freem(m); 275 goto done; 276 } 277 278 goto sendit; 279 } 280#ifndef notdef 281 /* 282 * If source address not specified yet, use address 283 * of outgoing interface. 284 */ 285 if (ip->ip_src.s_addr == INADDR_ANY) 286 ip->ip_src = IA_SIN(ia)->sin_addr; 287#endif 288 /* 289 * Verify that we have any chance at all of being able to queue 290 * the packet or packet fragments 291 */ 292 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 293 ifp->if_snd.ifq_maxlen) { 294 error = ENOBUFS; 295 goto bad; 296 } 297 298 /* 299 * Look for broadcast address and 300 * and verify user is allowed to send 301 * such a packet. 302 */ 303 if (in_broadcast(dst->sin_addr, ifp)) { 304 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 305 error = EADDRNOTAVAIL; 306 goto bad; 307 } 308 if ((flags & IP_ALLOWBROADCAST) == 0) { 309 error = EACCES; 310 goto bad; 311 } 312 /* don't allow broadcast messages to be fragmented */ 313 if ((u_short)ip->ip_len > ifp->if_mtu) { 314 error = EMSGSIZE; 315 goto bad; 316 } 317 m->m_flags |= M_BCAST; 318 } else 319 m->m_flags &= ~M_BCAST; 320 321sendit: 322 /* 323 * If small enough for interface, can just send directly. 324 */ 325 if ((u_short)ip->ip_len <= ifp->if_mtu) { 326 ip->ip_len = htons((u_short)ip->ip_len); 327 ip->ip_off = htons((u_short)ip->ip_off); 328 ip->ip_sum = 0; 329 ip->ip_sum = in_cksum(m, hlen); 330 error = (*ifp->if_output)(ifp, m, 331 (struct sockaddr *)dst, ro->ro_rt); 332 goto done; 333 } 334 /* 335 * Too large for interface; fragment if possible. 336 * Must be able to put at least 8 bytes per fragment. 337 */ 338 if (ip->ip_off & IP_DF) { 339 error = EMSGSIZE; 340#ifdef MTUDISC 341 /* 342 * This case can happen if the user changed the MTU 343 * of an interface after enabling IP on it. Because 344 * most netifs don't keep track of routes pointing to 345 * them, there is no way for one to update all its 346 * routes when the MTU is changed. 347 */ 348 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 349 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 350 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 351 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 352 } 353#endif /* MTUDISC */ 354 ipstat.ips_cantfrag++; 355 goto bad; 356 } 357 len = (ifp->if_mtu - hlen) &~ 7; 358 if (len < 8) { 359 error = EMSGSIZE; 360 goto bad; 361 } 362 363 { 364 int mhlen, firstlen = len; 365 struct mbuf **mnext = &m->m_nextpkt; 366 367 /* 368 * Loop through length of segment after first fragment, 369 * make new header and copy data of each part and link onto chain. 370 */ 371 m0 = m; 372 mhlen = sizeof (struct ip); 373 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 374 MGETHDR(m, M_DONTWAIT, MT_HEADER); 375 if (m == 0) { 376 error = ENOBUFS; 377 ipstat.ips_odropped++; 378 goto sendorfree; 379 } 380 m->m_data += max_linkhdr; 381 mhip = mtod(m, struct ip *); 382 *mhip = *ip; 383 if (hlen > sizeof (struct ip)) { 384 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 385 mhip->ip_hl = mhlen >> 2; 386 } 387 m->m_len = mhlen; 388 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 389 if (ip->ip_off & IP_MF) 390 mhip->ip_off |= IP_MF; 391 if (off + len >= (u_short)ip->ip_len) 392 len = (u_short)ip->ip_len - off; 393 else 394 mhip->ip_off |= IP_MF; 395 mhip->ip_len = htons((u_short)(len + mhlen)); 396 m->m_next = m_copy(m0, off, len); 397 if (m->m_next == 0) { 398 (void) m_free(m); 399 error = ENOBUFS; /* ??? */ 400 ipstat.ips_odropped++; 401 goto sendorfree; 402 } 403 m->m_pkthdr.len = mhlen + len; 404 m->m_pkthdr.rcvif = (struct ifnet *)0; 405 mhip->ip_off = htons((u_short)mhip->ip_off); 406 mhip->ip_sum = 0; 407 mhip->ip_sum = in_cksum(m, mhlen); 408 *mnext = m; 409 mnext = &m->m_nextpkt; 410 ipstat.ips_ofragments++; 411 } 412 /* 413 * Update first fragment by trimming what's been copied out 414 * and updating header, then send each fragment (in order). 415 */ 416 m = m0; 417 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 418 m->m_pkthdr.len = hlen + firstlen; 419 ip->ip_len = htons((u_short)m->m_pkthdr.len); 420 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 421 ip->ip_sum = 0; 422 ip->ip_sum = in_cksum(m, hlen); 423sendorfree: 424 for (m = m0; m; m = m0) { 425 m0 = m->m_nextpkt; 426 m->m_nextpkt = 0; 427 if (error == 0) 428 error = (*ifp->if_output)(ifp, m, 429 (struct sockaddr *)dst, ro->ro_rt); 430 else 431 m_freem(m); 432 } 433 434 if (error == 0) 435 ipstat.ips_fragmented++; 436 } 437done: 438 if (ro == &iproute && (flags & IP_ROUTETOIF) == 0 && ro->ro_rt) 439 RTFREE(ro->ro_rt); 440 /* 441 * Count outgoing packet,here we count both our packets and 442 * those we forward. 443 * Here we want to convert ip_len to host byte order when counting 444 * so we set 3rd arg to 1. 445 * This is locally generated packet so it has not 446 * incoming interface. 447 */ 448 if (ip_acct_cnt_ptr!=NULL) 449 (*ip_acct_cnt_ptr)(ip,NULL,ip_acct_chain,1); 450 451 return (error); 452bad: 453 m_freem(m0); 454 goto done; 455} 456 457/* 458 * Insert IP options into preformed packet. 459 * Adjust IP destination as required for IP source routing, 460 * as indicated by a non-zero in_addr at the start of the options. 461 */ 462static struct mbuf * 463ip_insertoptions(m, opt, phlen) 464 register struct mbuf *m; 465 struct mbuf *opt; 466 int *phlen; 467{ 468 register struct ipoption *p = mtod(opt, struct ipoption *); 469 struct mbuf *n; 470 register struct ip *ip = mtod(m, struct ip *); 471 unsigned optlen; 472 473 optlen = opt->m_len - sizeof(p->ipopt_dst); 474 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 475 return (m); /* XXX should fail */ 476 if (p->ipopt_dst.s_addr) 477 ip->ip_dst = p->ipopt_dst; 478 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 479 MGETHDR(n, M_DONTWAIT, MT_HEADER); 480 if (n == 0) 481 return (m); 482 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 483 m->m_len -= sizeof(struct ip); 484 m->m_data += sizeof(struct ip); 485 n->m_next = m; 486 m = n; 487 m->m_len = optlen + sizeof(struct ip); 488 m->m_data += max_linkhdr; 489 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 490 } else { 491 m->m_data -= optlen; 492 m->m_len += optlen; 493 m->m_pkthdr.len += optlen; 494 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 495 } 496 ip = mtod(m, struct ip *); 497 (void)memcpy(ip + 1, p->ipopt_list, (unsigned)optlen); 498 *phlen = sizeof(struct ip) + optlen; 499 ip->ip_len += optlen; 500 return (m); 501} 502 503/* 504 * Copy options from ip to jp, 505 * omitting those not copied during fragmentation. 506 */ 507static int 508ip_optcopy(ip, jp) 509 struct ip *ip, *jp; 510{ 511 register u_char *cp, *dp; 512 int opt, optlen, cnt; 513 514 cp = (u_char *)(ip + 1); 515 dp = (u_char *)(jp + 1); 516 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 517 for (; cnt > 0; cnt -= optlen, cp += optlen) { 518 opt = cp[0]; 519 if (opt == IPOPT_EOL) 520 break; 521 if (opt == IPOPT_NOP) { 522 /* Preserve for IP mcast tunnel's LSRR alignment. */ 523 *dp++ = IPOPT_NOP; 524 optlen = 1; 525 continue; 526 } else 527 optlen = cp[IPOPT_OLEN]; 528 /* bogus lengths should have been caught by ip_dooptions */ 529 if (optlen > cnt) 530 optlen = cnt; 531 if (IPOPT_COPIED(opt)) { 532 (void)memcpy(dp, cp, (unsigned)optlen); 533 dp += optlen; 534 } 535 } 536 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 537 *dp++ = IPOPT_EOL; 538 return (optlen); 539} 540 541/* 542 * IP socket option processing. 543 */ 544int 545ip_ctloutput(op, so, level, optname, mp) 546 int op; 547 struct socket *so; 548 int level, optname; 549 struct mbuf **mp; 550{ 551 register struct inpcb *inp = sotoinpcb(so); 552 register struct mbuf *m = *mp; 553 register int optval = 0; 554 int error = 0; 555 556 if (level != IPPROTO_IP) { 557 error = EINVAL; 558 if (op == PRCO_SETOPT && *mp) 559 (void) m_free(*mp); 560 } else switch (op) { 561 562 case PRCO_SETOPT: 563 switch (optname) { 564 case IP_OPTIONS: 565#ifdef notyet 566 case IP_RETOPTS: 567 return (ip_pcbopts(optname, &inp->inp_options, m)); 568#else 569 return (ip_pcbopts(&inp->inp_options, m)); 570#endif 571 572 case IP_TOS: 573 case IP_TTL: 574 case IP_RECVOPTS: 575 case IP_RECVRETOPTS: 576 case IP_RECVDSTADDR: 577 if (m == 0 || m->m_len != sizeof(int)) 578 error = EINVAL; 579 else { 580 optval = *mtod(m, int *); 581 switch (optname) { 582 583 case IP_TOS: 584 inp->inp_ip.ip_tos = optval; 585 break; 586 587 case IP_TTL: 588 inp->inp_ip.ip_ttl = optval; 589 break; 590#define OPTSET(bit) \ 591 if (optval) \ 592 inp->inp_flags |= bit; \ 593 else \ 594 inp->inp_flags &= ~bit; 595 596 case IP_RECVOPTS: 597 OPTSET(INP_RECVOPTS); 598 break; 599 600 case IP_RECVRETOPTS: 601 OPTSET(INP_RECVRETOPTS); 602 break; 603 604 case IP_RECVDSTADDR: 605 OPTSET(INP_RECVDSTADDR); 606 break; 607 } 608 } 609 break; 610#undef OPTSET 611 612 case IP_MULTICAST_IF: 613 case IP_MULTICAST_VIF: 614 case IP_MULTICAST_TTL: 615 case IP_MULTICAST_LOOP: 616 case IP_ADD_MEMBERSHIP: 617 case IP_DROP_MEMBERSHIP: 618 error = ip_setmoptions(optname, &inp->inp_moptions, m); 619 break; 620 621 default: 622 error = ENOPROTOOPT; 623 break; 624 } 625 if (m) 626 (void)m_free(m); 627 break; 628 629 case PRCO_GETOPT: 630 switch (optname) { 631 case IP_OPTIONS: 632 case IP_RETOPTS: 633 *mp = m = m_get(M_WAIT, MT_SOOPTS); 634 if (inp->inp_options) { 635 m->m_len = inp->inp_options->m_len; 636 (void)memcpy(mtod(m, void *), 637 mtod(inp->inp_options, void *), (unsigned)m->m_len); 638 } else 639 m->m_len = 0; 640 break; 641 642 case IP_TOS: 643 case IP_TTL: 644 case IP_RECVOPTS: 645 case IP_RECVRETOPTS: 646 case IP_RECVDSTADDR: 647 *mp = m = m_get(M_WAIT, MT_SOOPTS); 648 m->m_len = sizeof(int); 649 switch (optname) { 650 651 case IP_TOS: 652 optval = inp->inp_ip.ip_tos; 653 break; 654 655 case IP_TTL: 656 optval = inp->inp_ip.ip_ttl; 657 break; 658 659#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 660 661 case IP_RECVOPTS: 662 optval = OPTBIT(INP_RECVOPTS); 663 break; 664 665 case IP_RECVRETOPTS: 666 optval = OPTBIT(INP_RECVRETOPTS); 667 break; 668 669 case IP_RECVDSTADDR: 670 optval = OPTBIT(INP_RECVDSTADDR); 671 break; 672 } 673 *mtod(m, int *) = optval; 674 break; 675 676 case IP_MULTICAST_IF: 677 case IP_MULTICAST_VIF: 678 case IP_MULTICAST_TTL: 679 case IP_MULTICAST_LOOP: 680 case IP_ADD_MEMBERSHIP: 681 case IP_DROP_MEMBERSHIP: 682 error = ip_getmoptions(optname, inp->inp_moptions, mp); 683 break; 684 685 default: 686 error = ENOPROTOOPT; 687 break; 688 } 689 break; 690 } 691 return (error); 692} 693 694/* 695 * Set up IP options in pcb for insertion in output packets. 696 * Store in mbuf with pointer in pcbopt, adding pseudo-option 697 * with destination address if source routed. 698 */ 699static int 700#ifdef notyet 701ip_pcbopts(optname, pcbopt, m) 702 int optname; 703#else 704ip_pcbopts(pcbopt, m) 705#endif 706 struct mbuf **pcbopt; 707 register struct mbuf *m; 708{ 709 register cnt, optlen; 710 register u_char *cp; 711 u_char opt; 712 713 /* turn off any old options */ 714 if (*pcbopt) 715 (void)m_free(*pcbopt); 716 *pcbopt = 0; 717 if (m == (struct mbuf *)0 || m->m_len == 0) { 718 /* 719 * Only turning off any previous options. 720 */ 721 if (m) 722 (void)m_free(m); 723 return (0); 724 } 725 726#ifndef vax 727 if (m->m_len % sizeof(long)) 728 goto bad; 729#endif 730 /* 731 * IP first-hop destination address will be stored before 732 * actual options; move other options back 733 * and clear it when none present. 734 */ 735 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 736 goto bad; 737 cnt = m->m_len; 738 m->m_len += sizeof(struct in_addr); 739 cp = mtod(m, u_char *) + sizeof(struct in_addr); 740 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 741 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 742 743 for (; cnt > 0; cnt -= optlen, cp += optlen) { 744 opt = cp[IPOPT_OPTVAL]; 745 if (opt == IPOPT_EOL) 746 break; 747 if (opt == IPOPT_NOP) 748 optlen = 1; 749 else { 750 optlen = cp[IPOPT_OLEN]; 751 if (optlen <= IPOPT_OLEN || optlen > cnt) 752 goto bad; 753 } 754 switch (opt) { 755 756 default: 757 break; 758 759 case IPOPT_LSRR: 760 case IPOPT_SSRR: 761 /* 762 * user process specifies route as: 763 * ->A->B->C->D 764 * D must be our final destination (but we can't 765 * check that since we may not have connected yet). 766 * A is first hop destination, which doesn't appear in 767 * actual IP option, but is stored before the options. 768 */ 769 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 770 goto bad; 771 m->m_len -= sizeof(struct in_addr); 772 cnt -= sizeof(struct in_addr); 773 optlen -= sizeof(struct in_addr); 774 cp[IPOPT_OLEN] = optlen; 775 /* 776 * Move first hop before start of options. 777 */ 778 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 779 sizeof(struct in_addr)); 780 /* 781 * Then copy rest of options back 782 * to close up the deleted entry. 783 */ 784 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 785 sizeof(struct in_addr)), 786 (caddr_t)&cp[IPOPT_OFFSET+1], 787 (unsigned)cnt + sizeof(struct in_addr)); 788 break; 789 } 790 } 791 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 792 goto bad; 793 *pcbopt = m; 794 return (0); 795 796bad: 797 (void)m_free(m); 798 return (EINVAL); 799} 800 801/* 802 * Set the IP multicast options in response to user setsockopt(). 803 */ 804static int 805ip_setmoptions(optname, imop, m) 806 int optname; 807 struct ip_moptions **imop; 808 struct mbuf *m; 809{ 810 register int error = 0; 811 u_char loop; 812 register int i; 813 struct in_addr addr; 814 register struct ip_mreq *mreq; 815 register struct ifnet *ifp; 816 register struct ip_moptions *imo = *imop; 817 struct route ro; 818 register struct sockaddr_in *dst; 819 int s; 820 821 if (imo == NULL) { 822 /* 823 * No multicast option buffer attached to the pcb; 824 * allocate one and initialize to default values. 825 */ 826 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 827 M_WAITOK); 828 829 if (imo == NULL) 830 return (ENOBUFS); 831 *imop = imo; 832 imo->imo_multicast_ifp = NULL; 833 imo->imo_multicast_vif = -1; 834 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 835 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 836 imo->imo_num_memberships = 0; 837 } 838 839 switch (optname) { 840 /* store an index number for the vif you wanna use in the send */ 841 case IP_MULTICAST_VIF: 842 if (!legal_vif_num) { 843 error = EOPNOTSUPP; 844 break; 845 } 846 if (m == NULL || m->m_len != sizeof(int)) { 847 error = EINVAL; 848 break; 849 } 850 i = *(mtod(m, int *)); 851 if (!legal_vif_num(i) && (i != -1)) { 852 error = EINVAL; 853 break; 854 } 855 imo->imo_multicast_vif = i; 856 break; 857 858 case IP_MULTICAST_IF: 859 /* 860 * Select the interface for outgoing multicast packets. 861 */ 862 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 863 error = EINVAL; 864 break; 865 } 866 addr = *(mtod(m, struct in_addr *)); 867 /* 868 * INADDR_ANY is used to remove a previous selection. 869 * When no interface is selected, a default one is 870 * chosen every time a multicast packet is sent. 871 */ 872 if (addr.s_addr == INADDR_ANY) { 873 imo->imo_multicast_ifp = NULL; 874 break; 875 } 876 /* 877 * The selected interface is identified by its local 878 * IP address. Find the interface and confirm that 879 * it supports multicasting. 880 */ 881 s = splimp(); 882 INADDR_TO_IFP(addr, ifp); 883 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 884 error = EADDRNOTAVAIL; 885 break; 886 } 887 imo->imo_multicast_ifp = ifp; 888 splx(s); 889 break; 890 891 case IP_MULTICAST_TTL: 892 /* 893 * Set the IP time-to-live for outgoing multicast packets. 894 */ 895 if (m == NULL || m->m_len != 1) { 896 error = EINVAL; 897 break; 898 } 899 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 900 break; 901 902 case IP_MULTICAST_LOOP: 903 /* 904 * Set the loopback flag for outgoing multicast packets. 905 * Must be zero or one. 906 */ 907 if (m == NULL || m->m_len != 1 || 908 (loop = *(mtod(m, u_char *))) > 1) { 909 error = EINVAL; 910 break; 911 } 912 imo->imo_multicast_loop = loop; 913 break; 914 915 case IP_ADD_MEMBERSHIP: 916 /* 917 * Add a multicast group membership. 918 * Group must be a valid IP multicast address. 919 */ 920 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 921 error = EINVAL; 922 break; 923 } 924 mreq = mtod(m, struct ip_mreq *); 925 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 926 error = EINVAL; 927 break; 928 } 929 s = splimp(); 930 /* 931 * If no interface address was provided, use the interface of 932 * the route to the given multicast address. 933 */ 934 if (mreq->imr_interface.s_addr == INADDR_ANY) { 935 bzero((caddr_t)&ro, sizeof(ro)); 936 dst = (struct sockaddr_in *)&ro.ro_dst; 937 dst->sin_len = sizeof(*dst); 938 dst->sin_family = AF_INET; 939 dst->sin_addr = mreq->imr_multiaddr; 940 rtalloc(&ro); 941 if (ro.ro_rt == NULL) { 942 error = EADDRNOTAVAIL; 943 splx(s); 944 break; 945 } 946 ifp = ro.ro_rt->rt_ifp; 947 rtfree(ro.ro_rt); 948 } 949 else { 950 INADDR_TO_IFP(mreq->imr_interface, ifp); 951 } 952 953 /* 954 * See if we found an interface, and confirm that it 955 * supports multicast. 956 */ 957 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 958 error = EADDRNOTAVAIL; 959 splx(s); 960 break; 961 } 962 /* 963 * See if the membership already exists or if all the 964 * membership slots are full. 965 */ 966 for (i = 0; i < imo->imo_num_memberships; ++i) { 967 if (imo->imo_membership[i]->inm_ifp == ifp && 968 imo->imo_membership[i]->inm_addr.s_addr 969 == mreq->imr_multiaddr.s_addr) 970 break; 971 } 972 if (i < imo->imo_num_memberships) { 973 error = EADDRINUSE; 974 splx(s); 975 break; 976 } 977 if (i == IP_MAX_MEMBERSHIPS) { 978 error = ETOOMANYREFS; 979 splx(s); 980 break; 981 } 982 /* 983 * Everything looks good; add a new record to the multicast 984 * address list for the given interface. 985 */ 986 if ((imo->imo_membership[i] = 987 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 988 error = ENOBUFS; 989 splx(s); 990 break; 991 } 992 ++imo->imo_num_memberships; 993 splx(s); 994 break; 995 996 case IP_DROP_MEMBERSHIP: 997 /* 998 * Drop a multicast group membership. 999 * Group must be a valid IP multicast address. 1000 */ 1001 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1002 error = EINVAL; 1003 break; 1004 } 1005 mreq = mtod(m, struct ip_mreq *); 1006 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1007 error = EINVAL; 1008 break; 1009 } 1010 1011 s = splimp(); 1012 /* 1013 * If an interface address was specified, get a pointer 1014 * to its ifnet structure. 1015 */ 1016 if (mreq->imr_interface.s_addr == INADDR_ANY) 1017 ifp = NULL; 1018 else { 1019 INADDR_TO_IFP(mreq->imr_interface, ifp); 1020 if (ifp == NULL) { 1021 error = EADDRNOTAVAIL; 1022 splx(s); 1023 break; 1024 } 1025 } 1026 /* 1027 * Find the membership in the membership array. 1028 */ 1029 for (i = 0; i < imo->imo_num_memberships; ++i) { 1030 if ((ifp == NULL || 1031 imo->imo_membership[i]->inm_ifp == ifp) && 1032 imo->imo_membership[i]->inm_addr.s_addr == 1033 mreq->imr_multiaddr.s_addr) 1034 break; 1035 } 1036 if (i == imo->imo_num_memberships) { 1037 error = EADDRNOTAVAIL; 1038 splx(s); 1039 break; 1040 } 1041 /* 1042 * Give up the multicast address record to which the 1043 * membership points. 1044 */ 1045 in_delmulti(imo->imo_membership[i]); 1046 /* 1047 * Remove the gap in the membership array. 1048 */ 1049 for (++i; i < imo->imo_num_memberships; ++i) 1050 imo->imo_membership[i-1] = imo->imo_membership[i]; 1051 --imo->imo_num_memberships; 1052 splx(s); 1053 break; 1054 1055 default: 1056 error = EOPNOTSUPP; 1057 break; 1058 } 1059 1060 /* 1061 * If all options have default values, no need to keep the mbuf. 1062 */ 1063 if (imo->imo_multicast_ifp == NULL && 1064 imo->imo_multicast_vif == -1 && 1065 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1066 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1067 imo->imo_num_memberships == 0) { 1068 free(*imop, M_IPMOPTS); 1069 *imop = NULL; 1070 } 1071 1072 return (error); 1073} 1074 1075/* 1076 * Return the IP multicast options in response to user getsockopt(). 1077 */ 1078static int 1079ip_getmoptions(optname, imo, mp) 1080 int optname; 1081 register struct ip_moptions *imo; 1082 register struct mbuf **mp; 1083{ 1084 u_char *ttl; 1085 u_char *loop; 1086 struct in_addr *addr; 1087 struct in_ifaddr *ia; 1088 1089 *mp = m_get(M_WAIT, MT_SOOPTS); 1090 1091 switch (optname) { 1092 1093 case IP_MULTICAST_VIF: 1094 if (imo != NULL) 1095 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1096 else 1097 *(mtod(*mp, int *)) = -1; 1098 (*mp)->m_len = sizeof(int); 1099 return(0); 1100 1101 case IP_MULTICAST_IF: 1102 addr = mtod(*mp, struct in_addr *); 1103 (*mp)->m_len = sizeof(struct in_addr); 1104 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1105 addr->s_addr = INADDR_ANY; 1106 else { 1107 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1108 addr->s_addr = (ia == NULL) ? INADDR_ANY 1109 : IA_SIN(ia)->sin_addr.s_addr; 1110 } 1111 return (0); 1112 1113 case IP_MULTICAST_TTL: 1114 ttl = mtod(*mp, u_char *); 1115 (*mp)->m_len = 1; 1116 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1117 : imo->imo_multicast_ttl; 1118 return (0); 1119 1120 case IP_MULTICAST_LOOP: 1121 loop = mtod(*mp, u_char *); 1122 (*mp)->m_len = 1; 1123 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1124 : imo->imo_multicast_loop; 1125 return (0); 1126 1127 default: 1128 return (EOPNOTSUPP); 1129 } 1130} 1131 1132/* 1133 * Discard the IP multicast options. 1134 */ 1135void 1136ip_freemoptions(imo) 1137 register struct ip_moptions *imo; 1138{ 1139 register int i; 1140 1141 if (imo != NULL) { 1142 for (i = 0; i < imo->imo_num_memberships; ++i) 1143 in_delmulti(imo->imo_membership[i]); 1144 free(imo, M_IPMOPTS); 1145 } 1146} 1147 1148/* 1149 * Routine called from ip_output() to loop back a copy of an IP multicast 1150 * packet to the input queue of a specified interface. Note that this 1151 * calls the output routine of the loopback "driver", but with an interface 1152 * pointer that might NOT be a loopback interface -- evil, but easier than 1153 * replicating that code here. 1154 */ 1155static void 1156ip_mloopback(ifp, m, dst) 1157 struct ifnet *ifp; 1158 register struct mbuf *m; 1159 register struct sockaddr_in *dst; 1160{ 1161 register struct ip *ip; 1162 struct mbuf *copym; 1163 1164 copym = m_copy(m, 0, M_COPYALL); 1165 if (copym != NULL) { 1166 /* 1167 * We don't bother to fragment if the IP length is greater 1168 * than the interface's MTU. Can this possibly matter? 1169 */ 1170 ip = mtod(copym, struct ip *); 1171 ip->ip_len = htons((u_short)ip->ip_len); 1172 ip->ip_off = htons((u_short)ip->ip_off); 1173 ip->ip_sum = 0; 1174 ip->ip_sum = in_cksum(copym, ip->ip_hl << 2); 1175 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1176 } 1177} 1178