ip_output.c revision 15295
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 1.34 1996/04/03 13:52:20 phk Exp $ 35 */ 36 37#define _IP_VHL 38 39#include <sys/param.h> 40#include <sys/queue.h> 41#include <sys/systm.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/errno.h> 45#include <sys/protosw.h> 46#include <sys/socket.h> 47#include <sys/socketvar.h> 48 49#include <net/if.h> 50#include <net/route.h> 51 52#include <netinet/in.h> 53#include <netinet/in_systm.h> 54#include <netinet/ip.h> 55#include <netinet/in_pcb.h> 56#include <netinet/in_var.h> 57#include <netinet/ip_var.h> 58 59#ifdef vax 60#include <machine/mtpr.h> 61#endif 62#include <machine/in_cksum.h> 63 64u_short ip_id; 65 66static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 67static void ip_mloopback 68 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *)); 69static int ip_getmoptions 70 __P((int, struct ip_moptions *, struct mbuf **)); 71static int ip_optcopy __P((struct ip *, struct ip *)); 72static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 73static int ip_setmoptions 74 __P((int, struct ip_moptions **, struct mbuf *)); 75 76/* 77 * IP output. The packet in mbuf chain m contains a skeletal IP 78 * header (with len, off, ttl, proto, tos, src, dst). 79 * The mbuf chain containing the packet will be freed. 80 * The mbuf opt, if present, will not be freed. 81 */ 82int 83ip_output(m0, opt, ro, flags, imo) 84 struct mbuf *m0; 85 struct mbuf *opt; 86 struct route *ro; 87 int flags; 88 struct ip_moptions *imo; 89{ 90 struct ip *ip, *mhip; 91 struct ifnet *ifp; 92 struct mbuf *m = m0; 93 int hlen = sizeof (struct ip); 94 int len, off, error = 0; 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 if (!ro) 102 panic("ip_output no route, proto = %d", 103 mtod(m, struct ip *)->ip_p); 104#endif 105 if (opt) { 106 m = ip_insertoptions(m, opt, &len); 107 hlen = len; 108 } 109 ip = mtod(m, struct ip *); 110 /* 111 * Fill in IP header. 112 */ 113 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 114 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 115 ip->ip_off &= IP_DF; 116 ip->ip_id = htons(ip_id++); 117 ipstat.ips_localout++; 118 } else { 119 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 120 } 121 122 dst = (struct sockaddr_in *)&ro->ro_dst; 123 /* 124 * If there is a cached route, 125 * check that it is to the same destination 126 * and is still up. If not, free it and try again. 127 */ 128 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 129 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 130 RTFREE(ro->ro_rt); 131 ro->ro_rt = (struct rtentry *)0; 132 } 133 if (ro->ro_rt == 0) { 134 dst->sin_family = AF_INET; 135 dst->sin_len = sizeof(*dst); 136 dst->sin_addr = ip->ip_dst; 137 } 138 /* 139 * If routing to interface only, 140 * short circuit routing lookup. 141 */ 142#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 143#define sintosa(sin) ((struct sockaddr *)(sin)) 144 if (flags & IP_ROUTETOIF) { 145 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 146 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 147 ipstat.ips_noroute++; 148 error = ENETUNREACH; 149 goto bad; 150 } 151 ifp = ia->ia_ifp; 152 ip->ip_ttl = 1; 153 } else { 154 /* 155 * If this is the case, we probably don't want to allocate 156 * a protocol-cloned route since we didn't get one from the 157 * ULP. This lets TCP do its thing, while not burdening 158 * forwarding or ICMP with the overhead of cloning a route. 159 * Of course, we still want to do any cloning requested by 160 * the link layer, as this is probably required in all cases 161 * for correct operation (as it is for ARP). 162 */ 163 if (ro->ro_rt == 0) 164 rtalloc_ign(ro, RTF_PRCLONING); 165 if (ro->ro_rt == 0) { 166 ipstat.ips_noroute++; 167 error = EHOSTUNREACH; 168 goto bad; 169 } 170 ia = ifatoia(ro->ro_rt->rt_ifa); 171 ifp = ro->ro_rt->rt_ifp; 172 ro->ro_rt->rt_use++; 173 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 174 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 175 } 176 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 177 struct in_multi *inm; 178 179 m->m_flags |= M_MCAST; 180 /* 181 * IP destination address is multicast. Make sure "dst" 182 * still points to the address in "ro". (It may have been 183 * changed to point to a gateway address, above.) 184 */ 185 dst = (struct sockaddr_in *)&ro->ro_dst; 186 /* 187 * See if the caller provided any multicast options 188 */ 189 if (imo != NULL) { 190 ip->ip_ttl = imo->imo_multicast_ttl; 191 if (imo->imo_multicast_ifp != NULL) 192 ifp = imo->imo_multicast_ifp; 193 if (imo->imo_multicast_vif != -1) 194 ip->ip_src.s_addr = 195 ip_mcast_src(imo->imo_multicast_vif); 196 } else 197 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 198 /* 199 * Confirm that the outgoing interface supports multicast. 200 */ 201 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 202 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 203 ipstat.ips_noroute++; 204 error = ENETUNREACH; 205 goto bad; 206 } 207 } 208 /* 209 * If source address not specified yet, use address 210 * of outgoing interface. 211 */ 212 if (ip->ip_src.s_addr == INADDR_ANY) { 213 register struct in_ifaddr *ia; 214 215 for (ia = in_ifaddr; ia; ia = ia->ia_next) 216 if (ia->ia_ifp == ifp) { 217 ip->ip_src = IA_SIN(ia)->sin_addr; 218 break; 219 } 220 } 221 222 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 223 if (inm != NULL && 224 (imo == NULL || imo->imo_multicast_loop)) { 225 /* 226 * If we belong to the destination multicast group 227 * on the outgoing interface, and the caller did not 228 * forbid loopback, loop back a copy. 229 */ 230 ip_mloopback(ifp, m, dst); 231 } 232 else { 233 /* 234 * If we are acting as a multicast router, perform 235 * multicast forwarding as if the packet had just 236 * arrived on the interface to which we are about 237 * to send. The multicast forwarding function 238 * recursively calls this function, using the 239 * IP_FORWARDING flag to prevent infinite recursion. 240 * 241 * Multicasts that are looped back by ip_mloopback(), 242 * above, will be forwarded by the ip_input() routine, 243 * if necessary. 244 */ 245 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 246 /* 247 * Check if rsvp daemon is running. If not, don't 248 * set ip_moptions. This ensures that the packet 249 * is multicast and not just sent down one link 250 * as prescribed by rsvpd. 251 */ 252 if (!rsvp_on) 253 imo = NULL; 254 if (ip_mforward(ip, ifp, m, imo) != 0) { 255 m_freem(m); 256 goto done; 257 } 258 } 259 } 260 261 /* 262 * Multicasts with a time-to-live of zero may be looped- 263 * back, above, but must not be transmitted on a network. 264 * Also, multicasts addressed to the loopback interface 265 * are not sent -- the above call to ip_mloopback() will 266 * loop back a copy if this host actually belongs to the 267 * destination group on the loopback interface. 268 */ 269 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 270 m_freem(m); 271 goto done; 272 } 273 274 goto sendit; 275 } 276#ifndef notdef 277 /* 278 * If source address not specified yet, use address 279 * of outgoing interface. 280 */ 281 if (ip->ip_src.s_addr == INADDR_ANY) 282 ip->ip_src = IA_SIN(ia)->sin_addr; 283#endif 284 /* 285 * Verify that we have any chance at all of being able to queue 286 * the packet or packet fragments 287 */ 288 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 289 ifp->if_snd.ifq_maxlen) { 290 error = ENOBUFS; 291 goto bad; 292 } 293 294 /* 295 * Look for broadcast address and 296 * and verify user is allowed to send 297 * such a packet. 298 */ 299 if (in_broadcast(dst->sin_addr, ifp)) { 300 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 301 error = EADDRNOTAVAIL; 302 goto bad; 303 } 304 if ((flags & IP_ALLOWBROADCAST) == 0) { 305 error = EACCES; 306 goto bad; 307 } 308 /* don't allow broadcast messages to be fragmented */ 309 if ((u_short)ip->ip_len > ifp->if_mtu) { 310 error = EMSGSIZE; 311 goto bad; 312 } 313 m->m_flags |= M_BCAST; 314 } else 315 m->m_flags &= ~M_BCAST; 316 317sendit: 318 /* 319 * Check with the firewall... 320 */ 321 if (ip_fw_chk_ptr && !(*ip_fw_chk_ptr)(&ip, hlen, ifp, 1, &m)) { 322 error = EACCES; 323 goto done; 324 } 325 326 /* 327 * If small enough for interface, can just send directly. 328 */ 329 if ((u_short)ip->ip_len <= ifp->if_mtu) { 330 ip->ip_len = htons((u_short)ip->ip_len); 331 ip->ip_off = htons((u_short)ip->ip_off); 332 ip->ip_sum = 0; 333 if (ip->ip_vhl == IP_VHL_BORING) { 334 ip->ip_sum = in_cksum_hdr(ip); 335 } else { 336 ip->ip_sum = in_cksum(m, hlen); 337 } 338 error = (*ifp->if_output)(ifp, m, 339 (struct sockaddr *)dst, ro->ro_rt); 340 goto done; 341 } 342 /* 343 * Too large for interface; fragment if possible. 344 * Must be able to put at least 8 bytes per fragment. 345 */ 346 if (ip->ip_off & IP_DF) { 347 error = EMSGSIZE; 348#if 1 349 /* 350 * This case can happen if the user changed the MTU 351 * of an interface after enabling IP on it. Because 352 * most netifs don't keep track of routes pointing to 353 * them, there is no way for one to update all its 354 * routes when the MTU is changed. 355 */ 356 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 357 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 358 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 359 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 360 } 361#endif 362 ipstat.ips_cantfrag++; 363 goto bad; 364 } 365 len = (ifp->if_mtu - hlen) &~ 7; 366 if (len < 8) { 367 error = EMSGSIZE; 368 goto bad; 369 } 370 371 { 372 int mhlen, firstlen = len; 373 struct mbuf **mnext = &m->m_nextpkt; 374 375 /* 376 * Loop through length of segment after first fragment, 377 * make new header and copy data of each part and link onto chain. 378 */ 379 m0 = m; 380 mhlen = sizeof (struct ip); 381 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 382 MGETHDR(m, M_DONTWAIT, MT_HEADER); 383 if (m == 0) { 384 error = ENOBUFS; 385 ipstat.ips_odropped++; 386 goto sendorfree; 387 } 388 m->m_data += max_linkhdr; 389 mhip = mtod(m, struct ip *); 390 *mhip = *ip; 391 if (hlen > sizeof (struct ip)) { 392 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 393 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 394 } 395 m->m_len = mhlen; 396 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 397 if (ip->ip_off & IP_MF) 398 mhip->ip_off |= IP_MF; 399 if (off + len >= (u_short)ip->ip_len) 400 len = (u_short)ip->ip_len - off; 401 else 402 mhip->ip_off |= IP_MF; 403 mhip->ip_len = htons((u_short)(len + mhlen)); 404 m->m_next = m_copy(m0, off, len); 405 if (m->m_next == 0) { 406 (void) m_free(m); 407 error = ENOBUFS; /* ??? */ 408 ipstat.ips_odropped++; 409 goto sendorfree; 410 } 411 m->m_pkthdr.len = mhlen + len; 412 m->m_pkthdr.rcvif = (struct ifnet *)0; 413 mhip->ip_off = htons((u_short)mhip->ip_off); 414 mhip->ip_sum = 0; 415 if (mhip->ip_vhl == IP_VHL_BORING) { 416 mhip->ip_sum = in_cksum_hdr(ip); 417 } else { 418 mhip->ip_sum = in_cksum(m, mhlen); 419 } 420 *mnext = m; 421 mnext = &m->m_nextpkt; 422 ipstat.ips_ofragments++; 423 } 424 /* 425 * Update first fragment by trimming what's been copied out 426 * and updating header, then send each fragment (in order). 427 */ 428 m = m0; 429 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 430 m->m_pkthdr.len = hlen + firstlen; 431 ip->ip_len = htons((u_short)m->m_pkthdr.len); 432 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 433 ip->ip_sum = 0; 434 if (ip->ip_vhl == IP_VHL_BORING) { 435 ip->ip_sum = in_cksum_hdr(ip); 436 } else { 437 ip->ip_sum = in_cksum(m, hlen); 438 } 439sendorfree: 440 for (m = m0; m; m = m0) { 441 m0 = m->m_nextpkt; 442 m->m_nextpkt = 0; 443 if (error == 0) 444 error = (*ifp->if_output)(ifp, m, 445 (struct sockaddr *)dst, ro->ro_rt); 446 else 447 m_freem(m); 448 } 449 450 if (error == 0) 451 ipstat.ips_fragmented++; 452 } 453done: 454 return (error); 455bad: 456 m_freem(m0); 457 goto done; 458} 459 460/* 461 * Insert IP options into preformed packet. 462 * Adjust IP destination as required for IP source routing, 463 * as indicated by a non-zero in_addr at the start of the options. 464 * 465 * XXX This routine assumes that the packet has no options in place. 466 */ 467static struct mbuf * 468ip_insertoptions(m, opt, phlen) 469 register struct mbuf *m; 470 struct mbuf *opt; 471 int *phlen; 472{ 473 register struct ipoption *p = mtod(opt, struct ipoption *); 474 struct mbuf *n; 475 register struct ip *ip = mtod(m, struct ip *); 476 unsigned optlen; 477 478 optlen = opt->m_len - sizeof(p->ipopt_dst); 479 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 480 return (m); /* XXX should fail */ 481 if (p->ipopt_dst.s_addr) 482 ip->ip_dst = p->ipopt_dst; 483 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 484 MGETHDR(n, M_DONTWAIT, MT_HEADER); 485 if (n == 0) 486 return (m); 487 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 488 m->m_len -= sizeof(struct ip); 489 m->m_data += sizeof(struct ip); 490 n->m_next = m; 491 m = n; 492 m->m_len = optlen + sizeof(struct ip); 493 m->m_data += max_linkhdr; 494 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 495 } else { 496 m->m_data -= optlen; 497 m->m_len += optlen; 498 m->m_pkthdr.len += optlen; 499 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 500 } 501 ip = mtod(m, struct ip *); 502 (void)memcpy(ip + 1, p->ipopt_list, (unsigned)optlen); 503 *phlen = sizeof(struct ip) + optlen; 504 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 505 ip->ip_len += optlen; 506 return (m); 507} 508 509/* 510 * Copy options from ip to jp, 511 * omitting those not copied during fragmentation. 512 */ 513static int 514ip_optcopy(ip, jp) 515 struct ip *ip, *jp; 516{ 517 register u_char *cp, *dp; 518 int opt, optlen, cnt; 519 520 cp = (u_char *)(ip + 1); 521 dp = (u_char *)(jp + 1); 522 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 523 for (; cnt > 0; cnt -= optlen, cp += optlen) { 524 opt = cp[0]; 525 if (opt == IPOPT_EOL) 526 break; 527 if (opt == IPOPT_NOP) { 528 /* Preserve for IP mcast tunnel's LSRR alignment. */ 529 *dp++ = IPOPT_NOP; 530 optlen = 1; 531 continue; 532 } else 533 optlen = cp[IPOPT_OLEN]; 534 /* bogus lengths should have been caught by ip_dooptions */ 535 if (optlen > cnt) 536 optlen = cnt; 537 if (IPOPT_COPIED(opt)) { 538 (void)memcpy(dp, cp, (unsigned)optlen); 539 dp += optlen; 540 } 541 } 542 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 543 *dp++ = IPOPT_EOL; 544 return (optlen); 545} 546 547/* 548 * IP socket option processing. 549 */ 550int 551ip_ctloutput(op, so, level, optname, mp) 552 int op; 553 struct socket *so; 554 int level, optname; 555 struct mbuf **mp; 556{ 557 register struct inpcb *inp = sotoinpcb(so); 558 register struct mbuf *m = *mp; 559 register int optval = 0; 560 int error = 0; 561 562 if (level != IPPROTO_IP) { 563 error = EINVAL; 564 if (op == PRCO_SETOPT && *mp) 565 (void) m_free(*mp); 566 } else switch (op) { 567 568 case PRCO_SETOPT: 569 switch (optname) { 570 case IP_OPTIONS: 571#ifdef notyet 572 case IP_RETOPTS: 573 return (ip_pcbopts(optname, &inp->inp_options, m)); 574#else 575 return (ip_pcbopts(&inp->inp_options, m)); 576#endif 577 578 case IP_TOS: 579 case IP_TTL: 580 case IP_RECVOPTS: 581 case IP_RECVRETOPTS: 582 case IP_RECVDSTADDR: 583 if (m == 0 || m->m_len != sizeof(int)) 584 error = EINVAL; 585 else { 586 optval = *mtod(m, int *); 587 switch (optname) { 588 589 case IP_TOS: 590 inp->inp_ip.ip_tos = optval; 591 break; 592 593 case IP_TTL: 594 inp->inp_ip.ip_ttl = optval; 595 break; 596#define OPTSET(bit) \ 597 if (optval) \ 598 inp->inp_flags |= bit; \ 599 else \ 600 inp->inp_flags &= ~bit; 601 602 case IP_RECVOPTS: 603 OPTSET(INP_RECVOPTS); 604 break; 605 606 case IP_RECVRETOPTS: 607 OPTSET(INP_RECVRETOPTS); 608 break; 609 610 case IP_RECVDSTADDR: 611 OPTSET(INP_RECVDSTADDR); 612 break; 613 } 614 } 615 break; 616#undef OPTSET 617 618 case IP_MULTICAST_IF: 619 case IP_MULTICAST_VIF: 620 case IP_MULTICAST_TTL: 621 case IP_MULTICAST_LOOP: 622 case IP_ADD_MEMBERSHIP: 623 case IP_DROP_MEMBERSHIP: 624 error = ip_setmoptions(optname, &inp->inp_moptions, m); 625 break; 626 627 case IP_PORTRANGE: 628 if (m == 0 || m->m_len != sizeof(int)) 629 error = EINVAL; 630 else { 631 optval = *mtod(m, int *); 632 633 switch (optval) { 634 635 case IP_PORTRANGE_DEFAULT: 636 inp->inp_flags &= ~(INP_LOWPORT); 637 inp->inp_flags &= ~(INP_HIGHPORT); 638 break; 639 640 case IP_PORTRANGE_HIGH: 641 inp->inp_flags &= ~(INP_LOWPORT); 642 inp->inp_flags |= INP_HIGHPORT; 643 break; 644 645 case IP_PORTRANGE_LOW: 646 inp->inp_flags &= ~(INP_HIGHPORT); 647 inp->inp_flags |= INP_LOWPORT; 648 break; 649 650 default: 651 error = EINVAL; 652 break; 653 } 654 } 655 656 default: 657 error = ENOPROTOOPT; 658 break; 659 } 660 if (m) 661 (void)m_free(m); 662 break; 663 664 case PRCO_GETOPT: 665 switch (optname) { 666 case IP_OPTIONS: 667 case IP_RETOPTS: 668 *mp = m = m_get(M_WAIT, MT_SOOPTS); 669 if (inp->inp_options) { 670 m->m_len = inp->inp_options->m_len; 671 (void)memcpy(mtod(m, void *), 672 mtod(inp->inp_options, void *), (unsigned)m->m_len); 673 } else 674 m->m_len = 0; 675 break; 676 677 case IP_TOS: 678 case IP_TTL: 679 case IP_RECVOPTS: 680 case IP_RECVRETOPTS: 681 case IP_RECVDSTADDR: 682 *mp = m = m_get(M_WAIT, MT_SOOPTS); 683 m->m_len = sizeof(int); 684 switch (optname) { 685 686 case IP_TOS: 687 optval = inp->inp_ip.ip_tos; 688 break; 689 690 case IP_TTL: 691 optval = inp->inp_ip.ip_ttl; 692 break; 693 694#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 695 696 case IP_RECVOPTS: 697 optval = OPTBIT(INP_RECVOPTS); 698 break; 699 700 case IP_RECVRETOPTS: 701 optval = OPTBIT(INP_RECVRETOPTS); 702 break; 703 704 case IP_RECVDSTADDR: 705 optval = OPTBIT(INP_RECVDSTADDR); 706 break; 707 } 708 *mtod(m, int *) = optval; 709 break; 710 711 case IP_MULTICAST_IF: 712 case IP_MULTICAST_VIF: 713 case IP_MULTICAST_TTL: 714 case IP_MULTICAST_LOOP: 715 case IP_ADD_MEMBERSHIP: 716 case IP_DROP_MEMBERSHIP: 717 error = ip_getmoptions(optname, inp->inp_moptions, mp); 718 break; 719 720 case IP_PORTRANGE: 721 *mp = m = m_get(M_WAIT, MT_SOOPTS); 722 m->m_len = sizeof(int); 723 724 if (inp->inp_flags & INP_HIGHPORT) 725 optval = IP_PORTRANGE_HIGH; 726 else if (inp->inp_flags & INP_LOWPORT) 727 optval = IP_PORTRANGE_LOW; 728 else 729 optval = 0; 730 731 *mtod(m, int *) = optval; 732 break; 733 734 default: 735 error = ENOPROTOOPT; 736 break; 737 } 738 break; 739 } 740 return (error); 741} 742 743/* 744 * Set up IP options in pcb for insertion in output packets. 745 * Store in mbuf with pointer in pcbopt, adding pseudo-option 746 * with destination address if source routed. 747 */ 748static int 749#ifdef notyet 750ip_pcbopts(optname, pcbopt, m) 751 int optname; 752#else 753ip_pcbopts(pcbopt, m) 754#endif 755 struct mbuf **pcbopt; 756 register struct mbuf *m; 757{ 758 register cnt, optlen; 759 register u_char *cp; 760 u_char opt; 761 762 /* turn off any old options */ 763 if (*pcbopt) 764 (void)m_free(*pcbopt); 765 *pcbopt = 0; 766 if (m == (struct mbuf *)0 || m->m_len == 0) { 767 /* 768 * Only turning off any previous options. 769 */ 770 if (m) 771 (void)m_free(m); 772 return (0); 773 } 774 775#ifndef vax 776 if (m->m_len % sizeof(long)) 777 goto bad; 778#endif 779 /* 780 * IP first-hop destination address will be stored before 781 * actual options; move other options back 782 * and clear it when none present. 783 */ 784 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 785 goto bad; 786 cnt = m->m_len; 787 m->m_len += sizeof(struct in_addr); 788 cp = mtod(m, u_char *) + sizeof(struct in_addr); 789 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 790 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 791 792 for (; cnt > 0; cnt -= optlen, cp += optlen) { 793 opt = cp[IPOPT_OPTVAL]; 794 if (opt == IPOPT_EOL) 795 break; 796 if (opt == IPOPT_NOP) 797 optlen = 1; 798 else { 799 optlen = cp[IPOPT_OLEN]; 800 if (optlen <= IPOPT_OLEN || optlen > cnt) 801 goto bad; 802 } 803 switch (opt) { 804 805 default: 806 break; 807 808 case IPOPT_LSRR: 809 case IPOPT_SSRR: 810 /* 811 * user process specifies route as: 812 * ->A->B->C->D 813 * D must be our final destination (but we can't 814 * check that since we may not have connected yet). 815 * A is first hop destination, which doesn't appear in 816 * actual IP option, but is stored before the options. 817 */ 818 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 819 goto bad; 820 m->m_len -= sizeof(struct in_addr); 821 cnt -= sizeof(struct in_addr); 822 optlen -= sizeof(struct in_addr); 823 cp[IPOPT_OLEN] = optlen; 824 /* 825 * Move first hop before start of options. 826 */ 827 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 828 sizeof(struct in_addr)); 829 /* 830 * Then copy rest of options back 831 * to close up the deleted entry. 832 */ 833 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 834 sizeof(struct in_addr)), 835 (caddr_t)&cp[IPOPT_OFFSET+1], 836 (unsigned)cnt + sizeof(struct in_addr)); 837 break; 838 } 839 } 840 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 841 goto bad; 842 *pcbopt = m; 843 return (0); 844 845bad: 846 (void)m_free(m); 847 return (EINVAL); 848} 849 850/* 851 * Set the IP multicast options in response to user setsockopt(). 852 */ 853static int 854ip_setmoptions(optname, imop, m) 855 int optname; 856 struct ip_moptions **imop; 857 struct mbuf *m; 858{ 859 register int error = 0; 860 u_char loop; 861 register int i; 862 struct in_addr addr; 863 register struct ip_mreq *mreq; 864 register struct ifnet *ifp; 865 register struct ip_moptions *imo = *imop; 866 struct route ro; 867 register struct sockaddr_in *dst; 868 int s; 869 870 if (imo == NULL) { 871 /* 872 * No multicast option buffer attached to the pcb; 873 * allocate one and initialize to default values. 874 */ 875 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 876 M_WAITOK); 877 878 if (imo == NULL) 879 return (ENOBUFS); 880 *imop = imo; 881 imo->imo_multicast_ifp = NULL; 882 imo->imo_multicast_vif = -1; 883 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 884 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 885 imo->imo_num_memberships = 0; 886 } 887 888 switch (optname) { 889 /* store an index number for the vif you wanna use in the send */ 890 case IP_MULTICAST_VIF: 891 if (!legal_vif_num) { 892 error = EOPNOTSUPP; 893 break; 894 } 895 if (m == NULL || m->m_len != sizeof(int)) { 896 error = EINVAL; 897 break; 898 } 899 i = *(mtod(m, int *)); 900 if (!legal_vif_num(i) && (i != -1)) { 901 error = EINVAL; 902 break; 903 } 904 imo->imo_multicast_vif = i; 905 break; 906 907 case IP_MULTICAST_IF: 908 /* 909 * Select the interface for outgoing multicast packets. 910 */ 911 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 912 error = EINVAL; 913 break; 914 } 915 addr = *(mtod(m, struct in_addr *)); 916 /* 917 * INADDR_ANY is used to remove a previous selection. 918 * When no interface is selected, a default one is 919 * chosen every time a multicast packet is sent. 920 */ 921 if (addr.s_addr == INADDR_ANY) { 922 imo->imo_multicast_ifp = NULL; 923 break; 924 } 925 /* 926 * The selected interface is identified by its local 927 * IP address. Find the interface and confirm that 928 * it supports multicasting. 929 */ 930 s = splimp(); 931 INADDR_TO_IFP(addr, ifp); 932 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 933 splx(s); 934 error = EADDRNOTAVAIL; 935 break; 936 } 937 imo->imo_multicast_ifp = ifp; 938 splx(s); 939 break; 940 941 case IP_MULTICAST_TTL: 942 /* 943 * Set the IP time-to-live for outgoing multicast packets. 944 */ 945 if (m == NULL || m->m_len != 1) { 946 error = EINVAL; 947 break; 948 } 949 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 950 break; 951 952 case IP_MULTICAST_LOOP: 953 /* 954 * Set the loopback flag for outgoing multicast packets. 955 * Must be zero or one. 956 */ 957 if (m == NULL || m->m_len != 1 || 958 (loop = *(mtod(m, u_char *))) > 1) { 959 error = EINVAL; 960 break; 961 } 962 imo->imo_multicast_loop = loop; 963 break; 964 965 case IP_ADD_MEMBERSHIP: 966 /* 967 * Add a multicast group membership. 968 * Group must be a valid IP multicast address. 969 */ 970 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 971 error = EINVAL; 972 break; 973 } 974 mreq = mtod(m, struct ip_mreq *); 975 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 976 error = EINVAL; 977 break; 978 } 979 s = splimp(); 980 /* 981 * If no interface address was provided, use the interface of 982 * the route to the given multicast address. 983 */ 984 if (mreq->imr_interface.s_addr == INADDR_ANY) { 985 bzero((caddr_t)&ro, sizeof(ro)); 986 dst = (struct sockaddr_in *)&ro.ro_dst; 987 dst->sin_len = sizeof(*dst); 988 dst->sin_family = AF_INET; 989 dst->sin_addr = mreq->imr_multiaddr; 990 rtalloc(&ro); 991 if (ro.ro_rt == NULL) { 992 error = EADDRNOTAVAIL; 993 splx(s); 994 break; 995 } 996 ifp = ro.ro_rt->rt_ifp; 997 rtfree(ro.ro_rt); 998 } 999 else { 1000 INADDR_TO_IFP(mreq->imr_interface, ifp); 1001 } 1002 1003 /* 1004 * See if we found an interface, and confirm that it 1005 * supports multicast. 1006 */ 1007 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1008 error = EADDRNOTAVAIL; 1009 splx(s); 1010 break; 1011 } 1012 /* 1013 * See if the membership already exists or if all the 1014 * membership slots are full. 1015 */ 1016 for (i = 0; i < imo->imo_num_memberships; ++i) { 1017 if (imo->imo_membership[i]->inm_ifp == ifp && 1018 imo->imo_membership[i]->inm_addr.s_addr 1019 == mreq->imr_multiaddr.s_addr) 1020 break; 1021 } 1022 if (i < imo->imo_num_memberships) { 1023 error = EADDRINUSE; 1024 splx(s); 1025 break; 1026 } 1027 if (i == IP_MAX_MEMBERSHIPS) { 1028 error = ETOOMANYREFS; 1029 splx(s); 1030 break; 1031 } 1032 /* 1033 * Everything looks good; add a new record to the multicast 1034 * address list for the given interface. 1035 */ 1036 if ((imo->imo_membership[i] = 1037 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1038 error = ENOBUFS; 1039 splx(s); 1040 break; 1041 } 1042 ++imo->imo_num_memberships; 1043 splx(s); 1044 break; 1045 1046 case IP_DROP_MEMBERSHIP: 1047 /* 1048 * Drop a multicast group membership. 1049 * Group must be a valid IP multicast address. 1050 */ 1051 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1052 error = EINVAL; 1053 break; 1054 } 1055 mreq = mtod(m, struct ip_mreq *); 1056 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1057 error = EINVAL; 1058 break; 1059 } 1060 1061 s = splimp(); 1062 /* 1063 * If an interface address was specified, get a pointer 1064 * to its ifnet structure. 1065 */ 1066 if (mreq->imr_interface.s_addr == INADDR_ANY) 1067 ifp = NULL; 1068 else { 1069 INADDR_TO_IFP(mreq->imr_interface, ifp); 1070 if (ifp == NULL) { 1071 error = EADDRNOTAVAIL; 1072 splx(s); 1073 break; 1074 } 1075 } 1076 /* 1077 * Find the membership in the membership array. 1078 */ 1079 for (i = 0; i < imo->imo_num_memberships; ++i) { 1080 if ((ifp == NULL || 1081 imo->imo_membership[i]->inm_ifp == ifp) && 1082 imo->imo_membership[i]->inm_addr.s_addr == 1083 mreq->imr_multiaddr.s_addr) 1084 break; 1085 } 1086 if (i == imo->imo_num_memberships) { 1087 error = EADDRNOTAVAIL; 1088 splx(s); 1089 break; 1090 } 1091 /* 1092 * Give up the multicast address record to which the 1093 * membership points. 1094 */ 1095 in_delmulti(imo->imo_membership[i]); 1096 /* 1097 * Remove the gap in the membership array. 1098 */ 1099 for (++i; i < imo->imo_num_memberships; ++i) 1100 imo->imo_membership[i-1] = imo->imo_membership[i]; 1101 --imo->imo_num_memberships; 1102 splx(s); 1103 break; 1104 1105 default: 1106 error = EOPNOTSUPP; 1107 break; 1108 } 1109 1110 /* 1111 * If all options have default values, no need to keep the mbuf. 1112 */ 1113 if (imo->imo_multicast_ifp == NULL && 1114 imo->imo_multicast_vif == -1 && 1115 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1116 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1117 imo->imo_num_memberships == 0) { 1118 free(*imop, M_IPMOPTS); 1119 *imop = NULL; 1120 } 1121 1122 return (error); 1123} 1124 1125/* 1126 * Return the IP multicast options in response to user getsockopt(). 1127 */ 1128static int 1129ip_getmoptions(optname, imo, mp) 1130 int optname; 1131 register struct ip_moptions *imo; 1132 register struct mbuf **mp; 1133{ 1134 u_char *ttl; 1135 u_char *loop; 1136 struct in_addr *addr; 1137 struct in_ifaddr *ia; 1138 1139 *mp = m_get(M_WAIT, MT_SOOPTS); 1140 1141 switch (optname) { 1142 1143 case IP_MULTICAST_VIF: 1144 if (imo != NULL) 1145 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1146 else 1147 *(mtod(*mp, int *)) = -1; 1148 (*mp)->m_len = sizeof(int); 1149 return(0); 1150 1151 case IP_MULTICAST_IF: 1152 addr = mtod(*mp, struct in_addr *); 1153 (*mp)->m_len = sizeof(struct in_addr); 1154 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1155 addr->s_addr = INADDR_ANY; 1156 else { 1157 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1158 addr->s_addr = (ia == NULL) ? INADDR_ANY 1159 : IA_SIN(ia)->sin_addr.s_addr; 1160 } 1161 return (0); 1162 1163 case IP_MULTICAST_TTL: 1164 ttl = mtod(*mp, u_char *); 1165 (*mp)->m_len = 1; 1166 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1167 : imo->imo_multicast_ttl; 1168 return (0); 1169 1170 case IP_MULTICAST_LOOP: 1171 loop = mtod(*mp, u_char *); 1172 (*mp)->m_len = 1; 1173 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1174 : imo->imo_multicast_loop; 1175 return (0); 1176 1177 default: 1178 return (EOPNOTSUPP); 1179 } 1180} 1181 1182/* 1183 * Discard the IP multicast options. 1184 */ 1185void 1186ip_freemoptions(imo) 1187 register struct ip_moptions *imo; 1188{ 1189 register int i; 1190 1191 if (imo != NULL) { 1192 for (i = 0; i < imo->imo_num_memberships; ++i) 1193 in_delmulti(imo->imo_membership[i]); 1194 free(imo, M_IPMOPTS); 1195 } 1196} 1197 1198/* 1199 * Routine called from ip_output() to loop back a copy of an IP multicast 1200 * packet to the input queue of a specified interface. Note that this 1201 * calls the output routine of the loopback "driver", but with an interface 1202 * pointer that might NOT be a loopback interface -- evil, but easier than 1203 * replicating that code here. 1204 */ 1205static void 1206ip_mloopback(ifp, m, dst) 1207 struct ifnet *ifp; 1208 register struct mbuf *m; 1209 register struct sockaddr_in *dst; 1210{ 1211 register struct ip *ip; 1212 struct mbuf *copym; 1213 1214 copym = m_copy(m, 0, M_COPYALL); 1215 if (copym != NULL) { 1216 /* 1217 * We don't bother to fragment if the IP length is greater 1218 * than the interface's MTU. Can this possibly matter? 1219 */ 1220 ip = mtod(copym, struct ip *); 1221 ip->ip_len = htons((u_short)ip->ip_len); 1222 ip->ip_off = htons((u_short)ip->ip_off); 1223 ip->ip_sum = 0; 1224 if (ip->ip_vhl == IP_VHL_BORING) { 1225 ip->ip_sum = in_cksum_hdr(ip); 1226 } else { 1227 ip->ip_sum = in_cksum(copym, 1228 IP_VHL_HL(ip->ip_vhl) << 2); 1229 } 1230 /* 1231 * NB: 1232 * We can't simply call ip_input() directly because 1233 * the ip_mforward() depends on the `input interface' 1234 * being set to something unreasonable so that we don't 1235 * attempt to forward the looped-back copy. 1236 * It's also not clear whether there are any lingering 1237 * reentrancy problems in other areas which might be 1238 * exposed by this code. For the moment, we'll err 1239 * on the side of safety by continuing to abuse 1240 * loinput(). 1241 */ 1242#ifdef notdef 1243 copym->m_pkthdr.rcvif = &loif[0]; 1244 ip_input(copym) 1245#else 1246 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1247#endif 1248 } 1249} 1250