ip_output.c revision 14546
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 1.30 1996/02/24 00:17:35 phk Exp $ 35 */ 36 37#include <sys/param.h> 38#include <sys/queue.h> 39#include <sys/systm.h> 40#include <sys/malloc.h> 41#include <sys/mbuf.h> 42#include <sys/errno.h> 43#include <sys/protosw.h> 44#include <sys/socket.h> 45#include <sys/socketvar.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 /* 95 * It might seem obvious at first glance that one could easily 96 * make a one-behind cache out of this by simply making `iproute' 97 * static and eliminating the bzero() below. However, this turns 98 * out not to work, for two reasons: 99 * 100 * 1) This routine needs to be reentrant. It can be called 101 * recursively from encapsulating network interfaces, and it 102 * is always called recursively from ip_mforward(). 103 * 104 * 2) You turn out not to gain much. There is already a one- 105 * behind cache implemented for the specific case of forwarding, 106 * and sends on a connected socket will use a route associated 107 * with the PCB. The only cases left are sends on unconnected 108 * and raw sockets, and if these cases are really significant, 109 * something is seriously wrong. 110 */ 111 struct route iproute; 112 struct sockaddr_in *dst; 113 struct in_ifaddr *ia; 114 115#ifdef DIAGNOSTIC 116 if ((m->m_flags & M_PKTHDR) == 0) 117 panic("ip_output no HDR"); 118#endif 119 if (opt) { 120 m = ip_insertoptions(m, opt, &len); 121 hlen = len; 122 } 123 ip = mtod(m, struct ip *); 124 /* 125 * Fill in IP header. 126 */ 127 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 128 ip->ip_v = IPVERSION; 129 ip->ip_off &= IP_DF; 130 ip->ip_id = htons(ip_id++); 131 ip->ip_hl = hlen >> 2; 132 ipstat.ips_localout++; 133 } else { 134 hlen = ip->ip_hl << 2; 135 } 136 /* 137 * Route packet. 138 */ 139 if (ro == 0) { 140 ro = &iproute; 141 bzero((caddr_t)ro, sizeof (*ro)); 142 } 143 dst = (struct sockaddr_in *)&ro->ro_dst; 144 /* 145 * If there is a cached route, 146 * check that it is to the same destination 147 * and is still up. If not, free it and try again. 148 */ 149 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 150 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 151 RTFREE(ro->ro_rt); 152 ro->ro_rt = (struct rtentry *)0; 153 } 154 if (ro->ro_rt == 0) { 155 dst->sin_family = AF_INET; 156 dst->sin_len = sizeof(*dst); 157 dst->sin_addr = ip->ip_dst; 158 } 159 /* 160 * If routing to interface only, 161 * short circuit routing lookup. 162 */ 163#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 164#define sintosa(sin) ((struct sockaddr *)(sin)) 165 if (flags & IP_ROUTETOIF) { 166 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 167 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 168 ipstat.ips_noroute++; 169 error = ENETUNREACH; 170 goto bad; 171 } 172 ifp = ia->ia_ifp; 173 ip->ip_ttl = 1; 174 } else { 175 /* 176 * If this is the case, we probably don't want to allocate 177 * a protocol-cloned route since we didn't get one from the 178 * ULP. This lets TCP do its thing, while not burdening 179 * forwarding or ICMP with the overhead of cloning a route. 180 * Of course, we still want to do any cloning requested by 181 * the link layer, as this is probably required in all cases 182 * for correct operation (as it is for ARP). 183 */ 184 if (ro->ro_rt == 0) 185 rtalloc_ign(ro, RTF_PRCLONING); 186 if (ro->ro_rt == 0) { 187 ipstat.ips_noroute++; 188 error = EHOSTUNREACH; 189 goto bad; 190 } 191 ia = ifatoia(ro->ro_rt->rt_ifa); 192 ifp = ro->ro_rt->rt_ifp; 193 ro->ro_rt->rt_use++; 194 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 195 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 196 } 197 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 198 struct in_multi *inm; 199 200 m->m_flags |= M_MCAST; 201 /* 202 * IP destination address is multicast. Make sure "dst" 203 * still points to the address in "ro". (It may have been 204 * changed to point to a gateway address, above.) 205 */ 206 dst = (struct sockaddr_in *)&ro->ro_dst; 207 /* 208 * See if the caller provided any multicast options 209 */ 210 if (imo != NULL) { 211 ip->ip_ttl = imo->imo_multicast_ttl; 212 if (imo->imo_multicast_ifp != NULL) 213 ifp = imo->imo_multicast_ifp; 214 if (imo->imo_multicast_vif != -1) 215 ip->ip_src.s_addr = 216 ip_mcast_src(imo->imo_multicast_vif); 217 } else 218 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 219 /* 220 * Confirm that the outgoing interface supports multicast. 221 */ 222 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 223 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 224 ipstat.ips_noroute++; 225 error = ENETUNREACH; 226 goto bad; 227 } 228 } 229 /* 230 * If source address not specified yet, use address 231 * of outgoing interface. 232 */ 233 if (ip->ip_src.s_addr == INADDR_ANY) { 234 register struct in_ifaddr *ia; 235 236 for (ia = in_ifaddr; ia; ia = ia->ia_next) 237 if (ia->ia_ifp == ifp) { 238 ip->ip_src = IA_SIN(ia)->sin_addr; 239 break; 240 } 241 } 242 243 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 244 if (inm != NULL && 245 (imo == NULL || imo->imo_multicast_loop)) { 246 /* 247 * If we belong to the destination multicast group 248 * on the outgoing interface, and the caller did not 249 * forbid loopback, loop back a copy. 250 */ 251 ip_mloopback(ifp, m, dst); 252 } 253 else { 254 /* 255 * If we are acting as a multicast router, perform 256 * multicast forwarding as if the packet had just 257 * arrived on the interface to which we are about 258 * to send. The multicast forwarding function 259 * recursively calls this function, using the 260 * IP_FORWARDING flag to prevent infinite recursion. 261 * 262 * Multicasts that are looped back by ip_mloopback(), 263 * above, will be forwarded by the ip_input() routine, 264 * if necessary. 265 */ 266 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 267 /* 268 * Check if rsvp daemon is running. If not, don't 269 * set ip_moptions. This ensures that the packet 270 * is multicast and not just sent down one link 271 * as prescribed by rsvpd. 272 */ 273 if (!rsvp_on) 274 imo = NULL; 275 if (ip_mforward(ip, ifp, m, imo) != 0) { 276 m_freem(m); 277 goto done; 278 } 279 } 280 } 281 282 /* 283 * Multicasts with a time-to-live of zero may be looped- 284 * back, above, but must not be transmitted on a network. 285 * Also, multicasts addressed to the loopback interface 286 * are not sent -- the above call to ip_mloopback() will 287 * loop back a copy if this host actually belongs to the 288 * destination group on the loopback interface. 289 */ 290 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 291 m_freem(m); 292 goto done; 293 } 294 295 goto sendit; 296 } 297#ifndef notdef 298 /* 299 * If source address not specified yet, use address 300 * of outgoing interface. 301 */ 302 if (ip->ip_src.s_addr == INADDR_ANY) 303 ip->ip_src = IA_SIN(ia)->sin_addr; 304#endif 305 /* 306 * Verify that we have any chance at all of being able to queue 307 * the packet or packet fragments 308 */ 309 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 310 ifp->if_snd.ifq_maxlen) { 311 error = ENOBUFS; 312 goto bad; 313 } 314 315 /* 316 * Look for broadcast address and 317 * and verify user is allowed to send 318 * such a packet. 319 */ 320 if (in_broadcast(dst->sin_addr, ifp)) { 321 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 322 error = EADDRNOTAVAIL; 323 goto bad; 324 } 325 if ((flags & IP_ALLOWBROADCAST) == 0) { 326 error = EACCES; 327 goto bad; 328 } 329 /* don't allow broadcast messages to be fragmented */ 330 if ((u_short)ip->ip_len > ifp->if_mtu) { 331 error = EMSGSIZE; 332 goto bad; 333 } 334 m->m_flags |= M_BCAST; 335 } else 336 m->m_flags &= ~M_BCAST; 337 338sendit: 339 /* 340 * Check with the firewall... 341 */ 342 if (!(*ip_fw_chk_ptr)(m,ip,ifp,1)) { 343 error = EACCES; 344 goto done; 345 } 346 347 /* 348 * If small enough for interface, can just send directly. 349 */ 350 if ((u_short)ip->ip_len <= ifp->if_mtu) { 351 ip->ip_len = htons((u_short)ip->ip_len); 352 ip->ip_off = htons((u_short)ip->ip_off); 353 ip->ip_sum = 0; 354 ip->ip_sum = in_cksum(m, hlen); 355 error = (*ifp->if_output)(ifp, m, 356 (struct sockaddr *)dst, ro->ro_rt); 357 goto done; 358 } 359 /* 360 * Too large for interface; fragment if possible. 361 * Must be able to put at least 8 bytes per fragment. 362 */ 363 if (ip->ip_off & IP_DF) { 364 error = EMSGSIZE; 365#if 1 366 /* 367 * This case can happen if the user changed the MTU 368 * of an interface after enabling IP on it. Because 369 * most netifs don't keep track of routes pointing to 370 * them, there is no way for one to update all its 371 * routes when the MTU is changed. 372 */ 373 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 374 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 375 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 376 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 377 } 378#endif 379 ipstat.ips_cantfrag++; 380 goto bad; 381 } 382 len = (ifp->if_mtu - hlen) &~ 7; 383 if (len < 8) { 384 error = EMSGSIZE; 385 goto bad; 386 } 387 388 { 389 int mhlen, firstlen = len; 390 struct mbuf **mnext = &m->m_nextpkt; 391 392 /* 393 * Loop through length of segment after first fragment, 394 * make new header and copy data of each part and link onto chain. 395 */ 396 m0 = m; 397 mhlen = sizeof (struct ip); 398 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 399 MGETHDR(m, M_DONTWAIT, MT_HEADER); 400 if (m == 0) { 401 error = ENOBUFS; 402 ipstat.ips_odropped++; 403 goto sendorfree; 404 } 405 m->m_data += max_linkhdr; 406 mhip = mtod(m, struct ip *); 407 *mhip = *ip; 408 if (hlen > sizeof (struct ip)) { 409 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 410 mhip->ip_hl = mhlen >> 2; 411 } 412 m->m_len = mhlen; 413 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 414 if (ip->ip_off & IP_MF) 415 mhip->ip_off |= IP_MF; 416 if (off + len >= (u_short)ip->ip_len) 417 len = (u_short)ip->ip_len - off; 418 else 419 mhip->ip_off |= IP_MF; 420 mhip->ip_len = htons((u_short)(len + mhlen)); 421 m->m_next = m_copy(m0, off, len); 422 if (m->m_next == 0) { 423 (void) m_free(m); 424 error = ENOBUFS; /* ??? */ 425 ipstat.ips_odropped++; 426 goto sendorfree; 427 } 428 m->m_pkthdr.len = mhlen + len; 429 m->m_pkthdr.rcvif = (struct ifnet *)0; 430 mhip->ip_off = htons((u_short)mhip->ip_off); 431 mhip->ip_sum = 0; 432 mhip->ip_sum = in_cksum(m, mhlen); 433 *mnext = m; 434 mnext = &m->m_nextpkt; 435 ipstat.ips_ofragments++; 436 } 437 /* 438 * Update first fragment by trimming what's been copied out 439 * and updating header, then send each fragment (in order). 440 */ 441 m = m0; 442 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 443 m->m_pkthdr.len = hlen + firstlen; 444 ip->ip_len = htons((u_short)m->m_pkthdr.len); 445 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 446 ip->ip_sum = 0; 447 ip->ip_sum = in_cksum(m, hlen); 448sendorfree: 449 for (m = m0; m; m = m0) { 450 m0 = m->m_nextpkt; 451 m->m_nextpkt = 0; 452 if (error == 0) 453 error = (*ifp->if_output)(ifp, m, 454 (struct sockaddr *)dst, ro->ro_rt); 455 else 456 m_freem(m); 457 } 458 459 if (error == 0) 460 ipstat.ips_fragmented++; 461 } 462done: 463 if (ro == &iproute && (flags & IP_ROUTETOIF) == 0 && ro->ro_rt) 464 RTFREE(ro->ro_rt); 465 466 return (error); 467bad: 468 m_freem(m0); 469 goto done; 470} 471 472/* 473 * Insert IP options into preformed packet. 474 * Adjust IP destination as required for IP source routing, 475 * as indicated by a non-zero in_addr at the start of the options. 476 */ 477static struct mbuf * 478ip_insertoptions(m, opt, phlen) 479 register struct mbuf *m; 480 struct mbuf *opt; 481 int *phlen; 482{ 483 register struct ipoption *p = mtod(opt, struct ipoption *); 484 struct mbuf *n; 485 register struct ip *ip = mtod(m, struct ip *); 486 unsigned optlen; 487 488 optlen = opt->m_len - sizeof(p->ipopt_dst); 489 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 490 return (m); /* XXX should fail */ 491 if (p->ipopt_dst.s_addr) 492 ip->ip_dst = p->ipopt_dst; 493 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 494 MGETHDR(n, M_DONTWAIT, MT_HEADER); 495 if (n == 0) 496 return (m); 497 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 498 m->m_len -= sizeof(struct ip); 499 m->m_data += sizeof(struct ip); 500 n->m_next = m; 501 m = n; 502 m->m_len = optlen + sizeof(struct ip); 503 m->m_data += max_linkhdr; 504 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 505 } else { 506 m->m_data -= optlen; 507 m->m_len += optlen; 508 m->m_pkthdr.len += optlen; 509 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 510 } 511 ip = mtod(m, struct ip *); 512 (void)memcpy(ip + 1, p->ipopt_list, (unsigned)optlen); 513 *phlen = sizeof(struct ip) + optlen; 514 ip->ip_len += optlen; 515 return (m); 516} 517 518/* 519 * Copy options from ip to jp, 520 * omitting those not copied during fragmentation. 521 */ 522static int 523ip_optcopy(ip, jp) 524 struct ip *ip, *jp; 525{ 526 register u_char *cp, *dp; 527 int opt, optlen, cnt; 528 529 cp = (u_char *)(ip + 1); 530 dp = (u_char *)(jp + 1); 531 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 532 for (; cnt > 0; cnt -= optlen, cp += optlen) { 533 opt = cp[0]; 534 if (opt == IPOPT_EOL) 535 break; 536 if (opt == IPOPT_NOP) { 537 /* Preserve for IP mcast tunnel's LSRR alignment. */ 538 *dp++ = IPOPT_NOP; 539 optlen = 1; 540 continue; 541 } else 542 optlen = cp[IPOPT_OLEN]; 543 /* bogus lengths should have been caught by ip_dooptions */ 544 if (optlen > cnt) 545 optlen = cnt; 546 if (IPOPT_COPIED(opt)) { 547 (void)memcpy(dp, cp, (unsigned)optlen); 548 dp += optlen; 549 } 550 } 551 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 552 *dp++ = IPOPT_EOL; 553 return (optlen); 554} 555 556/* 557 * IP socket option processing. 558 */ 559int 560ip_ctloutput(op, so, level, optname, mp) 561 int op; 562 struct socket *so; 563 int level, optname; 564 struct mbuf **mp; 565{ 566 register struct inpcb *inp = sotoinpcb(so); 567 register struct mbuf *m = *mp; 568 register int optval = 0; 569 int error = 0; 570 571 if (level != IPPROTO_IP) { 572 error = EINVAL; 573 if (op == PRCO_SETOPT && *mp) 574 (void) m_free(*mp); 575 } else switch (op) { 576 577 case PRCO_SETOPT: 578 switch (optname) { 579 case IP_OPTIONS: 580#ifdef notyet 581 case IP_RETOPTS: 582 return (ip_pcbopts(optname, &inp->inp_options, m)); 583#else 584 return (ip_pcbopts(&inp->inp_options, m)); 585#endif 586 587 case IP_TOS: 588 case IP_TTL: 589 case IP_RECVOPTS: 590 case IP_RECVRETOPTS: 591 case IP_RECVDSTADDR: 592 if (m == 0 || m->m_len != sizeof(int)) 593 error = EINVAL; 594 else { 595 optval = *mtod(m, int *); 596 switch (optname) { 597 598 case IP_TOS: 599 inp->inp_ip.ip_tos = optval; 600 break; 601 602 case IP_TTL: 603 inp->inp_ip.ip_ttl = optval; 604 break; 605#define OPTSET(bit) \ 606 if (optval) \ 607 inp->inp_flags |= bit; \ 608 else \ 609 inp->inp_flags &= ~bit; 610 611 case IP_RECVOPTS: 612 OPTSET(INP_RECVOPTS); 613 break; 614 615 case IP_RECVRETOPTS: 616 OPTSET(INP_RECVRETOPTS); 617 break; 618 619 case IP_RECVDSTADDR: 620 OPTSET(INP_RECVDSTADDR); 621 break; 622 } 623 } 624 break; 625#undef OPTSET 626 627 case IP_MULTICAST_IF: 628 case IP_MULTICAST_VIF: 629 case IP_MULTICAST_TTL: 630 case IP_MULTICAST_LOOP: 631 case IP_ADD_MEMBERSHIP: 632 case IP_DROP_MEMBERSHIP: 633 error = ip_setmoptions(optname, &inp->inp_moptions, m); 634 break; 635 636 case IP_PORTRANGE: 637 if (m == 0 || m->m_len != sizeof(int)) 638 error = EINVAL; 639 else { 640 optval = *mtod(m, int *); 641 642 switch (optval) { 643 644 case IP_PORTRANGE_DEFAULT: 645 inp->inp_flags &= ~(INP_LOWPORT); 646 inp->inp_flags &= ~(INP_HIGHPORT); 647 break; 648 649 case IP_PORTRANGE_HIGH: 650 inp->inp_flags &= ~(INP_LOWPORT); 651 inp->inp_flags |= INP_HIGHPORT; 652 break; 653 654 case IP_PORTRANGE_LOW: 655 inp->inp_flags &= ~(INP_HIGHPORT); 656 inp->inp_flags |= INP_LOWPORT; 657 break; 658 659 default: 660 error = EINVAL; 661 break; 662 } 663 } 664 665 default: 666 error = ENOPROTOOPT; 667 break; 668 } 669 if (m) 670 (void)m_free(m); 671 break; 672 673 case PRCO_GETOPT: 674 switch (optname) { 675 case IP_OPTIONS: 676 case IP_RETOPTS: 677 *mp = m = m_get(M_WAIT, MT_SOOPTS); 678 if (inp->inp_options) { 679 m->m_len = inp->inp_options->m_len; 680 (void)memcpy(mtod(m, void *), 681 mtod(inp->inp_options, void *), (unsigned)m->m_len); 682 } else 683 m->m_len = 0; 684 break; 685 686 case IP_TOS: 687 case IP_TTL: 688 case IP_RECVOPTS: 689 case IP_RECVRETOPTS: 690 case IP_RECVDSTADDR: 691 *mp = m = m_get(M_WAIT, MT_SOOPTS); 692 m->m_len = sizeof(int); 693 switch (optname) { 694 695 case IP_TOS: 696 optval = inp->inp_ip.ip_tos; 697 break; 698 699 case IP_TTL: 700 optval = inp->inp_ip.ip_ttl; 701 break; 702 703#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 704 705 case IP_RECVOPTS: 706 optval = OPTBIT(INP_RECVOPTS); 707 break; 708 709 case IP_RECVRETOPTS: 710 optval = OPTBIT(INP_RECVRETOPTS); 711 break; 712 713 case IP_RECVDSTADDR: 714 optval = OPTBIT(INP_RECVDSTADDR); 715 break; 716 } 717 *mtod(m, int *) = optval; 718 break; 719 720 case IP_MULTICAST_IF: 721 case IP_MULTICAST_VIF: 722 case IP_MULTICAST_TTL: 723 case IP_MULTICAST_LOOP: 724 case IP_ADD_MEMBERSHIP: 725 case IP_DROP_MEMBERSHIP: 726 error = ip_getmoptions(optname, inp->inp_moptions, mp); 727 break; 728 729 case IP_PORTRANGE: 730 *mp = m = m_get(M_WAIT, MT_SOOPTS); 731 m->m_len = sizeof(int); 732 733 if (inp->inp_flags & INP_HIGHPORT) 734 optval = IP_PORTRANGE_HIGH; 735 else if (inp->inp_flags & INP_LOWPORT) 736 optval = IP_PORTRANGE_LOW; 737 else 738 optval = 0; 739 740 *mtod(m, int *) = optval; 741 break; 742 743 default: 744 error = ENOPROTOOPT; 745 break; 746 } 747 break; 748 } 749 return (error); 750} 751 752/* 753 * Set up IP options in pcb for insertion in output packets. 754 * Store in mbuf with pointer in pcbopt, adding pseudo-option 755 * with destination address if source routed. 756 */ 757static int 758#ifdef notyet 759ip_pcbopts(optname, pcbopt, m) 760 int optname; 761#else 762ip_pcbopts(pcbopt, m) 763#endif 764 struct mbuf **pcbopt; 765 register struct mbuf *m; 766{ 767 register cnt, optlen; 768 register u_char *cp; 769 u_char opt; 770 771 /* turn off any old options */ 772 if (*pcbopt) 773 (void)m_free(*pcbopt); 774 *pcbopt = 0; 775 if (m == (struct mbuf *)0 || m->m_len == 0) { 776 /* 777 * Only turning off any previous options. 778 */ 779 if (m) 780 (void)m_free(m); 781 return (0); 782 } 783 784#ifndef vax 785 if (m->m_len % sizeof(long)) 786 goto bad; 787#endif 788 /* 789 * IP first-hop destination address will be stored before 790 * actual options; move other options back 791 * and clear it when none present. 792 */ 793 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 794 goto bad; 795 cnt = m->m_len; 796 m->m_len += sizeof(struct in_addr); 797 cp = mtod(m, u_char *) + sizeof(struct in_addr); 798 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 799 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 800 801 for (; cnt > 0; cnt -= optlen, cp += optlen) { 802 opt = cp[IPOPT_OPTVAL]; 803 if (opt == IPOPT_EOL) 804 break; 805 if (opt == IPOPT_NOP) 806 optlen = 1; 807 else { 808 optlen = cp[IPOPT_OLEN]; 809 if (optlen <= IPOPT_OLEN || optlen > cnt) 810 goto bad; 811 } 812 switch (opt) { 813 814 default: 815 break; 816 817 case IPOPT_LSRR: 818 case IPOPT_SSRR: 819 /* 820 * user process specifies route as: 821 * ->A->B->C->D 822 * D must be our final destination (but we can't 823 * check that since we may not have connected yet). 824 * A is first hop destination, which doesn't appear in 825 * actual IP option, but is stored before the options. 826 */ 827 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 828 goto bad; 829 m->m_len -= sizeof(struct in_addr); 830 cnt -= sizeof(struct in_addr); 831 optlen -= sizeof(struct in_addr); 832 cp[IPOPT_OLEN] = optlen; 833 /* 834 * Move first hop before start of options. 835 */ 836 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 837 sizeof(struct in_addr)); 838 /* 839 * Then copy rest of options back 840 * to close up the deleted entry. 841 */ 842 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 843 sizeof(struct in_addr)), 844 (caddr_t)&cp[IPOPT_OFFSET+1], 845 (unsigned)cnt + sizeof(struct in_addr)); 846 break; 847 } 848 } 849 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 850 goto bad; 851 *pcbopt = m; 852 return (0); 853 854bad: 855 (void)m_free(m); 856 return (EINVAL); 857} 858 859/* 860 * Set the IP multicast options in response to user setsockopt(). 861 */ 862static int 863ip_setmoptions(optname, imop, m) 864 int optname; 865 struct ip_moptions **imop; 866 struct mbuf *m; 867{ 868 register int error = 0; 869 u_char loop; 870 register int i; 871 struct in_addr addr; 872 register struct ip_mreq *mreq; 873 register struct ifnet *ifp; 874 register struct ip_moptions *imo = *imop; 875 struct route ro; 876 register struct sockaddr_in *dst; 877 int s; 878 879 if (imo == NULL) { 880 /* 881 * No multicast option buffer attached to the pcb; 882 * allocate one and initialize to default values. 883 */ 884 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 885 M_WAITOK); 886 887 if (imo == NULL) 888 return (ENOBUFS); 889 *imop = imo; 890 imo->imo_multicast_ifp = NULL; 891 imo->imo_multicast_vif = -1; 892 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 893 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 894 imo->imo_num_memberships = 0; 895 } 896 897 switch (optname) { 898 /* store an index number for the vif you wanna use in the send */ 899 case IP_MULTICAST_VIF: 900 if (!legal_vif_num) { 901 error = EOPNOTSUPP; 902 break; 903 } 904 if (m == NULL || m->m_len != sizeof(int)) { 905 error = EINVAL; 906 break; 907 } 908 i = *(mtod(m, int *)); 909 if (!legal_vif_num(i) && (i != -1)) { 910 error = EINVAL; 911 break; 912 } 913 imo->imo_multicast_vif = i; 914 break; 915 916 case IP_MULTICAST_IF: 917 /* 918 * Select the interface for outgoing multicast packets. 919 */ 920 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 921 error = EINVAL; 922 break; 923 } 924 addr = *(mtod(m, struct in_addr *)); 925 /* 926 * INADDR_ANY is used to remove a previous selection. 927 * When no interface is selected, a default one is 928 * chosen every time a multicast packet is sent. 929 */ 930 if (addr.s_addr == INADDR_ANY) { 931 imo->imo_multicast_ifp = NULL; 932 break; 933 } 934 /* 935 * The selected interface is identified by its local 936 * IP address. Find the interface and confirm that 937 * it supports multicasting. 938 */ 939 s = splimp(); 940 INADDR_TO_IFP(addr, ifp); 941 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 942 error = EADDRNOTAVAIL; 943 break; 944 } 945 imo->imo_multicast_ifp = ifp; 946 splx(s); 947 break; 948 949 case IP_MULTICAST_TTL: 950 /* 951 * Set the IP time-to-live for outgoing multicast packets. 952 */ 953 if (m == NULL || m->m_len != 1) { 954 error = EINVAL; 955 break; 956 } 957 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 958 break; 959 960 case IP_MULTICAST_LOOP: 961 /* 962 * Set the loopback flag for outgoing multicast packets. 963 * Must be zero or one. 964 */ 965 if (m == NULL || m->m_len != 1 || 966 (loop = *(mtod(m, u_char *))) > 1) { 967 error = EINVAL; 968 break; 969 } 970 imo->imo_multicast_loop = loop; 971 break; 972 973 case IP_ADD_MEMBERSHIP: 974 /* 975 * Add a multicast group membership. 976 * Group must be a valid IP multicast address. 977 */ 978 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 979 error = EINVAL; 980 break; 981 } 982 mreq = mtod(m, struct ip_mreq *); 983 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 984 error = EINVAL; 985 break; 986 } 987 s = splimp(); 988 /* 989 * If no interface address was provided, use the interface of 990 * the route to the given multicast address. 991 */ 992 if (mreq->imr_interface.s_addr == INADDR_ANY) { 993 bzero((caddr_t)&ro, sizeof(ro)); 994 dst = (struct sockaddr_in *)&ro.ro_dst; 995 dst->sin_len = sizeof(*dst); 996 dst->sin_family = AF_INET; 997 dst->sin_addr = mreq->imr_multiaddr; 998 rtalloc(&ro); 999 if (ro.ro_rt == NULL) { 1000 error = EADDRNOTAVAIL; 1001 splx(s); 1002 break; 1003 } 1004 ifp = ro.ro_rt->rt_ifp; 1005 rtfree(ro.ro_rt); 1006 } 1007 else { 1008 INADDR_TO_IFP(mreq->imr_interface, ifp); 1009 } 1010 1011 /* 1012 * See if we found an interface, and confirm that it 1013 * supports multicast. 1014 */ 1015 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1016 error = EADDRNOTAVAIL; 1017 splx(s); 1018 break; 1019 } 1020 /* 1021 * See if the membership already exists or if all the 1022 * membership slots are full. 1023 */ 1024 for (i = 0; i < imo->imo_num_memberships; ++i) { 1025 if (imo->imo_membership[i]->inm_ifp == ifp && 1026 imo->imo_membership[i]->inm_addr.s_addr 1027 == mreq->imr_multiaddr.s_addr) 1028 break; 1029 } 1030 if (i < imo->imo_num_memberships) { 1031 error = EADDRINUSE; 1032 splx(s); 1033 break; 1034 } 1035 if (i == IP_MAX_MEMBERSHIPS) { 1036 error = ETOOMANYREFS; 1037 splx(s); 1038 break; 1039 } 1040 /* 1041 * Everything looks good; add a new record to the multicast 1042 * address list for the given interface. 1043 */ 1044 if ((imo->imo_membership[i] = 1045 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1046 error = ENOBUFS; 1047 splx(s); 1048 break; 1049 } 1050 ++imo->imo_num_memberships; 1051 splx(s); 1052 break; 1053 1054 case IP_DROP_MEMBERSHIP: 1055 /* 1056 * Drop a multicast group membership. 1057 * Group must be a valid IP multicast address. 1058 */ 1059 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1060 error = EINVAL; 1061 break; 1062 } 1063 mreq = mtod(m, struct ip_mreq *); 1064 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1065 error = EINVAL; 1066 break; 1067 } 1068 1069 s = splimp(); 1070 /* 1071 * If an interface address was specified, get a pointer 1072 * to its ifnet structure. 1073 */ 1074 if (mreq->imr_interface.s_addr == INADDR_ANY) 1075 ifp = NULL; 1076 else { 1077 INADDR_TO_IFP(mreq->imr_interface, ifp); 1078 if (ifp == NULL) { 1079 error = EADDRNOTAVAIL; 1080 splx(s); 1081 break; 1082 } 1083 } 1084 /* 1085 * Find the membership in the membership array. 1086 */ 1087 for (i = 0; i < imo->imo_num_memberships; ++i) { 1088 if ((ifp == NULL || 1089 imo->imo_membership[i]->inm_ifp == ifp) && 1090 imo->imo_membership[i]->inm_addr.s_addr == 1091 mreq->imr_multiaddr.s_addr) 1092 break; 1093 } 1094 if (i == imo->imo_num_memberships) { 1095 error = EADDRNOTAVAIL; 1096 splx(s); 1097 break; 1098 } 1099 /* 1100 * Give up the multicast address record to which the 1101 * membership points. 1102 */ 1103 in_delmulti(imo->imo_membership[i]); 1104 /* 1105 * Remove the gap in the membership array. 1106 */ 1107 for (++i; i < imo->imo_num_memberships; ++i) 1108 imo->imo_membership[i-1] = imo->imo_membership[i]; 1109 --imo->imo_num_memberships; 1110 splx(s); 1111 break; 1112 1113 default: 1114 error = EOPNOTSUPP; 1115 break; 1116 } 1117 1118 /* 1119 * If all options have default values, no need to keep the mbuf. 1120 */ 1121 if (imo->imo_multicast_ifp == NULL && 1122 imo->imo_multicast_vif == -1 && 1123 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1124 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1125 imo->imo_num_memberships == 0) { 1126 free(*imop, M_IPMOPTS); 1127 *imop = NULL; 1128 } 1129 1130 return (error); 1131} 1132 1133/* 1134 * Return the IP multicast options in response to user getsockopt(). 1135 */ 1136static int 1137ip_getmoptions(optname, imo, mp) 1138 int optname; 1139 register struct ip_moptions *imo; 1140 register struct mbuf **mp; 1141{ 1142 u_char *ttl; 1143 u_char *loop; 1144 struct in_addr *addr; 1145 struct in_ifaddr *ia; 1146 1147 *mp = m_get(M_WAIT, MT_SOOPTS); 1148 1149 switch (optname) { 1150 1151 case IP_MULTICAST_VIF: 1152 if (imo != NULL) 1153 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1154 else 1155 *(mtod(*mp, int *)) = -1; 1156 (*mp)->m_len = sizeof(int); 1157 return(0); 1158 1159 case IP_MULTICAST_IF: 1160 addr = mtod(*mp, struct in_addr *); 1161 (*mp)->m_len = sizeof(struct in_addr); 1162 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1163 addr->s_addr = INADDR_ANY; 1164 else { 1165 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1166 addr->s_addr = (ia == NULL) ? INADDR_ANY 1167 : IA_SIN(ia)->sin_addr.s_addr; 1168 } 1169 return (0); 1170 1171 case IP_MULTICAST_TTL: 1172 ttl = mtod(*mp, u_char *); 1173 (*mp)->m_len = 1; 1174 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1175 : imo->imo_multicast_ttl; 1176 return (0); 1177 1178 case IP_MULTICAST_LOOP: 1179 loop = mtod(*mp, u_char *); 1180 (*mp)->m_len = 1; 1181 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1182 : imo->imo_multicast_loop; 1183 return (0); 1184 1185 default: 1186 return (EOPNOTSUPP); 1187 } 1188} 1189 1190/* 1191 * Discard the IP multicast options. 1192 */ 1193void 1194ip_freemoptions(imo) 1195 register struct ip_moptions *imo; 1196{ 1197 register int i; 1198 1199 if (imo != NULL) { 1200 for (i = 0; i < imo->imo_num_memberships; ++i) 1201 in_delmulti(imo->imo_membership[i]); 1202 free(imo, M_IPMOPTS); 1203 } 1204} 1205 1206/* 1207 * Routine called from ip_output() to loop back a copy of an IP multicast 1208 * packet to the input queue of a specified interface. Note that this 1209 * calls the output routine of the loopback "driver", but with an interface 1210 * pointer that might NOT be a loopback interface -- evil, but easier than 1211 * replicating that code here. 1212 */ 1213static void 1214ip_mloopback(ifp, m, dst) 1215 struct ifnet *ifp; 1216 register struct mbuf *m; 1217 register struct sockaddr_in *dst; 1218{ 1219 register struct ip *ip; 1220 struct mbuf *copym; 1221 1222 copym = m_copy(m, 0, M_COPYALL); 1223 if (copym != NULL) { 1224 /* 1225 * We don't bother to fragment if the IP length is greater 1226 * than the interface's MTU. Can this possibly matter? 1227 */ 1228 ip = mtod(copym, struct ip *); 1229 ip->ip_len = htons((u_short)ip->ip_len); 1230 ip->ip_off = htons((u_short)ip->ip_off); 1231 ip->ip_sum = 0; 1232 ip->ip_sum = in_cksum(copym, ip->ip_hl << 2); 1233 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1234 } 1235} 1236