ip_output.c revision 17758
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 1.41 1996/07/10 19:44:26 julian 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 64#if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1 65#undef COMPAT_IPFW 66#define COMPAT_IPFW 1 67#else 68#undef COMPAT_IPFW 69#endif 70 71u_short ip_id; 72 73static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 74static void ip_mloopback 75 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *)); 76static int ip_getmoptions 77 __P((int, struct ip_moptions *, struct mbuf **)); 78static int ip_optcopy __P((struct ip *, struct ip *)); 79static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 80static int ip_setmoptions 81 __P((int, struct ip_moptions **, struct mbuf *)); 82 83extern struct protosw inetsw[]; 84 85/* 86 * IP output. The packet in mbuf chain m contains a skeletal IP 87 * header (with len, off, ttl, proto, tos, src, dst). 88 * The mbuf chain containing the packet will be freed. 89 * The mbuf opt, if present, will not be freed. 90 */ 91int 92ip_output(m0, opt, ro, flags, imo) 93 struct mbuf *m0; 94 struct mbuf *opt; 95 struct route *ro; 96 int flags; 97 struct ip_moptions *imo; 98{ 99 struct ip *ip, *mhip; 100 struct ifnet *ifp; 101 struct mbuf *m = m0; 102 int hlen = sizeof (struct ip); 103 int len, off, error = 0; 104 struct sockaddr_in *dst; 105 struct in_ifaddr *ia; 106 int isbroadcast; 107 108#ifdef DIAGNOSTIC 109 if ((m->m_flags & M_PKTHDR) == 0) 110 panic("ip_output no HDR"); 111 if (!ro) 112 panic("ip_output no route, proto = %d", 113 mtod(m, struct ip *)->ip_p); 114#endif 115 if (opt) { 116 m = ip_insertoptions(m, opt, &len); 117 hlen = len; 118 } 119 ip = mtod(m, struct ip *); 120 /* 121 * Fill in IP header. 122 */ 123 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 124 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 125 ip->ip_off &= IP_DF; 126 ip->ip_id = htons(ip_id++); 127 ipstat.ips_localout++; 128 } else { 129 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 130 } 131 132 dst = (struct sockaddr_in *)&ro->ro_dst; 133 /* 134 * If there is a cached route, 135 * check that it is to the same destination 136 * and is still up. If not, free it and try again. 137 */ 138 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 139 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 140 RTFREE(ro->ro_rt); 141 ro->ro_rt = (struct rtentry *)0; 142 } 143 if (ro->ro_rt == 0) { 144 dst->sin_family = AF_INET; 145 dst->sin_len = sizeof(*dst); 146 dst->sin_addr = ip->ip_dst; 147 } 148 /* 149 * If routing to interface only, 150 * short circuit routing lookup. 151 */ 152#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 153#define sintosa(sin) ((struct sockaddr *)(sin)) 154 if (flags & IP_ROUTETOIF) { 155 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 156 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 157 ipstat.ips_noroute++; 158 error = ENETUNREACH; 159 goto bad; 160 } 161 ifp = ia->ia_ifp; 162 ip->ip_ttl = 1; 163 isbroadcast = in_broadcast(dst->sin_addr, ifp); 164 } else { 165 /* 166 * If this is the case, we probably don't want to allocate 167 * a protocol-cloned route since we didn't get one from the 168 * ULP. This lets TCP do its thing, while not burdening 169 * forwarding or ICMP with the overhead of cloning a route. 170 * Of course, we still want to do any cloning requested by 171 * the link layer, as this is probably required in all cases 172 * for correct operation (as it is for ARP). 173 */ 174 if (ro->ro_rt == 0) 175 rtalloc_ign(ro, RTF_PRCLONING); 176 if (ro->ro_rt == 0) { 177 ipstat.ips_noroute++; 178 error = EHOSTUNREACH; 179 goto bad; 180 } 181 ia = ifatoia(ro->ro_rt->rt_ifa); 182 ifp = ro->ro_rt->rt_ifp; 183 ro->ro_rt->rt_use++; 184 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 185 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 186 if (ro->ro_rt->rt_flags & RTF_HOST) 187 isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST); 188 else 189 isbroadcast = in_broadcast(dst->sin_addr, ifp); 190 } 191 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 192 struct in_multi *inm; 193 194 m->m_flags |= M_MCAST; 195 /* 196 * IP destination address is multicast. Make sure "dst" 197 * still points to the address in "ro". (It may have been 198 * changed to point to a gateway address, above.) 199 */ 200 dst = (struct sockaddr_in *)&ro->ro_dst; 201 /* 202 * See if the caller provided any multicast options 203 */ 204 if (imo != NULL) { 205 ip->ip_ttl = imo->imo_multicast_ttl; 206 if (imo->imo_multicast_ifp != NULL) 207 ifp = imo->imo_multicast_ifp; 208 if (imo->imo_multicast_vif != -1) 209 ip->ip_src.s_addr = 210 ip_mcast_src(imo->imo_multicast_vif); 211 } else 212 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 213 /* 214 * Confirm that the outgoing interface supports multicast. 215 */ 216 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 217 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 218 ipstat.ips_noroute++; 219 error = ENETUNREACH; 220 goto bad; 221 } 222 } 223 /* 224 * If source address not specified yet, use address 225 * of outgoing interface. 226 */ 227 if (ip->ip_src.s_addr == INADDR_ANY) { 228 register struct in_ifaddr *ia; 229 230 for (ia = in_ifaddr; ia; ia = ia->ia_next) 231 if (ia->ia_ifp == ifp) { 232 ip->ip_src = IA_SIN(ia)->sin_addr; 233 break; 234 } 235 } 236 237 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 238 if (inm != NULL && 239 (imo == NULL || imo->imo_multicast_loop)) { 240 /* 241 * If we belong to the destination multicast group 242 * on the outgoing interface, and the caller did not 243 * forbid loopback, loop back a copy. 244 */ 245 ip_mloopback(ifp, m, dst); 246 } 247 else { 248 /* 249 * If we are acting as a multicast router, perform 250 * multicast forwarding as if the packet had just 251 * arrived on the interface to which we are about 252 * to send. The multicast forwarding function 253 * recursively calls this function, using the 254 * IP_FORWARDING flag to prevent infinite recursion. 255 * 256 * Multicasts that are looped back by ip_mloopback(), 257 * above, will be forwarded by the ip_input() routine, 258 * if necessary. 259 */ 260 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 261 /* 262 * Check if rsvp daemon is running. If not, don't 263 * set ip_moptions. This ensures that the packet 264 * is multicast and not just sent down one link 265 * as prescribed by rsvpd. 266 */ 267 if (!rsvp_on) 268 imo = NULL; 269 if (ip_mforward(ip, ifp, m, imo) != 0) { 270 m_freem(m); 271 goto done; 272 } 273 } 274 } 275 276 /* 277 * Multicasts with a time-to-live of zero may be looped- 278 * back, above, but must not be transmitted on a network. 279 * Also, multicasts addressed to the loopback interface 280 * are not sent -- the above call to ip_mloopback() will 281 * loop back a copy if this host actually belongs to the 282 * destination group on the loopback interface. 283 */ 284 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 285 m_freem(m); 286 goto done; 287 } 288 289 goto sendit; 290 } 291#ifndef notdef 292 /* 293 * If source address not specified yet, use address 294 * of outgoing interface. 295 */ 296 if (ip->ip_src.s_addr == INADDR_ANY) 297 ip->ip_src = IA_SIN(ia)->sin_addr; 298#endif 299 /* 300 * Verify that we have any chance at all of being able to queue 301 * the packet or packet fragments 302 */ 303 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 304 ifp->if_snd.ifq_maxlen) { 305 error = ENOBUFS; 306 goto bad; 307 } 308 309 /* 310 * Look for broadcast address and 311 * and verify user is allowed to send 312 * such a packet. 313 */ 314 if (isbroadcast) { 315 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 316 error = EADDRNOTAVAIL; 317 goto bad; 318 } 319 if ((flags & IP_ALLOWBROADCAST) == 0) { 320 error = EACCES; 321 goto bad; 322 } 323 /* don't allow broadcast messages to be fragmented */ 324 if ((u_short)ip->ip_len > ifp->if_mtu) { 325 error = EMSGSIZE; 326 goto bad; 327 } 328 m->m_flags |= M_BCAST; 329 } else { 330 m->m_flags &= ~M_BCAST; 331 } 332 333sendit: 334 /* 335 * IpHack's section. 336 * - Xlate: translate packet's addr/port (NAT). 337 * - Firewall: deny/allow 338 * - Wrap: fake packet's addr/port <unimpl.> 339 * - Encapsulate: put it in another IP and send out. <unimp.> 340 */ 341 342 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, IP_NAT_OUT)) { 343 error = EACCES; 344 goto done; 345 } 346 347#ifdef COMPAT_IPFW 348 /* 349 * Check with the firewall... 350 */ 351 if (ip_fw_chk_ptr) { 352 int action; 353 354#ifdef IPDIVERT 355 action = (*ip_fw_chk_ptr)(&ip, 356 hlen, ifp, (~0 << 16) | ip_divert_ignore, &m); 357#else 358 action = (*ip_fw_chk_ptr)(&ip, hlen, ifp, (~0 << 16), &m); 359#endif 360 if (action == -1) { 361 error = EACCES; /* XXX is this appropriate? */ 362 goto done; 363 } else if (action != 0) { 364#ifdef IPDIVERT 365 ip_divert_port = action; /* divert to port */ 366 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 367 goto done; 368#else 369 m_freem(m); /* ipfw says divert, but we can't */ 370 goto done; 371#endif 372 } 373 } 374#endif /* COMPAT_IPFW */ 375 376 /* 377 * If small enough for interface, can just send directly. 378 */ 379 if ((u_short)ip->ip_len <= ifp->if_mtu) { 380 ip->ip_len = htons((u_short)ip->ip_len); 381 ip->ip_off = htons((u_short)ip->ip_off); 382 ip->ip_sum = 0; 383 if (ip->ip_vhl == IP_VHL_BORING) { 384 ip->ip_sum = in_cksum_hdr(ip); 385 } else { 386 ip->ip_sum = in_cksum(m, hlen); 387 } 388 error = (*ifp->if_output)(ifp, m, 389 (struct sockaddr *)dst, ro->ro_rt); 390 goto done; 391 } 392 /* 393 * Too large for interface; fragment if possible. 394 * Must be able to put at least 8 bytes per fragment. 395 */ 396 if (ip->ip_off & IP_DF) { 397 error = EMSGSIZE; 398 /* 399 * This case can happen if the user changed the MTU 400 * of an interface after enabling IP on it. Because 401 * most netifs don't keep track of routes pointing to 402 * them, there is no way for one to update all its 403 * routes when the MTU is changed. 404 */ 405 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 406 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 407 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 408 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 409 } 410 ipstat.ips_cantfrag++; 411 goto bad; 412 } 413 len = (ifp->if_mtu - hlen) &~ 7; 414 if (len < 8) { 415 error = EMSGSIZE; 416 goto bad; 417 } 418 419 { 420 int mhlen, firstlen = len; 421 struct mbuf **mnext = &m->m_nextpkt; 422 423 /* 424 * Loop through length of segment after first fragment, 425 * make new header and copy data of each part and link onto chain. 426 */ 427 m0 = m; 428 mhlen = sizeof (struct ip); 429 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 430 MGETHDR(m, M_DONTWAIT, MT_HEADER); 431 if (m == 0) { 432 error = ENOBUFS; 433 ipstat.ips_odropped++; 434 goto sendorfree; 435 } 436 m->m_data += max_linkhdr; 437 mhip = mtod(m, struct ip *); 438 *mhip = *ip; 439 if (hlen > sizeof (struct ip)) { 440 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 441 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 442 } 443 m->m_len = mhlen; 444 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 445 if (ip->ip_off & IP_MF) 446 mhip->ip_off |= IP_MF; 447 if (off + len >= (u_short)ip->ip_len) 448 len = (u_short)ip->ip_len - off; 449 else 450 mhip->ip_off |= IP_MF; 451 mhip->ip_len = htons((u_short)(len + mhlen)); 452 m->m_next = m_copy(m0, off, len); 453 if (m->m_next == 0) { 454 (void) m_free(m); 455 error = ENOBUFS; /* ??? */ 456 ipstat.ips_odropped++; 457 goto sendorfree; 458 } 459 m->m_pkthdr.len = mhlen + len; 460 m->m_pkthdr.rcvif = (struct ifnet *)0; 461 mhip->ip_off = htons((u_short)mhip->ip_off); 462 mhip->ip_sum = 0; 463 if (mhip->ip_vhl == IP_VHL_BORING) { 464 mhip->ip_sum = in_cksum_hdr(mhip); 465 } else { 466 mhip->ip_sum = in_cksum(m, mhlen); 467 } 468 *mnext = m; 469 mnext = &m->m_nextpkt; 470 ipstat.ips_ofragments++; 471 } 472 /* 473 * Update first fragment by trimming what's been copied out 474 * and updating header, then send each fragment (in order). 475 */ 476 m = m0; 477 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 478 m->m_pkthdr.len = hlen + firstlen; 479 ip->ip_len = htons((u_short)m->m_pkthdr.len); 480 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 481 ip->ip_sum = 0; 482 if (ip->ip_vhl == IP_VHL_BORING) { 483 ip->ip_sum = in_cksum_hdr(ip); 484 } else { 485 ip->ip_sum = in_cksum(m, hlen); 486 } 487sendorfree: 488 for (m = m0; m; m = m0) { 489 m0 = m->m_nextpkt; 490 m->m_nextpkt = 0; 491 if (error == 0) 492 error = (*ifp->if_output)(ifp, m, 493 (struct sockaddr *)dst, ro->ro_rt); 494 else 495 m_freem(m); 496 } 497 498 if (error == 0) 499 ipstat.ips_fragmented++; 500 } 501done: 502 return (error); 503bad: 504 m_freem(m0); 505 goto done; 506} 507 508/* 509 * Insert IP options into preformed packet. 510 * Adjust IP destination as required for IP source routing, 511 * as indicated by a non-zero in_addr at the start of the options. 512 * 513 * XXX This routine assumes that the packet has no options in place. 514 */ 515static struct mbuf * 516ip_insertoptions(m, opt, phlen) 517 register struct mbuf *m; 518 struct mbuf *opt; 519 int *phlen; 520{ 521 register struct ipoption *p = mtod(opt, struct ipoption *); 522 struct mbuf *n; 523 register struct ip *ip = mtod(m, struct ip *); 524 unsigned optlen; 525 526 optlen = opt->m_len - sizeof(p->ipopt_dst); 527 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 528 return (m); /* XXX should fail */ 529 if (p->ipopt_dst.s_addr) 530 ip->ip_dst = p->ipopt_dst; 531 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 532 MGETHDR(n, M_DONTWAIT, MT_HEADER); 533 if (n == 0) 534 return (m); 535 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 536 m->m_len -= sizeof(struct ip); 537 m->m_data += sizeof(struct ip); 538 n->m_next = m; 539 m = n; 540 m->m_len = optlen + sizeof(struct ip); 541 m->m_data += max_linkhdr; 542 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 543 } else { 544 m->m_data -= optlen; 545 m->m_len += optlen; 546 m->m_pkthdr.len += optlen; 547 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 548 } 549 ip = mtod(m, struct ip *); 550 bcopy(p->ipopt_list, ip + 1, optlen); 551 *phlen = sizeof(struct ip) + optlen; 552 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 553 ip->ip_len += optlen; 554 return (m); 555} 556 557/* 558 * Copy options from ip to jp, 559 * omitting those not copied during fragmentation. 560 */ 561static int 562ip_optcopy(ip, jp) 563 struct ip *ip, *jp; 564{ 565 register u_char *cp, *dp; 566 int opt, optlen, cnt; 567 568 cp = (u_char *)(ip + 1); 569 dp = (u_char *)(jp + 1); 570 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 571 for (; cnt > 0; cnt -= optlen, cp += optlen) { 572 opt = cp[0]; 573 if (opt == IPOPT_EOL) 574 break; 575 if (opt == IPOPT_NOP) { 576 /* Preserve for IP mcast tunnel's LSRR alignment. */ 577 *dp++ = IPOPT_NOP; 578 optlen = 1; 579 continue; 580 } else 581 optlen = cp[IPOPT_OLEN]; 582 /* bogus lengths should have been caught by ip_dooptions */ 583 if (optlen > cnt) 584 optlen = cnt; 585 if (IPOPT_COPIED(opt)) { 586 bcopy(cp, dp, optlen); 587 dp += optlen; 588 } 589 } 590 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 591 *dp++ = IPOPT_EOL; 592 return (optlen); 593} 594 595/* 596 * IP socket option processing. 597 */ 598int 599ip_ctloutput(op, so, level, optname, mp) 600 int op; 601 struct socket *so; 602 int level, optname; 603 struct mbuf **mp; 604{ 605 register struct inpcb *inp = sotoinpcb(so); 606 register struct mbuf *m = *mp; 607 register int optval = 0; 608 int error = 0; 609 610 if (level != IPPROTO_IP) { 611 error = EINVAL; 612 if (op == PRCO_SETOPT && *mp) 613 (void) m_free(*mp); 614 } else switch (op) { 615 616 case PRCO_SETOPT: 617 switch (optname) { 618 case IP_OPTIONS: 619#ifdef notyet 620 case IP_RETOPTS: 621 return (ip_pcbopts(optname, &inp->inp_options, m)); 622#else 623 return (ip_pcbopts(&inp->inp_options, m)); 624#endif 625 626 case IP_TOS: 627 case IP_TTL: 628 case IP_RECVOPTS: 629 case IP_RECVRETOPTS: 630 case IP_RECVDSTADDR: 631 if (m == 0 || m->m_len != sizeof(int)) 632 error = EINVAL; 633 else { 634 optval = *mtod(m, int *); 635 switch (optname) { 636 637 case IP_TOS: 638 inp->inp_ip.ip_tos = optval; 639 break; 640 641 case IP_TTL: 642 inp->inp_ip.ip_ttl = optval; 643 break; 644#define OPTSET(bit) \ 645 if (optval) \ 646 inp->inp_flags |= bit; \ 647 else \ 648 inp->inp_flags &= ~bit; 649 650 case IP_RECVOPTS: 651 OPTSET(INP_RECVOPTS); 652 break; 653 654 case IP_RECVRETOPTS: 655 OPTSET(INP_RECVRETOPTS); 656 break; 657 658 case IP_RECVDSTADDR: 659 OPTSET(INP_RECVDSTADDR); 660 break; 661 } 662 } 663 break; 664#undef OPTSET 665 666 case IP_MULTICAST_IF: 667 case IP_MULTICAST_VIF: 668 case IP_MULTICAST_TTL: 669 case IP_MULTICAST_LOOP: 670 case IP_ADD_MEMBERSHIP: 671 case IP_DROP_MEMBERSHIP: 672 error = ip_setmoptions(optname, &inp->inp_moptions, m); 673 break; 674 675 case IP_PORTRANGE: 676 if (m == 0 || m->m_len != sizeof(int)) 677 error = EINVAL; 678 else { 679 optval = *mtod(m, int *); 680 681 switch (optval) { 682 683 case IP_PORTRANGE_DEFAULT: 684 inp->inp_flags &= ~(INP_LOWPORT); 685 inp->inp_flags &= ~(INP_HIGHPORT); 686 break; 687 688 case IP_PORTRANGE_HIGH: 689 inp->inp_flags &= ~(INP_LOWPORT); 690 inp->inp_flags |= INP_HIGHPORT; 691 break; 692 693 case IP_PORTRANGE_LOW: 694 inp->inp_flags &= ~(INP_HIGHPORT); 695 inp->inp_flags |= INP_LOWPORT; 696 break; 697 698 default: 699 error = EINVAL; 700 break; 701 } 702 } 703 break; 704 705 default: 706 error = ENOPROTOOPT; 707 break; 708 } 709 if (m) 710 (void)m_free(m); 711 break; 712 713 case PRCO_GETOPT: 714 switch (optname) { 715 case IP_OPTIONS: 716 case IP_RETOPTS: 717 *mp = m = m_get(M_WAIT, MT_SOOPTS); 718 if (inp->inp_options) { 719 m->m_len = inp->inp_options->m_len; 720 bcopy(mtod(inp->inp_options, void *), 721 mtod(m, void *), m->m_len); 722 } else 723 m->m_len = 0; 724 break; 725 726 case IP_TOS: 727 case IP_TTL: 728 case IP_RECVOPTS: 729 case IP_RECVRETOPTS: 730 case IP_RECVDSTADDR: 731 *mp = m = m_get(M_WAIT, MT_SOOPTS); 732 m->m_len = sizeof(int); 733 switch (optname) { 734 735 case IP_TOS: 736 optval = inp->inp_ip.ip_tos; 737 break; 738 739 case IP_TTL: 740 optval = inp->inp_ip.ip_ttl; 741 break; 742 743#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 744 745 case IP_RECVOPTS: 746 optval = OPTBIT(INP_RECVOPTS); 747 break; 748 749 case IP_RECVRETOPTS: 750 optval = OPTBIT(INP_RECVRETOPTS); 751 break; 752 753 case IP_RECVDSTADDR: 754 optval = OPTBIT(INP_RECVDSTADDR); 755 break; 756 } 757 *mtod(m, int *) = optval; 758 break; 759 760 case IP_MULTICAST_IF: 761 case IP_MULTICAST_VIF: 762 case IP_MULTICAST_TTL: 763 case IP_MULTICAST_LOOP: 764 case IP_ADD_MEMBERSHIP: 765 case IP_DROP_MEMBERSHIP: 766 error = ip_getmoptions(optname, inp->inp_moptions, mp); 767 break; 768 769 case IP_PORTRANGE: 770 *mp = m = m_get(M_WAIT, MT_SOOPTS); 771 m->m_len = sizeof(int); 772 773 if (inp->inp_flags & INP_HIGHPORT) 774 optval = IP_PORTRANGE_HIGH; 775 else if (inp->inp_flags & INP_LOWPORT) 776 optval = IP_PORTRANGE_LOW; 777 else 778 optval = 0; 779 780 *mtod(m, int *) = optval; 781 break; 782 783 default: 784 error = ENOPROTOOPT; 785 break; 786 } 787 break; 788 } 789 return (error); 790} 791 792/* 793 * Set up IP options in pcb for insertion in output packets. 794 * Store in mbuf with pointer in pcbopt, adding pseudo-option 795 * with destination address if source routed. 796 */ 797static int 798#ifdef notyet 799ip_pcbopts(optname, pcbopt, m) 800 int optname; 801#else 802ip_pcbopts(pcbopt, m) 803#endif 804 struct mbuf **pcbopt; 805 register struct mbuf *m; 806{ 807 register cnt, optlen; 808 register u_char *cp; 809 u_char opt; 810 811 /* turn off any old options */ 812 if (*pcbopt) 813 (void)m_free(*pcbopt); 814 *pcbopt = 0; 815 if (m == (struct mbuf *)0 || m->m_len == 0) { 816 /* 817 * Only turning off any previous options. 818 */ 819 if (m) 820 (void)m_free(m); 821 return (0); 822 } 823 824#ifndef vax 825 if (m->m_len % sizeof(long)) 826 goto bad; 827#endif 828 /* 829 * IP first-hop destination address will be stored before 830 * actual options; move other options back 831 * and clear it when none present. 832 */ 833 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 834 goto bad; 835 cnt = m->m_len; 836 m->m_len += sizeof(struct in_addr); 837 cp = mtod(m, u_char *) + sizeof(struct in_addr); 838 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 839 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 840 841 for (; cnt > 0; cnt -= optlen, cp += optlen) { 842 opt = cp[IPOPT_OPTVAL]; 843 if (opt == IPOPT_EOL) 844 break; 845 if (opt == IPOPT_NOP) 846 optlen = 1; 847 else { 848 optlen = cp[IPOPT_OLEN]; 849 if (optlen <= IPOPT_OLEN || optlen > cnt) 850 goto bad; 851 } 852 switch (opt) { 853 854 default: 855 break; 856 857 case IPOPT_LSRR: 858 case IPOPT_SSRR: 859 /* 860 * user process specifies route as: 861 * ->A->B->C->D 862 * D must be our final destination (but we can't 863 * check that since we may not have connected yet). 864 * A is first hop destination, which doesn't appear in 865 * actual IP option, but is stored before the options. 866 */ 867 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 868 goto bad; 869 m->m_len -= sizeof(struct in_addr); 870 cnt -= sizeof(struct in_addr); 871 optlen -= sizeof(struct in_addr); 872 cp[IPOPT_OLEN] = optlen; 873 /* 874 * Move first hop before start of options. 875 */ 876 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 877 sizeof(struct in_addr)); 878 /* 879 * Then copy rest of options back 880 * to close up the deleted entry. 881 */ 882 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 883 sizeof(struct in_addr)), 884 (caddr_t)&cp[IPOPT_OFFSET+1], 885 (unsigned)cnt + sizeof(struct in_addr)); 886 break; 887 } 888 } 889 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 890 goto bad; 891 *pcbopt = m; 892 return (0); 893 894bad: 895 (void)m_free(m); 896 return (EINVAL); 897} 898 899/* 900 * Set the IP multicast options in response to user setsockopt(). 901 */ 902static int 903ip_setmoptions(optname, imop, m) 904 int optname; 905 struct ip_moptions **imop; 906 struct mbuf *m; 907{ 908 register int error = 0; 909 u_char loop; 910 register int i; 911 struct in_addr addr; 912 register struct ip_mreq *mreq; 913 register struct ifnet *ifp; 914 register struct ip_moptions *imo = *imop; 915 struct route ro; 916 register struct sockaddr_in *dst; 917 int s; 918 919 if (imo == NULL) { 920 /* 921 * No multicast option buffer attached to the pcb; 922 * allocate one and initialize to default values. 923 */ 924 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 925 M_WAITOK); 926 927 if (imo == NULL) 928 return (ENOBUFS); 929 *imop = imo; 930 imo->imo_multicast_ifp = NULL; 931 imo->imo_multicast_vif = -1; 932 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 933 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 934 imo->imo_num_memberships = 0; 935 } 936 937 switch (optname) { 938 /* store an index number for the vif you wanna use in the send */ 939 case IP_MULTICAST_VIF: 940 if (!legal_vif_num) { 941 error = EOPNOTSUPP; 942 break; 943 } 944 if (m == NULL || m->m_len != sizeof(int)) { 945 error = EINVAL; 946 break; 947 } 948 i = *(mtod(m, int *)); 949 if (!legal_vif_num(i) && (i != -1)) { 950 error = EINVAL; 951 break; 952 } 953 imo->imo_multicast_vif = i; 954 break; 955 956 case IP_MULTICAST_IF: 957 /* 958 * Select the interface for outgoing multicast packets. 959 */ 960 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 961 error = EINVAL; 962 break; 963 } 964 addr = *(mtod(m, struct in_addr *)); 965 /* 966 * INADDR_ANY is used to remove a previous selection. 967 * When no interface is selected, a default one is 968 * chosen every time a multicast packet is sent. 969 */ 970 if (addr.s_addr == INADDR_ANY) { 971 imo->imo_multicast_ifp = NULL; 972 break; 973 } 974 /* 975 * The selected interface is identified by its local 976 * IP address. Find the interface and confirm that 977 * it supports multicasting. 978 */ 979 s = splimp(); 980 INADDR_TO_IFP(addr, ifp); 981 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 982 splx(s); 983 error = EADDRNOTAVAIL; 984 break; 985 } 986 imo->imo_multicast_ifp = ifp; 987 splx(s); 988 break; 989 990 case IP_MULTICAST_TTL: 991 /* 992 * Set the IP time-to-live for outgoing multicast packets. 993 */ 994 if (m == NULL || m->m_len != 1) { 995 error = EINVAL; 996 break; 997 } 998 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 999 break; 1000 1001 case IP_MULTICAST_LOOP: 1002 /* 1003 * Set the loopback flag for outgoing multicast packets. 1004 * Must be zero or one. 1005 */ 1006 if (m == NULL || m->m_len != 1 || 1007 (loop = *(mtod(m, u_char *))) > 1) { 1008 error = EINVAL; 1009 break; 1010 } 1011 imo->imo_multicast_loop = loop; 1012 break; 1013 1014 case IP_ADD_MEMBERSHIP: 1015 /* 1016 * Add a multicast group membership. 1017 * Group must be a valid IP multicast address. 1018 */ 1019 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1020 error = EINVAL; 1021 break; 1022 } 1023 mreq = mtod(m, struct ip_mreq *); 1024 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1025 error = EINVAL; 1026 break; 1027 } 1028 s = splimp(); 1029 /* 1030 * If no interface address was provided, use the interface of 1031 * the route to the given multicast address. 1032 */ 1033 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1034 bzero((caddr_t)&ro, sizeof(ro)); 1035 dst = (struct sockaddr_in *)&ro.ro_dst; 1036 dst->sin_len = sizeof(*dst); 1037 dst->sin_family = AF_INET; 1038 dst->sin_addr = mreq->imr_multiaddr; 1039 rtalloc(&ro); 1040 if (ro.ro_rt == NULL) { 1041 error = EADDRNOTAVAIL; 1042 splx(s); 1043 break; 1044 } 1045 ifp = ro.ro_rt->rt_ifp; 1046 rtfree(ro.ro_rt); 1047 } 1048 else { 1049 INADDR_TO_IFP(mreq->imr_interface, ifp); 1050 } 1051 1052 /* 1053 * See if we found an interface, and confirm that it 1054 * supports multicast. 1055 */ 1056 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1057 error = EADDRNOTAVAIL; 1058 splx(s); 1059 break; 1060 } 1061 /* 1062 * See if the membership already exists or if all the 1063 * membership slots are full. 1064 */ 1065 for (i = 0; i < imo->imo_num_memberships; ++i) { 1066 if (imo->imo_membership[i]->inm_ifp == ifp && 1067 imo->imo_membership[i]->inm_addr.s_addr 1068 == mreq->imr_multiaddr.s_addr) 1069 break; 1070 } 1071 if (i < imo->imo_num_memberships) { 1072 error = EADDRINUSE; 1073 splx(s); 1074 break; 1075 } 1076 if (i == IP_MAX_MEMBERSHIPS) { 1077 error = ETOOMANYREFS; 1078 splx(s); 1079 break; 1080 } 1081 /* 1082 * Everything looks good; add a new record to the multicast 1083 * address list for the given interface. 1084 */ 1085 if ((imo->imo_membership[i] = 1086 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1087 error = ENOBUFS; 1088 splx(s); 1089 break; 1090 } 1091 ++imo->imo_num_memberships; 1092 splx(s); 1093 break; 1094 1095 case IP_DROP_MEMBERSHIP: 1096 /* 1097 * Drop a multicast group membership. 1098 * Group must be a valid IP multicast address. 1099 */ 1100 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1101 error = EINVAL; 1102 break; 1103 } 1104 mreq = mtod(m, struct ip_mreq *); 1105 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1106 error = EINVAL; 1107 break; 1108 } 1109 1110 s = splimp(); 1111 /* 1112 * If an interface address was specified, get a pointer 1113 * to its ifnet structure. 1114 */ 1115 if (mreq->imr_interface.s_addr == INADDR_ANY) 1116 ifp = NULL; 1117 else { 1118 INADDR_TO_IFP(mreq->imr_interface, ifp); 1119 if (ifp == NULL) { 1120 error = EADDRNOTAVAIL; 1121 splx(s); 1122 break; 1123 } 1124 } 1125 /* 1126 * Find the membership in the membership array. 1127 */ 1128 for (i = 0; i < imo->imo_num_memberships; ++i) { 1129 if ((ifp == NULL || 1130 imo->imo_membership[i]->inm_ifp == ifp) && 1131 imo->imo_membership[i]->inm_addr.s_addr == 1132 mreq->imr_multiaddr.s_addr) 1133 break; 1134 } 1135 if (i == imo->imo_num_memberships) { 1136 error = EADDRNOTAVAIL; 1137 splx(s); 1138 break; 1139 } 1140 /* 1141 * Give up the multicast address record to which the 1142 * membership points. 1143 */ 1144 in_delmulti(imo->imo_membership[i]); 1145 /* 1146 * Remove the gap in the membership array. 1147 */ 1148 for (++i; i < imo->imo_num_memberships; ++i) 1149 imo->imo_membership[i-1] = imo->imo_membership[i]; 1150 --imo->imo_num_memberships; 1151 splx(s); 1152 break; 1153 1154 default: 1155 error = EOPNOTSUPP; 1156 break; 1157 } 1158 1159 /* 1160 * If all options have default values, no need to keep the mbuf. 1161 */ 1162 if (imo->imo_multicast_ifp == NULL && 1163 imo->imo_multicast_vif == -1 && 1164 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1165 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1166 imo->imo_num_memberships == 0) { 1167 free(*imop, M_IPMOPTS); 1168 *imop = NULL; 1169 } 1170 1171 return (error); 1172} 1173 1174/* 1175 * Return the IP multicast options in response to user getsockopt(). 1176 */ 1177static int 1178ip_getmoptions(optname, imo, mp) 1179 int optname; 1180 register struct ip_moptions *imo; 1181 register struct mbuf **mp; 1182{ 1183 u_char *ttl; 1184 u_char *loop; 1185 struct in_addr *addr; 1186 struct in_ifaddr *ia; 1187 1188 *mp = m_get(M_WAIT, MT_SOOPTS); 1189 1190 switch (optname) { 1191 1192 case IP_MULTICAST_VIF: 1193 if (imo != NULL) 1194 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1195 else 1196 *(mtod(*mp, int *)) = -1; 1197 (*mp)->m_len = sizeof(int); 1198 return(0); 1199 1200 case IP_MULTICAST_IF: 1201 addr = mtod(*mp, struct in_addr *); 1202 (*mp)->m_len = sizeof(struct in_addr); 1203 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1204 addr->s_addr = INADDR_ANY; 1205 else { 1206 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1207 addr->s_addr = (ia == NULL) ? INADDR_ANY 1208 : IA_SIN(ia)->sin_addr.s_addr; 1209 } 1210 return (0); 1211 1212 case IP_MULTICAST_TTL: 1213 ttl = mtod(*mp, u_char *); 1214 (*mp)->m_len = 1; 1215 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1216 : imo->imo_multicast_ttl; 1217 return (0); 1218 1219 case IP_MULTICAST_LOOP: 1220 loop = mtod(*mp, u_char *); 1221 (*mp)->m_len = 1; 1222 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1223 : imo->imo_multicast_loop; 1224 return (0); 1225 1226 default: 1227 return (EOPNOTSUPP); 1228 } 1229} 1230 1231/* 1232 * Discard the IP multicast options. 1233 */ 1234void 1235ip_freemoptions(imo) 1236 register struct ip_moptions *imo; 1237{ 1238 register int i; 1239 1240 if (imo != NULL) { 1241 for (i = 0; i < imo->imo_num_memberships; ++i) 1242 in_delmulti(imo->imo_membership[i]); 1243 free(imo, M_IPMOPTS); 1244 } 1245} 1246 1247/* 1248 * Routine called from ip_output() to loop back a copy of an IP multicast 1249 * packet to the input queue of a specified interface. Note that this 1250 * calls the output routine of the loopback "driver", but with an interface 1251 * pointer that might NOT be a loopback interface -- evil, but easier than 1252 * replicating that code here. 1253 */ 1254static void 1255ip_mloopback(ifp, m, dst) 1256 struct ifnet *ifp; 1257 register struct mbuf *m; 1258 register struct sockaddr_in *dst; 1259{ 1260 register struct ip *ip; 1261 struct mbuf *copym; 1262 1263 copym = m_copy(m, 0, M_COPYALL); 1264 if (copym != NULL) { 1265 /* 1266 * We don't bother to fragment if the IP length is greater 1267 * than the interface's MTU. Can this possibly matter? 1268 */ 1269 ip = mtod(copym, struct ip *); 1270 ip->ip_len = htons((u_short)ip->ip_len); 1271 ip->ip_off = htons((u_short)ip->ip_off); 1272 ip->ip_sum = 0; 1273 if (ip->ip_vhl == IP_VHL_BORING) { 1274 ip->ip_sum = in_cksum_hdr(ip); 1275 } else { 1276 ip->ip_sum = in_cksum(copym, 1277 IP_VHL_HL(ip->ip_vhl) << 2); 1278 } 1279 /* 1280 * NB: 1281 * We can't simply call ip_input() directly because 1282 * the ip_mforward() depends on the `input interface' 1283 * being set to something unreasonable so that we don't 1284 * attempt to forward the looped-back copy. 1285 * It's also not clear whether there are any lingering 1286 * reentrancy problems in other areas which might be 1287 * exposed by this code. For the moment, we'll err 1288 * on the side of safety by continuing to abuse 1289 * loinput(). 1290 */ 1291#ifdef notdef 1292 copym->m_pkthdr.rcvif = &loif[0]; 1293 ip_input(copym) 1294#else 1295 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1296#endif 1297 } 1298} 1299