ip_output.c revision 22212
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 * $FreeBSD: head/sys/netinet/ip_output.c 22212 1997-02-02 16:33:12Z brian $ 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_ifaddrhead.tqh_first; ia; 231 ia = ia->ia_link.tqe_next) 232 if (ia->ia_ifp == ifp) { 233 ip->ip_src = IA_SIN(ia)->sin_addr; 234 break; 235 } 236 } 237 238 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 239 if (inm != NULL && 240 (imo == NULL || imo->imo_multicast_loop)) { 241 /* 242 * If we belong to the destination multicast group 243 * on the outgoing interface, and the caller did not 244 * forbid loopback, loop back a copy. 245 */ 246 ip_mloopback(ifp, m, dst); 247 } 248 else { 249 /* 250 * If we are acting as a multicast router, perform 251 * multicast forwarding as if the packet had just 252 * arrived on the interface to which we are about 253 * to send. The multicast forwarding function 254 * recursively calls this function, using the 255 * IP_FORWARDING flag to prevent infinite recursion. 256 * 257 * Multicasts that are looped back by ip_mloopback(), 258 * above, will be forwarded by the ip_input() routine, 259 * if necessary. 260 */ 261 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 262 /* 263 * Check if rsvp daemon is running. If not, don't 264 * set ip_moptions. This ensures that the packet 265 * is multicast and not just sent down one link 266 * as prescribed by rsvpd. 267 */ 268 if (!rsvp_on) 269 imo = NULL; 270 if (ip_mforward(ip, ifp, m, imo) != 0) { 271 m_freem(m); 272 goto done; 273 } 274 } 275 } 276 277 /* 278 * Multicasts with a time-to-live of zero may be looped- 279 * back, above, but must not be transmitted on a network. 280 * Also, multicasts addressed to the loopback interface 281 * are not sent -- the above call to ip_mloopback() will 282 * loop back a copy if this host actually belongs to the 283 * destination group on the loopback interface. 284 */ 285 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 286 m_freem(m); 287 goto done; 288 } 289 290 goto sendit; 291 } 292#ifndef notdef 293 /* 294 * If source address not specified yet, use address 295 * of outgoing interface. 296 */ 297 if (ip->ip_src.s_addr == INADDR_ANY) 298 ip->ip_src = IA_SIN(ia)->sin_addr; 299#endif 300 /* 301 * Verify that we have any chance at all of being able to queue 302 * the packet or packet fragments 303 */ 304 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 305 ifp->if_snd.ifq_maxlen) { 306 error = ENOBUFS; 307 goto bad; 308 } 309 310 /* 311 * Look for broadcast address and 312 * and verify user is allowed to send 313 * such a packet. 314 */ 315 if (isbroadcast) { 316 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 317 error = EADDRNOTAVAIL; 318 goto bad; 319 } 320 if ((flags & IP_ALLOWBROADCAST) == 0) { 321 error = EACCES; 322 goto bad; 323 } 324 /* don't allow broadcast messages to be fragmented */ 325 if ((u_short)ip->ip_len > ifp->if_mtu) { 326 error = EMSGSIZE; 327 goto bad; 328 } 329 m->m_flags |= M_BCAST; 330 } else { 331 m->m_flags &= ~M_BCAST; 332 } 333 334sendit: 335 /* 336 * IpHack's section. 337 * - Xlate: translate packet's addr/port (NAT). 338 * - Firewall: deny/allow 339 * - Wrap: fake packet's addr/port <unimpl.> 340 * - Encapsulate: put it in another IP and send out. <unimp.> 341 */ 342 343#ifdef COMPAT_IPFW 344 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, ifp, IP_NAT_OUT)) { 345 error = EACCES; 346 goto done; 347 } 348 349 /* 350 * Check with the firewall... 351 */ 352 if (ip_fw_chk_ptr) { 353 int action; 354 355#ifdef IPDIVERT 356 action = (*ip_fw_chk_ptr)(&ip, 357 hlen, ifp, (~0 << 16) | ip_divert_ignore, &m); 358 ip_divert_ignore = 0; 359#else 360 action = (*ip_fw_chk_ptr)(&ip, hlen, ifp, (~0 << 16), &m); 361#endif 362 if (action == -1) { 363 error = EACCES; /* XXX is this appropriate? */ 364 goto done; 365 } else if (action != 0) { 366#ifdef IPDIVERT 367 ip_divert_port = action; /* divert to port */ 368 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 369 goto done; 370#else 371 m_freem(m); /* ipfw says divert, but we can't */ 372 goto done; 373#endif 374 } 375 } 376#endif /* COMPAT_IPFW */ 377 378 /* 379 * If small enough for interface, can just send directly. 380 */ 381 if ((u_short)ip->ip_len <= ifp->if_mtu) { 382 ip->ip_len = htons((u_short)ip->ip_len); 383 ip->ip_off = htons((u_short)ip->ip_off); 384 ip->ip_sum = 0; 385 if (ip->ip_vhl == IP_VHL_BORING) { 386 ip->ip_sum = in_cksum_hdr(ip); 387 } else { 388 ip->ip_sum = in_cksum(m, hlen); 389 } 390 error = (*ifp->if_output)(ifp, m, 391 (struct sockaddr *)dst, ro->ro_rt); 392 goto done; 393 } 394 /* 395 * Too large for interface; fragment if possible. 396 * Must be able to put at least 8 bytes per fragment. 397 */ 398 if (ip->ip_off & IP_DF) { 399 error = EMSGSIZE; 400 /* 401 * This case can happen if the user changed the MTU 402 * of an interface after enabling IP on it. Because 403 * most netifs don't keep track of routes pointing to 404 * them, there is no way for one to update all its 405 * routes when the MTU is changed. 406 */ 407 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 408 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 409 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 410 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 411 } 412 ipstat.ips_cantfrag++; 413 goto bad; 414 } 415 len = (ifp->if_mtu - hlen) &~ 7; 416 if (len < 8) { 417 error = EMSGSIZE; 418 goto bad; 419 } 420 421 { 422 int mhlen, firstlen = len; 423 struct mbuf **mnext = &m->m_nextpkt; 424 425 /* 426 * Loop through length of segment after first fragment, 427 * make new header and copy data of each part and link onto chain. 428 */ 429 m0 = m; 430 mhlen = sizeof (struct ip); 431 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 432 MGETHDR(m, M_DONTWAIT, MT_HEADER); 433 if (m == 0) { 434 error = ENOBUFS; 435 ipstat.ips_odropped++; 436 goto sendorfree; 437 } 438 m->m_data += max_linkhdr; 439 mhip = mtod(m, struct ip *); 440 *mhip = *ip; 441 if (hlen > sizeof (struct ip)) { 442 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 443 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 444 } 445 m->m_len = mhlen; 446 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 447 if (ip->ip_off & IP_MF) 448 mhip->ip_off |= IP_MF; 449 if (off + len >= (u_short)ip->ip_len) 450 len = (u_short)ip->ip_len - off; 451 else 452 mhip->ip_off |= IP_MF; 453 mhip->ip_len = htons((u_short)(len + mhlen)); 454 m->m_next = m_copy(m0, off, len); 455 if (m->m_next == 0) { 456 (void) m_free(m); 457 error = ENOBUFS; /* ??? */ 458 ipstat.ips_odropped++; 459 goto sendorfree; 460 } 461 m->m_pkthdr.len = mhlen + len; 462 m->m_pkthdr.rcvif = (struct ifnet *)0; 463 mhip->ip_off = htons((u_short)mhip->ip_off); 464 mhip->ip_sum = 0; 465 if (mhip->ip_vhl == IP_VHL_BORING) { 466 mhip->ip_sum = in_cksum_hdr(mhip); 467 } else { 468 mhip->ip_sum = in_cksum(m, mhlen); 469 } 470 *mnext = m; 471 mnext = &m->m_nextpkt; 472 ipstat.ips_ofragments++; 473 } 474 /* 475 * Update first fragment by trimming what's been copied out 476 * and updating header, then send each fragment (in order). 477 */ 478 m = m0; 479 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 480 m->m_pkthdr.len = hlen + firstlen; 481 ip->ip_len = htons((u_short)m->m_pkthdr.len); 482 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 483 ip->ip_sum = 0; 484 if (ip->ip_vhl == IP_VHL_BORING) { 485 ip->ip_sum = in_cksum_hdr(ip); 486 } else { 487 ip->ip_sum = in_cksum(m, hlen); 488 } 489sendorfree: 490 for (m = m0; m; m = m0) { 491 m0 = m->m_nextpkt; 492 m->m_nextpkt = 0; 493 if (error == 0) 494 error = (*ifp->if_output)(ifp, m, 495 (struct sockaddr *)dst, ro->ro_rt); 496 else 497 m_freem(m); 498 } 499 500 if (error == 0) 501 ipstat.ips_fragmented++; 502 } 503done: 504 return (error); 505bad: 506 m_freem(m0); 507 goto done; 508} 509 510/* 511 * Insert IP options into preformed packet. 512 * Adjust IP destination as required for IP source routing, 513 * as indicated by a non-zero in_addr at the start of the options. 514 * 515 * XXX This routine assumes that the packet has no options in place. 516 */ 517static struct mbuf * 518ip_insertoptions(m, opt, phlen) 519 register struct mbuf *m; 520 struct mbuf *opt; 521 int *phlen; 522{ 523 register struct ipoption *p = mtod(opt, struct ipoption *); 524 struct mbuf *n; 525 register struct ip *ip = mtod(m, struct ip *); 526 unsigned optlen; 527 528 optlen = opt->m_len - sizeof(p->ipopt_dst); 529 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 530 return (m); /* XXX should fail */ 531 if (p->ipopt_dst.s_addr) 532 ip->ip_dst = p->ipopt_dst; 533 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 534 MGETHDR(n, M_DONTWAIT, MT_HEADER); 535 if (n == 0) 536 return (m); 537 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 538 m->m_len -= sizeof(struct ip); 539 m->m_data += sizeof(struct ip); 540 n->m_next = m; 541 m = n; 542 m->m_len = optlen + sizeof(struct ip); 543 m->m_data += max_linkhdr; 544 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 545 } else { 546 m->m_data -= optlen; 547 m->m_len += optlen; 548 m->m_pkthdr.len += optlen; 549 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 550 } 551 ip = mtod(m, struct ip *); 552 bcopy(p->ipopt_list, ip + 1, optlen); 553 *phlen = sizeof(struct ip) + optlen; 554 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 555 ip->ip_len += optlen; 556 return (m); 557} 558 559/* 560 * Copy options from ip to jp, 561 * omitting those not copied during fragmentation. 562 */ 563static int 564ip_optcopy(ip, jp) 565 struct ip *ip, *jp; 566{ 567 register u_char *cp, *dp; 568 int opt, optlen, cnt; 569 570 cp = (u_char *)(ip + 1); 571 dp = (u_char *)(jp + 1); 572 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 573 for (; cnt > 0; cnt -= optlen, cp += optlen) { 574 opt = cp[0]; 575 if (opt == IPOPT_EOL) 576 break; 577 if (opt == IPOPT_NOP) { 578 /* Preserve for IP mcast tunnel's LSRR alignment. */ 579 *dp++ = IPOPT_NOP; 580 optlen = 1; 581 continue; 582 } else 583 optlen = cp[IPOPT_OLEN]; 584 /* bogus lengths should have been caught by ip_dooptions */ 585 if (optlen > cnt) 586 optlen = cnt; 587 if (IPOPT_COPIED(opt)) { 588 bcopy(cp, dp, optlen); 589 dp += optlen; 590 } 591 } 592 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 593 *dp++ = IPOPT_EOL; 594 return (optlen); 595} 596 597/* 598 * IP socket option processing. 599 */ 600int 601ip_ctloutput(op, so, level, optname, mp) 602 int op; 603 struct socket *so; 604 int level, optname; 605 struct mbuf **mp; 606{ 607 register struct inpcb *inp = sotoinpcb(so); 608 register struct mbuf *m = *mp; 609 register int optval = 0; 610 int error = 0; 611 612 if (level != IPPROTO_IP) { 613 error = EINVAL; 614 if (op == PRCO_SETOPT && *mp) 615 (void) m_free(*mp); 616 } else switch (op) { 617 618 case PRCO_SETOPT: 619 switch (optname) { 620 case IP_OPTIONS: 621#ifdef notyet 622 case IP_RETOPTS: 623 return (ip_pcbopts(optname, &inp->inp_options, m)); 624#else 625 return (ip_pcbopts(&inp->inp_options, m)); 626#endif 627 628 case IP_TOS: 629 case IP_TTL: 630 case IP_RECVOPTS: 631 case IP_RECVRETOPTS: 632 case IP_RECVDSTADDR: 633 case IP_RECVIF: 634 if (m == 0 || m->m_len != sizeof(int)) 635 error = EINVAL; 636 else { 637 optval = *mtod(m, int *); 638 switch (optname) { 639 640 case IP_TOS: 641 inp->inp_ip.ip_tos = optval; 642 break; 643 644 case IP_TTL: 645 inp->inp_ip.ip_ttl = optval; 646 break; 647#define OPTSET(bit) \ 648 if (optval) \ 649 inp->inp_flags |= bit; \ 650 else \ 651 inp->inp_flags &= ~bit; 652 653 case IP_RECVOPTS: 654 OPTSET(INP_RECVOPTS); 655 break; 656 657 case IP_RECVRETOPTS: 658 OPTSET(INP_RECVRETOPTS); 659 break; 660 661 case IP_RECVDSTADDR: 662 OPTSET(INP_RECVDSTADDR); 663 break; 664 665 case IP_RECVIF: 666 OPTSET(INP_RECVIF); 667 break; 668 } 669 } 670 break; 671#undef OPTSET 672 673 case IP_MULTICAST_IF: 674 case IP_MULTICAST_VIF: 675 case IP_MULTICAST_TTL: 676 case IP_MULTICAST_LOOP: 677 case IP_ADD_MEMBERSHIP: 678 case IP_DROP_MEMBERSHIP: 679 error = ip_setmoptions(optname, &inp->inp_moptions, m); 680 break; 681 682 case IP_PORTRANGE: 683 if (m == 0 || m->m_len != sizeof(int)) 684 error = EINVAL; 685 else { 686 optval = *mtod(m, int *); 687 688 switch (optval) { 689 690 case IP_PORTRANGE_DEFAULT: 691 inp->inp_flags &= ~(INP_LOWPORT); 692 inp->inp_flags &= ~(INP_HIGHPORT); 693 break; 694 695 case IP_PORTRANGE_HIGH: 696 inp->inp_flags &= ~(INP_LOWPORT); 697 inp->inp_flags |= INP_HIGHPORT; 698 break; 699 700 case IP_PORTRANGE_LOW: 701 inp->inp_flags &= ~(INP_HIGHPORT); 702 inp->inp_flags |= INP_LOWPORT; 703 break; 704 705 default: 706 error = EINVAL; 707 break; 708 } 709 } 710 break; 711 712 default: 713 error = ENOPROTOOPT; 714 break; 715 } 716 if (m) 717 (void)m_free(m); 718 break; 719 720 case PRCO_GETOPT: 721 switch (optname) { 722 case IP_OPTIONS: 723 case IP_RETOPTS: 724 *mp = m = m_get(M_WAIT, MT_SOOPTS); 725 if (inp->inp_options) { 726 m->m_len = inp->inp_options->m_len; 727 bcopy(mtod(inp->inp_options, void *), 728 mtod(m, void *), m->m_len); 729 } else 730 m->m_len = 0; 731 break; 732 733 case IP_TOS: 734 case IP_TTL: 735 case IP_RECVOPTS: 736 case IP_RECVRETOPTS: 737 case IP_RECVDSTADDR: 738 case IP_RECVIF: 739 *mp = m = m_get(M_WAIT, MT_SOOPTS); 740 m->m_len = sizeof(int); 741 switch (optname) { 742 743 case IP_TOS: 744 optval = inp->inp_ip.ip_tos; 745 break; 746 747 case IP_TTL: 748 optval = inp->inp_ip.ip_ttl; 749 break; 750 751#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 752 753 case IP_RECVOPTS: 754 optval = OPTBIT(INP_RECVOPTS); 755 break; 756 757 case IP_RECVRETOPTS: 758 optval = OPTBIT(INP_RECVRETOPTS); 759 break; 760 761 case IP_RECVDSTADDR: 762 optval = OPTBIT(INP_RECVDSTADDR); 763 break; 764 765 case IP_RECVIF: 766 optval = OPTBIT(INP_RECVIF); 767 break; 768 } 769 *mtod(m, int *) = optval; 770 break; 771 772 case IP_MULTICAST_IF: 773 case IP_MULTICAST_VIF: 774 case IP_MULTICAST_TTL: 775 case IP_MULTICAST_LOOP: 776 case IP_ADD_MEMBERSHIP: 777 case IP_DROP_MEMBERSHIP: 778 error = ip_getmoptions(optname, inp->inp_moptions, mp); 779 break; 780 781 case IP_PORTRANGE: 782 *mp = m = m_get(M_WAIT, MT_SOOPTS); 783 m->m_len = sizeof(int); 784 785 if (inp->inp_flags & INP_HIGHPORT) 786 optval = IP_PORTRANGE_HIGH; 787 else if (inp->inp_flags & INP_LOWPORT) 788 optval = IP_PORTRANGE_LOW; 789 else 790 optval = 0; 791 792 *mtod(m, int *) = optval; 793 break; 794 795 default: 796 error = ENOPROTOOPT; 797 break; 798 } 799 break; 800 } 801 return (error); 802} 803 804/* 805 * Set up IP options in pcb for insertion in output packets. 806 * Store in mbuf with pointer in pcbopt, adding pseudo-option 807 * with destination address if source routed. 808 */ 809static int 810#ifdef notyet 811ip_pcbopts(optname, pcbopt, m) 812 int optname; 813#else 814ip_pcbopts(pcbopt, m) 815#endif 816 struct mbuf **pcbopt; 817 register struct mbuf *m; 818{ 819 register cnt, optlen; 820 register u_char *cp; 821 u_char opt; 822 823 /* turn off any old options */ 824 if (*pcbopt) 825 (void)m_free(*pcbopt); 826 *pcbopt = 0; 827 if (m == (struct mbuf *)0 || m->m_len == 0) { 828 /* 829 * Only turning off any previous options. 830 */ 831 if (m) 832 (void)m_free(m); 833 return (0); 834 } 835 836#ifndef vax 837 if (m->m_len % sizeof(long)) 838 goto bad; 839#endif 840 /* 841 * IP first-hop destination address will be stored before 842 * actual options; move other options back 843 * and clear it when none present. 844 */ 845 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 846 goto bad; 847 cnt = m->m_len; 848 m->m_len += sizeof(struct in_addr); 849 cp = mtod(m, u_char *) + sizeof(struct in_addr); 850 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 851 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 852 853 for (; cnt > 0; cnt -= optlen, cp += optlen) { 854 opt = cp[IPOPT_OPTVAL]; 855 if (opt == IPOPT_EOL) 856 break; 857 if (opt == IPOPT_NOP) 858 optlen = 1; 859 else { 860 optlen = cp[IPOPT_OLEN]; 861 if (optlen <= IPOPT_OLEN || optlen > cnt) 862 goto bad; 863 } 864 switch (opt) { 865 866 default: 867 break; 868 869 case IPOPT_LSRR: 870 case IPOPT_SSRR: 871 /* 872 * user process specifies route as: 873 * ->A->B->C->D 874 * D must be our final destination (but we can't 875 * check that since we may not have connected yet). 876 * A is first hop destination, which doesn't appear in 877 * actual IP option, but is stored before the options. 878 */ 879 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 880 goto bad; 881 m->m_len -= sizeof(struct in_addr); 882 cnt -= sizeof(struct in_addr); 883 optlen -= sizeof(struct in_addr); 884 cp[IPOPT_OLEN] = optlen; 885 /* 886 * Move first hop before start of options. 887 */ 888 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 889 sizeof(struct in_addr)); 890 /* 891 * Then copy rest of options back 892 * to close up the deleted entry. 893 */ 894 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 895 sizeof(struct in_addr)), 896 (caddr_t)&cp[IPOPT_OFFSET+1], 897 (unsigned)cnt + sizeof(struct in_addr)); 898 break; 899 } 900 } 901 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 902 goto bad; 903 *pcbopt = m; 904 return (0); 905 906bad: 907 (void)m_free(m); 908 return (EINVAL); 909} 910 911/* 912 * Set the IP multicast options in response to user setsockopt(). 913 */ 914static int 915ip_setmoptions(optname, imop, m) 916 int optname; 917 struct ip_moptions **imop; 918 struct mbuf *m; 919{ 920 register int error = 0; 921 u_char loop; 922 register int i; 923 struct in_addr addr; 924 register struct ip_mreq *mreq; 925 register struct ifnet *ifp; 926 register struct ip_moptions *imo = *imop; 927 struct route ro; 928 register struct sockaddr_in *dst; 929 int s; 930 931 if (imo == NULL) { 932 /* 933 * No multicast option buffer attached to the pcb; 934 * allocate one and initialize to default values. 935 */ 936 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 937 M_WAITOK); 938 939 if (imo == NULL) 940 return (ENOBUFS); 941 *imop = imo; 942 imo->imo_multicast_ifp = NULL; 943 imo->imo_multicast_vif = -1; 944 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 945 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 946 imo->imo_num_memberships = 0; 947 } 948 949 switch (optname) { 950 /* store an index number for the vif you wanna use in the send */ 951 case IP_MULTICAST_VIF: 952 if (!legal_vif_num) { 953 error = EOPNOTSUPP; 954 break; 955 } 956 if (m == NULL || m->m_len != sizeof(int)) { 957 error = EINVAL; 958 break; 959 } 960 i = *(mtod(m, int *)); 961 if (!legal_vif_num(i) && (i != -1)) { 962 error = EINVAL; 963 break; 964 } 965 imo->imo_multicast_vif = i; 966 break; 967 968 case IP_MULTICAST_IF: 969 /* 970 * Select the interface for outgoing multicast packets. 971 */ 972 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 973 error = EINVAL; 974 break; 975 } 976 addr = *(mtod(m, struct in_addr *)); 977 /* 978 * INADDR_ANY is used to remove a previous selection. 979 * When no interface is selected, a default one is 980 * chosen every time a multicast packet is sent. 981 */ 982 if (addr.s_addr == INADDR_ANY) { 983 imo->imo_multicast_ifp = NULL; 984 break; 985 } 986 /* 987 * The selected interface is identified by its local 988 * IP address. Find the interface and confirm that 989 * it supports multicasting. 990 */ 991 s = splimp(); 992 INADDR_TO_IFP(addr, ifp); 993 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 994 splx(s); 995 error = EADDRNOTAVAIL; 996 break; 997 } 998 imo->imo_multicast_ifp = ifp; 999 splx(s); 1000 break; 1001 1002 case IP_MULTICAST_TTL: 1003 /* 1004 * Set the IP time-to-live for outgoing multicast packets. 1005 */ 1006 if (m == NULL || m->m_len != 1) { 1007 error = EINVAL; 1008 break; 1009 } 1010 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1011 break; 1012 1013 case IP_MULTICAST_LOOP: 1014 /* 1015 * Set the loopback flag for outgoing multicast packets. 1016 * Must be zero or one. 1017 */ 1018 if (m == NULL || m->m_len != 1 || 1019 (loop = *(mtod(m, u_char *))) > 1) { 1020 error = EINVAL; 1021 break; 1022 } 1023 imo->imo_multicast_loop = loop; 1024 break; 1025 1026 case IP_ADD_MEMBERSHIP: 1027 /* 1028 * Add a multicast group membership. 1029 * Group must be a valid IP multicast address. 1030 */ 1031 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1032 error = EINVAL; 1033 break; 1034 } 1035 mreq = mtod(m, struct ip_mreq *); 1036 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1037 error = EINVAL; 1038 break; 1039 } 1040 s = splimp(); 1041 /* 1042 * If no interface address was provided, use the interface of 1043 * the route to the given multicast address. 1044 */ 1045 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1046 bzero((caddr_t)&ro, sizeof(ro)); 1047 dst = (struct sockaddr_in *)&ro.ro_dst; 1048 dst->sin_len = sizeof(*dst); 1049 dst->sin_family = AF_INET; 1050 dst->sin_addr = mreq->imr_multiaddr; 1051 rtalloc(&ro); 1052 if (ro.ro_rt == NULL) { 1053 error = EADDRNOTAVAIL; 1054 splx(s); 1055 break; 1056 } 1057 ifp = ro.ro_rt->rt_ifp; 1058 rtfree(ro.ro_rt); 1059 } 1060 else { 1061 INADDR_TO_IFP(mreq->imr_interface, ifp); 1062 } 1063 1064 /* 1065 * See if we found an interface, and confirm that it 1066 * supports multicast. 1067 */ 1068 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1069 error = EADDRNOTAVAIL; 1070 splx(s); 1071 break; 1072 } 1073 /* 1074 * See if the membership already exists or if all the 1075 * membership slots are full. 1076 */ 1077 for (i = 0; i < imo->imo_num_memberships; ++i) { 1078 if (imo->imo_membership[i]->inm_ifp == ifp && 1079 imo->imo_membership[i]->inm_addr.s_addr 1080 == mreq->imr_multiaddr.s_addr) 1081 break; 1082 } 1083 if (i < imo->imo_num_memberships) { 1084 error = EADDRINUSE; 1085 splx(s); 1086 break; 1087 } 1088 if (i == IP_MAX_MEMBERSHIPS) { 1089 error = ETOOMANYREFS; 1090 splx(s); 1091 break; 1092 } 1093 /* 1094 * Everything looks good; add a new record to the multicast 1095 * address list for the given interface. 1096 */ 1097 if ((imo->imo_membership[i] = 1098 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1099 error = ENOBUFS; 1100 splx(s); 1101 break; 1102 } 1103 ++imo->imo_num_memberships; 1104 splx(s); 1105 break; 1106 1107 case IP_DROP_MEMBERSHIP: 1108 /* 1109 * Drop a multicast group membership. 1110 * Group must be a valid IP multicast address. 1111 */ 1112 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1113 error = EINVAL; 1114 break; 1115 } 1116 mreq = mtod(m, struct ip_mreq *); 1117 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1118 error = EINVAL; 1119 break; 1120 } 1121 1122 s = splimp(); 1123 /* 1124 * If an interface address was specified, get a pointer 1125 * to its ifnet structure. 1126 */ 1127 if (mreq->imr_interface.s_addr == INADDR_ANY) 1128 ifp = NULL; 1129 else { 1130 INADDR_TO_IFP(mreq->imr_interface, ifp); 1131 if (ifp == NULL) { 1132 error = EADDRNOTAVAIL; 1133 splx(s); 1134 break; 1135 } 1136 } 1137 /* 1138 * Find the membership in the membership array. 1139 */ 1140 for (i = 0; i < imo->imo_num_memberships; ++i) { 1141 if ((ifp == NULL || 1142 imo->imo_membership[i]->inm_ifp == ifp) && 1143 imo->imo_membership[i]->inm_addr.s_addr == 1144 mreq->imr_multiaddr.s_addr) 1145 break; 1146 } 1147 if (i == imo->imo_num_memberships) { 1148 error = EADDRNOTAVAIL; 1149 splx(s); 1150 break; 1151 } 1152 /* 1153 * Give up the multicast address record to which the 1154 * membership points. 1155 */ 1156 in_delmulti(imo->imo_membership[i]); 1157 /* 1158 * Remove the gap in the membership array. 1159 */ 1160 for (++i; i < imo->imo_num_memberships; ++i) 1161 imo->imo_membership[i-1] = imo->imo_membership[i]; 1162 --imo->imo_num_memberships; 1163 splx(s); 1164 break; 1165 1166 default: 1167 error = EOPNOTSUPP; 1168 break; 1169 } 1170 1171 /* 1172 * If all options have default values, no need to keep the mbuf. 1173 */ 1174 if (imo->imo_multicast_ifp == NULL && 1175 imo->imo_multicast_vif == -1 && 1176 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1177 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1178 imo->imo_num_memberships == 0) { 1179 free(*imop, M_IPMOPTS); 1180 *imop = NULL; 1181 } 1182 1183 return (error); 1184} 1185 1186/* 1187 * Return the IP multicast options in response to user getsockopt(). 1188 */ 1189static int 1190ip_getmoptions(optname, imo, mp) 1191 int optname; 1192 register struct ip_moptions *imo; 1193 register struct mbuf **mp; 1194{ 1195 u_char *ttl; 1196 u_char *loop; 1197 struct in_addr *addr; 1198 struct in_ifaddr *ia; 1199 1200 *mp = m_get(M_WAIT, MT_SOOPTS); 1201 1202 switch (optname) { 1203 1204 case IP_MULTICAST_VIF: 1205 if (imo != NULL) 1206 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1207 else 1208 *(mtod(*mp, int *)) = -1; 1209 (*mp)->m_len = sizeof(int); 1210 return(0); 1211 1212 case IP_MULTICAST_IF: 1213 addr = mtod(*mp, struct in_addr *); 1214 (*mp)->m_len = sizeof(struct in_addr); 1215 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1216 addr->s_addr = INADDR_ANY; 1217 else { 1218 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1219 addr->s_addr = (ia == NULL) ? INADDR_ANY 1220 : IA_SIN(ia)->sin_addr.s_addr; 1221 } 1222 return (0); 1223 1224 case IP_MULTICAST_TTL: 1225 ttl = mtod(*mp, u_char *); 1226 (*mp)->m_len = 1; 1227 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1228 : imo->imo_multicast_ttl; 1229 return (0); 1230 1231 case IP_MULTICAST_LOOP: 1232 loop = mtod(*mp, u_char *); 1233 (*mp)->m_len = 1; 1234 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1235 : imo->imo_multicast_loop; 1236 return (0); 1237 1238 default: 1239 return (EOPNOTSUPP); 1240 } 1241} 1242 1243/* 1244 * Discard the IP multicast options. 1245 */ 1246void 1247ip_freemoptions(imo) 1248 register struct ip_moptions *imo; 1249{ 1250 register int i; 1251 1252 if (imo != NULL) { 1253 for (i = 0; i < imo->imo_num_memberships; ++i) 1254 in_delmulti(imo->imo_membership[i]); 1255 free(imo, M_IPMOPTS); 1256 } 1257} 1258 1259/* 1260 * Routine called from ip_output() to loop back a copy of an IP multicast 1261 * packet to the input queue of a specified interface. Note that this 1262 * calls the output routine of the loopback "driver", but with an interface 1263 * pointer that might NOT be a loopback interface -- evil, but easier than 1264 * replicating that code here. 1265 */ 1266static void 1267ip_mloopback(ifp, m, dst) 1268 struct ifnet *ifp; 1269 register struct mbuf *m; 1270 register struct sockaddr_in *dst; 1271{ 1272 register struct ip *ip; 1273 struct mbuf *copym; 1274 1275 copym = m_copy(m, 0, M_COPYALL); 1276 if (copym != NULL) { 1277 /* 1278 * We don't bother to fragment if the IP length is greater 1279 * than the interface's MTU. Can this possibly matter? 1280 */ 1281 ip = mtod(copym, struct ip *); 1282 ip->ip_len = htons((u_short)ip->ip_len); 1283 ip->ip_off = htons((u_short)ip->ip_off); 1284 ip->ip_sum = 0; 1285 if (ip->ip_vhl == IP_VHL_BORING) { 1286 ip->ip_sum = in_cksum_hdr(ip); 1287 } else { 1288 ip->ip_sum = in_cksum(copym, 1289 IP_VHL_HL(ip->ip_vhl) << 2); 1290 } 1291 /* 1292 * NB: 1293 * We can't simply call ip_input() directly because 1294 * the ip_mforward() depends on the `input interface' 1295 * being set to something unreasonable so that we don't 1296 * attempt to forward the looped-back copy. 1297 * It's also not clear whether there are any lingering 1298 * reentrancy problems in other areas which might be 1299 * exposed by this code. For the moment, we'll err 1300 * on the side of safety by continuing to abuse 1301 * loinput(). 1302 */ 1303#ifdef notdef 1304 copym->m_pkthdr.rcvif = &loif[0]; 1305 ip_input(copym) 1306#else 1307 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1308#endif 1309 } 1310} 1311