ip_output.c revision 7684
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 1.15 1995/03/20 18:31:51 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 *)); 68 69/* 70 * IP output. The packet in mbuf chain m contains a skeletal IP 71 * header (with len, off, ttl, proto, tos, src, dst). 72 * The mbuf chain containing the packet will be freed. 73 * The mbuf opt, if present, will not be freed. 74 */ 75int 76ip_output(m0, opt, ro, flags, imo) 77 struct mbuf *m0; 78 struct mbuf *opt; 79 struct route *ro; 80 int flags; 81 struct ip_moptions *imo; 82{ 83 register struct ip *ip, *mhip; 84 register struct ifnet *ifp; 85 register struct mbuf *m = m0; 86 register int hlen = sizeof (struct ip); 87 int len, off, error = 0; 88 struct route iproute; 89 struct sockaddr_in *dst; 90 struct in_ifaddr *ia; 91 92#ifdef DIAGNOSTIC 93 if ((m->m_flags & M_PKTHDR) == 0) 94 panic("ip_output no HDR"); 95#endif 96 if (opt) { 97 m = ip_insertoptions(m, opt, &len); 98 hlen = len; 99 } 100 ip = mtod(m, struct ip *); 101 /* 102 * Fill in IP header. 103 */ 104 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 105 ip->ip_v = IPVERSION; 106 ip->ip_off &= IP_DF; 107 ip->ip_id = htons(ip_id++); 108 ip->ip_hl = hlen >> 2; 109 ipstat.ips_localout++; 110 } else { 111 hlen = ip->ip_hl << 2; 112 } 113 /* 114 * Route packet. 115 */ 116 if (ro == 0) { 117 ro = &iproute; 118 bzero((caddr_t)ro, sizeof (*ro)); 119 } 120 dst = (struct sockaddr_in *)&ro->ro_dst; 121 /* 122 * If there is a cached route, 123 * check that it is to the same destination 124 * and is still up. If not, free it and try again. 125 */ 126 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 127 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 128 RTFREE(ro->ro_rt); 129 ro->ro_rt = (struct rtentry *)0; 130 } 131 if (ro->ro_rt == 0) { 132 dst->sin_family = AF_INET; 133 dst->sin_len = sizeof(*dst); 134 dst->sin_addr = ip->ip_dst; 135 } 136 /* 137 * If routing to interface only, 138 * short circuit routing lookup. 139 */ 140#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 141#define sintosa(sin) ((struct sockaddr *)(sin)) 142 if (flags & IP_ROUTETOIF) { 143 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 144 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 145 ipstat.ips_noroute++; 146 error = ENETUNREACH; 147 goto bad; 148 } 149 ifp = ia->ia_ifp; 150 ip->ip_ttl = 1; 151 } else { 152 /* 153 * If this is the case, we probably don't want to allocate 154 * a protocol-cloned route since we didn't get one from the 155 * ULP. This lets TCP do its thing, while not burdening 156 * forwarding or ICMP with the overhead of cloning a route. 157 * Of course, we still want to do any cloning requested by 158 * the link layer, as this is probably required in all cases 159 * for correct operation (as it is for ARP). 160 */ 161 if (ro->ro_rt == 0) 162 rtalloc_ign(ro, RTF_PRCLONING); 163 if (ro->ro_rt == 0) { 164 ipstat.ips_noroute++; 165 error = EHOSTUNREACH; 166 goto bad; 167 } 168 ia = ifatoia(ro->ro_rt->rt_ifa); 169 ifp = ro->ro_rt->rt_ifp; 170 ro->ro_rt->rt_use++; 171 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 172 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 173 } 174 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 175 struct in_multi *inm; 176 177 m->m_flags |= M_MCAST; 178 /* 179 * IP destination address is multicast. Make sure "dst" 180 * still points to the address in "ro". (It may have been 181 * changed to point to a gateway address, above.) 182 */ 183 dst = (struct sockaddr_in *)&ro->ro_dst; 184 /* 185 * See if the caller provided any multicast options 186 */ 187 if (imo != NULL) { 188 ip->ip_ttl = imo->imo_multicast_ttl; 189 if (imo->imo_multicast_ifp != NULL) 190 ifp = imo->imo_multicast_ifp; 191 } else 192 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 193 /* 194 * Confirm that the outgoing interface supports multicast. 195 */ 196 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 197 ipstat.ips_noroute++; 198 error = ENETUNREACH; 199 goto bad; 200 } 201 /* 202 * If source address not specified yet, use address 203 * of outgoing interface. 204 */ 205 if (ip->ip_src.s_addr == INADDR_ANY) { 206 register struct in_ifaddr *ia; 207 208 for (ia = in_ifaddr; ia; ia = ia->ia_next) 209 if (ia->ia_ifp == ifp) { 210 ip->ip_src = IA_SIN(ia)->sin_addr; 211 break; 212 } 213 } 214 215 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 216 if (inm != NULL && 217 (imo == NULL || imo->imo_multicast_loop)) { 218 /* 219 * If we belong to the destination multicast group 220 * on the outgoing interface, and the caller did not 221 * forbid loopback, loop back a copy. 222 */ 223 ip_mloopback(ifp, m, dst); 224 } 225 else { 226 /* 227 * If we are acting as a multicast router, perform 228 * multicast forwarding as if the packet had just 229 * arrived on the interface to which we are about 230 * to send. The multicast forwarding function 231 * recursively calls this function, using the 232 * IP_FORWARDING flag to prevent infinite recursion. 233 * 234 * Multicasts that are looped back by ip_mloopback(), 235 * above, will be forwarded by the ip_input() routine, 236 * if necessary. 237 */ 238 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 239 /* 240 * Check if rsvp daemon is running. If not, don't 241 * set ip_moptions. This ensures that the packet 242 * is multicast and not just sent down one link 243 * as prescribed by rsvpd. 244 */ 245 if (ip_rsvpd == NULL) 246 imo = NULL; 247 if (ip_mforward(ip, ifp, m, imo) != 0) { 248 m_freem(m); 249 goto done; 250 } 251 } 252 } 253 254 /* 255 * Multicasts with a time-to-live of zero may be looped- 256 * back, above, but must not be transmitted on a network. 257 * Also, multicasts addressed to the loopback interface 258 * are not sent -- the above call to ip_mloopback() will 259 * loop back a copy if this host actually belongs to the 260 * destination group on the loopback interface. 261 */ 262 if (ip->ip_ttl == 0 || ifp == &loif) { 263 m_freem(m); 264 goto done; 265 } 266 267 goto sendit; 268 } 269#ifndef notdef 270 /* 271 * If source address not specified yet, use address 272 * of outgoing interface. 273 */ 274 if (ip->ip_src.s_addr == INADDR_ANY) 275 ip->ip_src = IA_SIN(ia)->sin_addr; 276#endif 277 /* 278 * Verify that we have any chance at all of being able to queue 279 * the packet or packet fragments 280 */ 281 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 282 ifp->if_snd.ifq_maxlen) { 283 error = ENOBUFS; 284 goto bad; 285 } 286 287 /* 288 * Look for broadcast address and 289 * and verify user is allowed to send 290 * such a packet. 291 */ 292 if (in_broadcast(dst->sin_addr, ifp)) { 293 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 294 error = EADDRNOTAVAIL; 295 goto bad; 296 } 297 if ((flags & IP_ALLOWBROADCAST) == 0) { 298 error = EACCES; 299 goto bad; 300 } 301 /* don't allow broadcast messages to be fragmented */ 302 if ((u_short)ip->ip_len > ifp->if_mtu) { 303 error = EMSGSIZE; 304 goto bad; 305 } 306 m->m_flags |= M_BCAST; 307 } else 308 m->m_flags &= ~M_BCAST; 309 310sendit: 311 /* 312 * If small enough for interface, can just send directly. 313 */ 314 if ((u_short)ip->ip_len <= ifp->if_mtu) { 315 ip->ip_len = htons((u_short)ip->ip_len); 316 ip->ip_off = htons((u_short)ip->ip_off); 317 ip->ip_sum = 0; 318 ip->ip_sum = in_cksum(m, hlen); 319 error = (*ifp->if_output)(ifp, m, 320 (struct sockaddr *)dst, ro->ro_rt); 321 goto done; 322 } 323 /* 324 * Too large for interface; fragment if possible. 325 * Must be able to put at least 8 bytes per fragment. 326 */ 327 if (ip->ip_off & IP_DF) { 328 error = EMSGSIZE; 329 ipstat.ips_cantfrag++; 330 goto bad; 331 } 332 len = (ifp->if_mtu - hlen) &~ 7; 333 if (len < 8) { 334 error = EMSGSIZE; 335 goto bad; 336 } 337 338 { 339 int mhlen, firstlen = len; 340 struct mbuf **mnext = &m->m_nextpkt; 341 342 /* 343 * Loop through length of segment after first fragment, 344 * make new header and copy data of each part and link onto chain. 345 */ 346 m0 = m; 347 mhlen = sizeof (struct ip); 348 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 349 MGETHDR(m, M_DONTWAIT, MT_HEADER); 350 if (m == 0) { 351 error = ENOBUFS; 352 ipstat.ips_odropped++; 353 goto sendorfree; 354 } 355 m->m_data += max_linkhdr; 356 mhip = mtod(m, struct ip *); 357 *mhip = *ip; 358 if (hlen > sizeof (struct ip)) { 359 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 360 mhip->ip_hl = mhlen >> 2; 361 } 362 m->m_len = mhlen; 363 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 364 if (ip->ip_off & IP_MF) 365 mhip->ip_off |= IP_MF; 366 if (off + len >= (u_short)ip->ip_len) 367 len = (u_short)ip->ip_len - off; 368 else 369 mhip->ip_off |= IP_MF; 370 mhip->ip_len = htons((u_short)(len + mhlen)); 371 m->m_next = m_copy(m0, off, len); 372 if (m->m_next == 0) { 373 (void) m_free(m); 374 error = ENOBUFS; /* ??? */ 375 ipstat.ips_odropped++; 376 goto sendorfree; 377 } 378 m->m_pkthdr.len = mhlen + len; 379 m->m_pkthdr.rcvif = (struct ifnet *)0; 380 mhip->ip_off = htons((u_short)mhip->ip_off); 381 mhip->ip_sum = 0; 382 mhip->ip_sum = in_cksum(m, mhlen); 383 *mnext = m; 384 mnext = &m->m_nextpkt; 385 ipstat.ips_ofragments++; 386 } 387 /* 388 * Update first fragment by trimming what's been copied out 389 * and updating header, then send each fragment (in order). 390 */ 391 m = m0; 392 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 393 m->m_pkthdr.len = hlen + firstlen; 394 ip->ip_len = htons((u_short)m->m_pkthdr.len); 395 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 396 ip->ip_sum = 0; 397 ip->ip_sum = in_cksum(m, hlen); 398sendorfree: 399 for (m = m0; m; m = m0) { 400 m0 = m->m_nextpkt; 401 m->m_nextpkt = 0; 402 if (error == 0) 403 error = (*ifp->if_output)(ifp, m, 404 (struct sockaddr *)dst, ro->ro_rt); 405 else 406 m_freem(m); 407 } 408 409 if (error == 0) 410 ipstat.ips_fragmented++; 411 } 412done: 413 if (ro == &iproute && (flags & IP_ROUTETOIF) == 0 && ro->ro_rt) 414 RTFREE(ro->ro_rt); 415 /* 416 * Count outgoing packet,here we count both our packets and 417 * those we forward. 418 * Here we want to convert ip_len to host byte order when counting 419 * so we set 3rd arg to 1. 420 * This is locally generated packet so it has not 421 * incoming interface. 422 */ 423 if (ip_acct_cnt_ptr!=NULL) 424 (*ip_acct_cnt_ptr)(ip,NULL,ip_acct_chain,1); 425 426 return (error); 427bad: 428 m_freem(m0); 429 goto done; 430} 431 432/* 433 * Insert IP options into preformed packet. 434 * Adjust IP destination as required for IP source routing, 435 * as indicated by a non-zero in_addr at the start of the options. 436 */ 437static struct mbuf * 438ip_insertoptions(m, opt, phlen) 439 register struct mbuf *m; 440 struct mbuf *opt; 441 int *phlen; 442{ 443 register struct ipoption *p = mtod(opt, struct ipoption *); 444 struct mbuf *n; 445 register struct ip *ip = mtod(m, struct ip *); 446 unsigned optlen; 447 448 optlen = opt->m_len - sizeof(p->ipopt_dst); 449 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 450 return (m); /* XXX should fail */ 451 if (p->ipopt_dst.s_addr) 452 ip->ip_dst = p->ipopt_dst; 453 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 454 MGETHDR(n, M_DONTWAIT, MT_HEADER); 455 if (n == 0) 456 return (m); 457 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 458 m->m_len -= sizeof(struct ip); 459 m->m_data += sizeof(struct ip); 460 n->m_next = m; 461 m = n; 462 m->m_len = optlen + sizeof(struct ip); 463 m->m_data += max_linkhdr; 464 bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 465 } else { 466 m->m_data -= optlen; 467 m->m_len += optlen; 468 m->m_pkthdr.len += optlen; 469 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 470 } 471 ip = mtod(m, struct ip *); 472 bcopy((caddr_t)p->ipopt_list, (caddr_t)(ip + 1), (unsigned)optlen); 473 *phlen = sizeof(struct ip) + optlen; 474 ip->ip_len += optlen; 475 return (m); 476} 477 478/* 479 * Copy options from ip to jp, 480 * omitting those not copied during fragmentation. 481 */ 482int 483ip_optcopy(ip, jp) 484 struct ip *ip, *jp; 485{ 486 register u_char *cp, *dp; 487 int opt, optlen, cnt; 488 489 cp = (u_char *)(ip + 1); 490 dp = (u_char *)(jp + 1); 491 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 492 for (; cnt > 0; cnt -= optlen, cp += optlen) { 493 opt = cp[0]; 494 if (opt == IPOPT_EOL) 495 break; 496 if (opt == IPOPT_NOP) { 497 /* Preserve for IP mcast tunnel's LSRR alignment. */ 498 *dp++ = IPOPT_NOP; 499 optlen = 1; 500 continue; 501 } else 502 optlen = cp[IPOPT_OLEN]; 503 /* bogus lengths should have been caught by ip_dooptions */ 504 if (optlen > cnt) 505 optlen = cnt; 506 if (IPOPT_COPIED(opt)) { 507 bcopy((caddr_t)cp, (caddr_t)dp, (unsigned)optlen); 508 dp += optlen; 509 } 510 } 511 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 512 *dp++ = IPOPT_EOL; 513 return (optlen); 514} 515 516/* 517 * IP socket option processing. 518 */ 519int 520ip_ctloutput(op, so, level, optname, mp) 521 int op; 522 struct socket *so; 523 int level, optname; 524 struct mbuf **mp; 525{ 526 register struct inpcb *inp = sotoinpcb(so); 527 register struct mbuf *m = *mp; 528 register int optval = 0; 529 int error = 0; 530 531 if (level != IPPROTO_IP) { 532 error = EINVAL; 533 if (op == PRCO_SETOPT && *mp) 534 (void) m_free(*mp); 535 } else switch (op) { 536 537 case PRCO_SETOPT: 538 switch (optname) { 539 case IP_OPTIONS: 540#ifdef notyet 541 case IP_RETOPTS: 542 return (ip_pcbopts(optname, &inp->inp_options, m)); 543#else 544 return (ip_pcbopts(&inp->inp_options, m)); 545#endif 546 547 case IP_TOS: 548 case IP_TTL: 549 case IP_RECVOPTS: 550 case IP_RECVRETOPTS: 551 case IP_RECVDSTADDR: 552 if (m->m_len != sizeof(int)) 553 error = EINVAL; 554 else { 555 optval = *mtod(m, int *); 556 switch (optname) { 557 558 case IP_TOS: 559 inp->inp_ip.ip_tos = optval; 560 break; 561 562 case IP_TTL: 563 inp->inp_ip.ip_ttl = optval; 564 break; 565#define OPTSET(bit) \ 566 if (optval) \ 567 inp->inp_flags |= bit; \ 568 else \ 569 inp->inp_flags &= ~bit; 570 571 case IP_RECVOPTS: 572 OPTSET(INP_RECVOPTS); 573 break; 574 575 case IP_RECVRETOPTS: 576 OPTSET(INP_RECVRETOPTS); 577 break; 578 579 case IP_RECVDSTADDR: 580 OPTSET(INP_RECVDSTADDR); 581 break; 582 } 583 } 584 break; 585#undef OPTSET 586 587 case IP_MULTICAST_IF: 588 case IP_MULTICAST_VIF: 589 case IP_MULTICAST_TTL: 590 case IP_MULTICAST_LOOP: 591 case IP_ADD_MEMBERSHIP: 592 case IP_DROP_MEMBERSHIP: 593 error = ip_setmoptions(optname, &inp->inp_moptions, m); 594 break; 595 596 default: 597 error = ENOPROTOOPT; 598 break; 599 } 600 if (m) 601 (void)m_free(m); 602 break; 603 604 case PRCO_GETOPT: 605 switch (optname) { 606 case IP_OPTIONS: 607 case IP_RETOPTS: 608 *mp = m = m_get(M_WAIT, MT_SOOPTS); 609 if (inp->inp_options) { 610 m->m_len = inp->inp_options->m_len; 611 bcopy(mtod(inp->inp_options, caddr_t), 612 mtod(m, caddr_t), (unsigned)m->m_len); 613 } else 614 m->m_len = 0; 615 break; 616 617 case IP_TOS: 618 case IP_TTL: 619 case IP_RECVOPTS: 620 case IP_RECVRETOPTS: 621 case IP_RECVDSTADDR: 622 *mp = m = m_get(M_WAIT, MT_SOOPTS); 623 m->m_len = sizeof(int); 624 switch (optname) { 625 626 case IP_TOS: 627 optval = inp->inp_ip.ip_tos; 628 break; 629 630 case IP_TTL: 631 optval = inp->inp_ip.ip_ttl; 632 break; 633 634#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 635 636 case IP_RECVOPTS: 637 optval = OPTBIT(INP_RECVOPTS); 638 break; 639 640 case IP_RECVRETOPTS: 641 optval = OPTBIT(INP_RECVRETOPTS); 642 break; 643 644 case IP_RECVDSTADDR: 645 optval = OPTBIT(INP_RECVDSTADDR); 646 break; 647 } 648 *mtod(m, int *) = optval; 649 break; 650 651 case IP_MULTICAST_IF: 652 case IP_MULTICAST_VIF: 653 case IP_MULTICAST_TTL: 654 case IP_MULTICAST_LOOP: 655 case IP_ADD_MEMBERSHIP: 656 case IP_DROP_MEMBERSHIP: 657 error = ip_getmoptions(optname, inp->inp_moptions, mp); 658 break; 659 660 default: 661 error = ENOPROTOOPT; 662 break; 663 } 664 break; 665 } 666 return (error); 667} 668 669/* 670 * Set up IP options in pcb for insertion in output packets. 671 * Store in mbuf with pointer in pcbopt, adding pseudo-option 672 * with destination address if source routed. 673 */ 674int 675#ifdef notyet 676ip_pcbopts(optname, pcbopt, m) 677 int optname; 678#else 679ip_pcbopts(pcbopt, m) 680#endif 681 struct mbuf **pcbopt; 682 register struct mbuf *m; 683{ 684 register cnt, optlen; 685 register u_char *cp; 686 u_char opt; 687 688 /* turn off any old options */ 689 if (*pcbopt) 690 (void)m_free(*pcbopt); 691 *pcbopt = 0; 692 if (m == (struct mbuf *)0 || m->m_len == 0) { 693 /* 694 * Only turning off any previous options. 695 */ 696 if (m) 697 (void)m_free(m); 698 return (0); 699 } 700 701#ifndef vax 702 if (m->m_len % sizeof(long)) 703 goto bad; 704#endif 705 /* 706 * IP first-hop destination address will be stored before 707 * actual options; move other options back 708 * and clear it when none present. 709 */ 710 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 711 goto bad; 712 cnt = m->m_len; 713 m->m_len += sizeof(struct in_addr); 714 cp = mtod(m, u_char *) + sizeof(struct in_addr); 715 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 716 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 717 718 for (; cnt > 0; cnt -= optlen, cp += optlen) { 719 opt = cp[IPOPT_OPTVAL]; 720 if (opt == IPOPT_EOL) 721 break; 722 if (opt == IPOPT_NOP) 723 optlen = 1; 724 else { 725 optlen = cp[IPOPT_OLEN]; 726 if (optlen <= IPOPT_OLEN || optlen > cnt) 727 goto bad; 728 } 729 switch (opt) { 730 731 default: 732 break; 733 734 case IPOPT_LSRR: 735 case IPOPT_SSRR: 736 /* 737 * user process specifies route as: 738 * ->A->B->C->D 739 * D must be our final destination (but we can't 740 * check that since we may not have connected yet). 741 * A is first hop destination, which doesn't appear in 742 * actual IP option, but is stored before the options. 743 */ 744 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 745 goto bad; 746 m->m_len -= sizeof(struct in_addr); 747 cnt -= sizeof(struct in_addr); 748 optlen -= sizeof(struct in_addr); 749 cp[IPOPT_OLEN] = optlen; 750 /* 751 * Move first hop before start of options. 752 */ 753 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 754 sizeof(struct in_addr)); 755 /* 756 * Then copy rest of options back 757 * to close up the deleted entry. 758 */ 759 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 760 sizeof(struct in_addr)), 761 (caddr_t)&cp[IPOPT_OFFSET+1], 762 (unsigned)cnt + sizeof(struct in_addr)); 763 break; 764 } 765 } 766 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 767 goto bad; 768 *pcbopt = m; 769 return (0); 770 771bad: 772 (void)m_free(m); 773 return (EINVAL); 774} 775 776/* 777 * Set the IP multicast options in response to user setsockopt(). 778 */ 779int 780ip_setmoptions(optname, imop, m) 781 int optname; 782 struct ip_moptions **imop; 783 struct mbuf *m; 784{ 785 register int error = 0; 786 u_char loop; 787 register int i; 788 struct in_addr addr; 789 register struct ip_mreq *mreq; 790 register struct ifnet *ifp; 791 register struct ip_moptions *imo = *imop; 792 struct route ro; 793 register struct sockaddr_in *dst; 794 int s; 795 796 if (imo == NULL) { 797 /* 798 * No multicast option buffer attached to the pcb; 799 * allocate one and initialize to default values. 800 */ 801 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 802 M_WAITOK); 803 804 if (imo == NULL) 805 return (ENOBUFS); 806 *imop = imo; 807 imo->imo_multicast_ifp = NULL; 808 imo->imo_multicast_vif = 0; 809 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 810 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 811 imo->imo_num_memberships = 0; 812 } 813 814 switch (optname) { 815 /* store an index number for the vif you wanna use in the send */ 816 case IP_MULTICAST_VIF: 817 if (!legal_vif_num) { 818 error = EOPNOTSUPP; 819 break; 820 } 821 if (m == NULL || m->m_len != sizeof(int)) { 822 error = EINVAL; 823 break; 824 } 825 i = *(mtod(m, int *)); 826 if (!legal_vif_num(i)) { 827 error = EINVAL; 828 break; 829 } 830 imo->imo_multicast_vif = i; 831 break; 832 833 case IP_MULTICAST_IF: 834 /* 835 * Select the interface for outgoing multicast packets. 836 */ 837 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 838 error = EINVAL; 839 break; 840 } 841 addr = *(mtod(m, struct in_addr *)); 842 /* 843 * INADDR_ANY is used to remove a previous selection. 844 * When no interface is selected, a default one is 845 * chosen every time a multicast packet is sent. 846 */ 847 if (addr.s_addr == INADDR_ANY) { 848 imo->imo_multicast_ifp = NULL; 849 break; 850 } 851 /* 852 * The selected interface is identified by its local 853 * IP address. Find the interface and confirm that 854 * it supports multicasting. 855 */ 856 s = splimp(); 857 INADDR_TO_IFP(addr, ifp); 858 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 859 error = EADDRNOTAVAIL; 860 break; 861 } 862 imo->imo_multicast_ifp = ifp; 863 splx(s); 864 break; 865 866 case IP_MULTICAST_TTL: 867 /* 868 * Set the IP time-to-live for outgoing multicast packets. 869 */ 870 if (m == NULL || m->m_len != 1) { 871 error = EINVAL; 872 break; 873 } 874 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 875 break; 876 877 case IP_MULTICAST_LOOP: 878 /* 879 * Set the loopback flag for outgoing multicast packets. 880 * Must be zero or one. 881 */ 882 if (m == NULL || m->m_len != 1 || 883 (loop = *(mtod(m, u_char *))) > 1) { 884 error = EINVAL; 885 break; 886 } 887 imo->imo_multicast_loop = loop; 888 break; 889 890 case IP_ADD_MEMBERSHIP: 891 /* 892 * Add a multicast group membership. 893 * Group must be a valid IP multicast address. 894 */ 895 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 896 error = EINVAL; 897 break; 898 } 899 mreq = mtod(m, struct ip_mreq *); 900 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 901 error = EINVAL; 902 break; 903 } 904 s = splimp(); 905 /* 906 * If no interface address was provided, use the interface of 907 * the route to the given multicast address. 908 */ 909 if (mreq->imr_interface.s_addr == INADDR_ANY) { 910 ro.ro_rt = NULL; 911 dst = (struct sockaddr_in *)&ro.ro_dst; 912 dst->sin_len = sizeof(*dst); 913 dst->sin_family = AF_INET; 914 dst->sin_addr = mreq->imr_multiaddr; 915 rtalloc(&ro); 916 if (ro.ro_rt == NULL) { 917 error = EADDRNOTAVAIL; 918 splx(s); 919 break; 920 } 921 ifp = ro.ro_rt->rt_ifp; 922 rtfree(ro.ro_rt); 923 } 924 else { 925 INADDR_TO_IFP(mreq->imr_interface, ifp); 926 } 927 928 /* 929 * See if we found an interface, and confirm that it 930 * supports multicast. 931 */ 932 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 933 error = EADDRNOTAVAIL; 934 splx(s); 935 break; 936 } 937 /* 938 * See if the membership already exists or if all the 939 * membership slots are full. 940 */ 941 for (i = 0; i < imo->imo_num_memberships; ++i) { 942 if (imo->imo_membership[i]->inm_ifp == ifp && 943 imo->imo_membership[i]->inm_addr.s_addr 944 == mreq->imr_multiaddr.s_addr) 945 break; 946 } 947 if (i < imo->imo_num_memberships) { 948 error = EADDRINUSE; 949 splx(s); 950 break; 951 } 952 if (i == IP_MAX_MEMBERSHIPS) { 953 error = ETOOMANYREFS; 954 splx(s); 955 break; 956 } 957 /* 958 * Everything looks good; add a new record to the multicast 959 * address list for the given interface. 960 */ 961 if ((imo->imo_membership[i] = 962 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 963 error = ENOBUFS; 964 splx(s); 965 break; 966 } 967 ++imo->imo_num_memberships; 968 splx(s); 969 break; 970 971 case IP_DROP_MEMBERSHIP: 972 /* 973 * Drop a multicast group membership. 974 * Group must be a valid IP multicast address. 975 */ 976 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 977 error = EINVAL; 978 break; 979 } 980 mreq = mtod(m, struct ip_mreq *); 981 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 982 error = EINVAL; 983 break; 984 } 985 986 s = splimp(); 987 /* 988 * If an interface address was specified, get a pointer 989 * to its ifnet structure. 990 */ 991 if (mreq->imr_interface.s_addr == INADDR_ANY) 992 ifp = NULL; 993 else { 994 INADDR_TO_IFP(mreq->imr_interface, ifp); 995 if (ifp == NULL) { 996 error = EADDRNOTAVAIL; 997 splx(s); 998 break; 999 } 1000 } 1001 /* 1002 * Find the membership in the membership array. 1003 */ 1004 for (i = 0; i < imo->imo_num_memberships; ++i) { 1005 if ((ifp == NULL || 1006 imo->imo_membership[i]->inm_ifp == ifp) && 1007 imo->imo_membership[i]->inm_addr.s_addr == 1008 mreq->imr_multiaddr.s_addr) 1009 break; 1010 } 1011 if (i == imo->imo_num_memberships) { 1012 error = EADDRNOTAVAIL; 1013 splx(s); 1014 break; 1015 } 1016 /* 1017 * Give up the multicast address record to which the 1018 * membership points. 1019 */ 1020 in_delmulti(imo->imo_membership[i]); 1021 /* 1022 * Remove the gap in the membership array. 1023 */ 1024 for (++i; i < imo->imo_num_memberships; ++i) 1025 imo->imo_membership[i-1] = imo->imo_membership[i]; 1026 --imo->imo_num_memberships; 1027 splx(s); 1028 break; 1029 1030 default: 1031 error = EOPNOTSUPP; 1032 break; 1033 } 1034 1035 /* 1036 * If all options have default values, no need to keep the mbuf. 1037 */ 1038 if (imo->imo_multicast_ifp == NULL && 1039 imo->imo_multicast_vif == 0 && 1040 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1041 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1042 imo->imo_num_memberships == 0) { 1043 free(*imop, M_IPMOPTS); 1044 *imop = NULL; 1045 } 1046 1047 return (error); 1048} 1049 1050/* 1051 * Return the IP multicast options in response to user getsockopt(). 1052 */ 1053int 1054ip_getmoptions(optname, imo, mp) 1055 int optname; 1056 register struct ip_moptions *imo; 1057 register struct mbuf **mp; 1058{ 1059 u_char *ttl; 1060 u_char *loop; 1061 struct in_addr *addr; 1062 struct in_ifaddr *ia; 1063 1064 *mp = m_get(M_WAIT, MT_SOOPTS); 1065 1066 switch (optname) { 1067 1068 case IP_MULTICAST_VIF: 1069 if (imo != NULL) 1070 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1071 else 1072 *(mtod(*mp, int *)) = 7890; 1073 (*mp)->m_len = sizeof(int); 1074 return(0); 1075 1076 case IP_MULTICAST_IF: 1077 addr = mtod(*mp, struct in_addr *); 1078 (*mp)->m_len = sizeof(struct in_addr); 1079 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1080 addr->s_addr = INADDR_ANY; 1081 else { 1082 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1083 addr->s_addr = (ia == NULL) ? INADDR_ANY 1084 : IA_SIN(ia)->sin_addr.s_addr; 1085 } 1086 return (0); 1087 1088 case IP_MULTICAST_TTL: 1089 ttl = mtod(*mp, u_char *); 1090 (*mp)->m_len = 1; 1091 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1092 : imo->imo_multicast_ttl; 1093 return (0); 1094 1095 case IP_MULTICAST_LOOP: 1096 loop = mtod(*mp, u_char *); 1097 (*mp)->m_len = 1; 1098 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1099 : imo->imo_multicast_loop; 1100 return (0); 1101 1102 default: 1103 return (EOPNOTSUPP); 1104 } 1105} 1106 1107/* 1108 * Discard the IP multicast options. 1109 */ 1110void 1111ip_freemoptions(imo) 1112 register struct ip_moptions *imo; 1113{ 1114 register int i; 1115 1116 if (imo != NULL) { 1117 for (i = 0; i < imo->imo_num_memberships; ++i) 1118 in_delmulti(imo->imo_membership[i]); 1119 free(imo, M_IPMOPTS); 1120 } 1121} 1122 1123/* 1124 * Routine called from ip_output() to loop back a copy of an IP multicast 1125 * packet to the input queue of a specified interface. Note that this 1126 * calls the output routine of the loopback "driver", but with an interface 1127 * pointer that might NOT be &loif -- easier than replicating that code here. 1128 */ 1129static void 1130ip_mloopback(ifp, m, dst) 1131 struct ifnet *ifp; 1132 register struct mbuf *m; 1133 register struct sockaddr_in *dst; 1134{ 1135 register struct ip *ip; 1136 struct mbuf *copym; 1137 1138 copym = m_copy(m, 0, M_COPYALL); 1139 if (copym != NULL) { 1140 /* 1141 * We don't bother to fragment if the IP length is greater 1142 * than the interface's MTU. Can this possibly matter? 1143 */ 1144 ip = mtod(copym, struct ip *); 1145 ip->ip_len = htons((u_short)ip->ip_len); 1146 ip->ip_off = htons((u_short)ip->ip_off); 1147 ip->ip_sum = 0; 1148 ip->ip_sum = in_cksum(copym, ip->ip_hl << 2); 1149 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1150 } 1151} 1152