ip_output.c revision 20407
1170890Syar/* 2170890Syar * Copyright (c) 1982, 1986, 1988, 1990, 1993 3170890Syar * The Regents of the University of California. All rights reserved. 4170890Syar * 5170890Syar * Redistribution and use in source and binary forms, with or without 6170890Syar * modification, are permitted provided that the following conditions 7170890Syar * are met: 8170890Syar * 1. Redistributions of source code must retain the above copyright 9170890Syar * 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 1.45 1996/11/11 04:56:19 fenner 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_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#else 359 action = (*ip_fw_chk_ptr)(&ip, hlen, ifp, (~0 << 16), &m); 360#endif 361 if (action == -1) { 362 error = EACCES; /* XXX is this appropriate? */ 363 goto done; 364 } else if (action != 0) { 365#ifdef IPDIVERT 366 ip_divert_port = action; /* divert to port */ 367 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 368 goto done; 369#else 370 m_freem(m); /* ipfw says divert, but we can't */ 371 goto done; 372#endif 373 } 374 } 375#endif /* COMPAT_IPFW */ 376 377 /* 378 * If small enough for interface, can just send directly. 379 */ 380 if ((u_short)ip->ip_len <= ifp->if_mtu) { 381 ip->ip_len = htons((u_short)ip->ip_len); 382 ip->ip_off = htons((u_short)ip->ip_off); 383 ip->ip_sum = 0; 384 if (ip->ip_vhl == IP_VHL_BORING) { 385 ip->ip_sum = in_cksum_hdr(ip); 386 } else { 387 ip->ip_sum = in_cksum(m, hlen); 388 } 389 error = (*ifp->if_output)(ifp, m, 390 (struct sockaddr *)dst, ro->ro_rt); 391 goto done; 392 } 393 /* 394 * Too large for interface; fragment if possible. 395 * Must be able to put at least 8 bytes per fragment. 396 */ 397 if (ip->ip_off & IP_DF) { 398 error = EMSGSIZE; 399 /* 400 * This case can happen if the user changed the MTU 401 * of an interface after enabling IP on it. Because 402 * most netifs don't keep track of routes pointing to 403 * them, there is no way for one to update all its 404 * routes when the MTU is changed. 405 */ 406 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 407 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 408 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 409 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 410 } 411 ipstat.ips_cantfrag++; 412 goto bad; 413 } 414 len = (ifp->if_mtu - hlen) &~ 7; 415 if (len < 8) { 416 error = EMSGSIZE; 417 goto bad; 418 } 419 420 { 421 int mhlen, firstlen = len; 422 struct mbuf **mnext = &m->m_nextpkt; 423 424 /* 425 * Loop through length of segment after first fragment, 426 * make new header and copy data of each part and link onto chain. 427 */ 428 m0 = m; 429 mhlen = sizeof (struct ip); 430 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 431 MGETHDR(m, M_DONTWAIT, MT_HEADER); 432 if (m == 0) { 433 error = ENOBUFS; 434 ipstat.ips_odropped++; 435 goto sendorfree; 436 } 437 m->m_data += max_linkhdr; 438 mhip = mtod(m, struct ip *); 439 *mhip = *ip; 440 if (hlen > sizeof (struct ip)) { 441 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 442 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 443 } 444 m->m_len = mhlen; 445 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 446 if (ip->ip_off & IP_MF) 447 mhip->ip_off |= IP_MF; 448 if (off + len >= (u_short)ip->ip_len) 449 len = (u_short)ip->ip_len - off; 450 else 451 mhip->ip_off |= IP_MF; 452 mhip->ip_len = htons((u_short)(len + mhlen)); 453 m->m_next = m_copy(m0, off, len); 454 if (m->m_next == 0) { 455 (void) m_free(m); 456 error = ENOBUFS; /* ??? */ 457 ipstat.ips_odropped++; 458 goto sendorfree; 459 } 460 m->m_pkthdr.len = mhlen + len; 461 m->m_pkthdr.rcvif = (struct ifnet *)0; 462 mhip->ip_off = htons((u_short)mhip->ip_off); 463 mhip->ip_sum = 0; 464 if (mhip->ip_vhl == IP_VHL_BORING) { 465 mhip->ip_sum = in_cksum_hdr(mhip); 466 } else { 467 mhip->ip_sum = in_cksum(m, mhlen); 468 } 469 *mnext = m; 470 mnext = &m->m_nextpkt; 471 ipstat.ips_ofragments++; 472 } 473 /* 474 * Update first fragment by trimming what's been copied out 475 * and updating header, then send each fragment (in order). 476 */ 477 m = m0; 478 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 479 m->m_pkthdr.len = hlen + firstlen; 480 ip->ip_len = htons((u_short)m->m_pkthdr.len); 481 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 482 ip->ip_sum = 0; 483 if (ip->ip_vhl == IP_VHL_BORING) { 484 ip->ip_sum = in_cksum_hdr(ip); 485 } else { 486 ip->ip_sum = in_cksum(m, hlen); 487 } 488sendorfree: 489 for (m = m0; m; m = m0) { 490 m0 = m->m_nextpkt; 491 m->m_nextpkt = 0; 492 if (error == 0) 493 error = (*ifp->if_output)(ifp, m, 494 (struct sockaddr *)dst, ro->ro_rt); 495 else 496 m_freem(m); 497 } 498 499 if (error == 0) 500 ipstat.ips_fragmented++; 501 } 502done: 503 return (error); 504bad: 505 m_freem(m0); 506 goto done; 507} 508 509/* 510 * Insert IP options into preformed packet. 511 * Adjust IP destination as required for IP source routing, 512 * as indicated by a non-zero in_addr at the start of the options. 513 * 514 * XXX This routine assumes that the packet has no options in place. 515 */ 516static struct mbuf * 517ip_insertoptions(m, opt, phlen) 518 register struct mbuf *m; 519 struct mbuf *opt; 520 int *phlen; 521{ 522 register struct ipoption *p = mtod(opt, struct ipoption *); 523 struct mbuf *n; 524 register struct ip *ip = mtod(m, struct ip *); 525 unsigned optlen; 526 527 optlen = opt->m_len - sizeof(p->ipopt_dst); 528 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 529 return (m); /* XXX should fail */ 530 if (p->ipopt_dst.s_addr) 531 ip->ip_dst = p->ipopt_dst; 532 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 533 MGETHDR(n, M_DONTWAIT, MT_HEADER); 534 if (n == 0) 535 return (m); 536 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 537 m->m_len -= sizeof(struct ip); 538 m->m_data += sizeof(struct ip); 539 n->m_next = m; 540 m = n; 541 m->m_len = optlen + sizeof(struct ip); 542 m->m_data += max_linkhdr; 543 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 544 } else { 545 m->m_data -= optlen; 546 m->m_len += optlen; 547 m->m_pkthdr.len += optlen; 548 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 549 } 550 ip = mtod(m, struct ip *); 551 bcopy(p->ipopt_list, ip + 1, optlen); 552 *phlen = sizeof(struct ip) + optlen; 553 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 554 ip->ip_len += optlen; 555 return (m); 556} 557 558/* 559 * Copy options from ip to jp, 560 * omitting those not copied during fragmentation. 561 */ 562static int 563ip_optcopy(ip, jp) 564 struct ip *ip, *jp; 565{ 566 register u_char *cp, *dp; 567 int opt, optlen, cnt; 568 569 cp = (u_char *)(ip + 1); 570 dp = (u_char *)(jp + 1); 571 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 572 for (; cnt > 0; cnt -= optlen, cp += optlen) { 573 opt = cp[0]; 574 if (opt == IPOPT_EOL) 575 break; 576 if (opt == IPOPT_NOP) { 577 /* Preserve for IP mcast tunnel's LSRR alignment. */ 578 *dp++ = IPOPT_NOP; 579 optlen = 1; 580 continue; 581 } else 582 optlen = cp[IPOPT_OLEN]; 583 /* bogus lengths should have been caught by ip_dooptions */ 584 if (optlen > cnt) 585 optlen = cnt; 586 if (IPOPT_COPIED(opt)) { 587 bcopy(cp, dp, optlen); 588 dp += optlen; 589 } 590 } 591 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 592 *dp++ = IPOPT_EOL; 593 return (optlen); 594} 595 596/* 597 * IP socket option processing. 598 */ 599int 600ip_ctloutput(op, so, level, optname, mp) 601 int op; 602 struct socket *so; 603 int level, optname; 604 struct mbuf **mp; 605{ 606 register struct inpcb *inp = sotoinpcb(so); 607 register struct mbuf *m = *mp; 608 register int optval = 0; 609 int error = 0; 610 611 if (level != IPPROTO_IP) { 612 error = EINVAL; 613 if (op == PRCO_SETOPT && *mp) 614 (void) m_free(*mp); 615 } else switch (op) { 616 617 case PRCO_SETOPT: 618 switch (optname) { 619 case IP_OPTIONS: 620#ifdef notyet 621 case IP_RETOPTS: 622 return (ip_pcbopts(optname, &inp->inp_options, m)); 623#else 624 return (ip_pcbopts(&inp->inp_options, m)); 625#endif 626 627 case IP_TOS: 628 case IP_TTL: 629 case IP_RECVOPTS: 630 case IP_RECVRETOPTS: 631 case IP_RECVDSTADDR: 632 case IP_RECVIF: 633 if (m == 0 || m->m_len != sizeof(int)) 634 error = EINVAL; 635 else { 636 optval = *mtod(m, int *); 637 switch (optname) { 638 639 case IP_TOS: 640 inp->inp_ip.ip_tos = optval; 641 break; 642 643 case IP_TTL: 644 inp->inp_ip.ip_ttl = optval; 645 break; 646#define OPTSET(bit) \ 647 if (optval) \ 648 inp->inp_flags |= bit; \ 649 else \ 650 inp->inp_flags &= ~bit; 651 652 case IP_RECVOPTS: 653 OPTSET(INP_RECVOPTS); 654 break; 655 656 case IP_RECVRETOPTS: 657 OPTSET(INP_RECVRETOPTS); 658 break; 659 660 case IP_RECVDSTADDR: 661 OPTSET(INP_RECVDSTADDR); 662 break; 663 664 case IP_RECVIF: 665 OPTSET(INP_RECVIF); 666 break; 667 } 668 } 669 break; 670#undef OPTSET 671 672 case IP_MULTICAST_IF: 673 case IP_MULTICAST_VIF: 674 case IP_MULTICAST_TTL: 675 case IP_MULTICAST_LOOP: 676 case IP_ADD_MEMBERSHIP: 677 case IP_DROP_MEMBERSHIP: 678 error = ip_setmoptions(optname, &inp->inp_moptions, m); 679 break; 680 681 case IP_PORTRANGE: 682 if (m == 0 || m->m_len != sizeof(int)) 683 error = EINVAL; 684 else { 685 optval = *mtod(m, int *); 686 687 switch (optval) { 688 689 case IP_PORTRANGE_DEFAULT: 690 inp->inp_flags &= ~(INP_LOWPORT); 691 inp->inp_flags &= ~(INP_HIGHPORT); 692 break; 693 694 case IP_PORTRANGE_HIGH: 695 inp->inp_flags &= ~(INP_LOWPORT); 696 inp->inp_flags |= INP_HIGHPORT; 697 break; 698 699 case IP_PORTRANGE_LOW: 700 inp->inp_flags &= ~(INP_HIGHPORT); 701 inp->inp_flags |= INP_LOWPORT; 702 break; 703 704 default: 705 error = EINVAL; 706 break; 707 } 708 } 709 break; 710 711 default: 712 error = ENOPROTOOPT; 713 break; 714 } 715 if (m) 716 (void)m_free(m); 717 break; 718 719 case PRCO_GETOPT: 720 switch (optname) { 721 case IP_OPTIONS: 722 case IP_RETOPTS: 723 *mp = m = m_get(M_WAIT, MT_SOOPTS); 724 if (inp->inp_options) { 725 m->m_len = inp->inp_options->m_len; 726 bcopy(mtod(inp->inp_options, void *), 727 mtod(m, void *), m->m_len); 728 } else 729 m->m_len = 0; 730 break; 731 732 case IP_TOS: 733 case IP_TTL: 734 case IP_RECVOPTS: 735 case IP_RECVRETOPTS: 736 case IP_RECVDSTADDR: 737 case IP_RECVIF: 738 *mp = m = m_get(M_WAIT, MT_SOOPTS); 739 m->m_len = sizeof(int); 740 switch (optname) { 741 742 case IP_TOS: 743 optval = inp->inp_ip.ip_tos; 744 break; 745 746 case IP_TTL: 747 optval = inp->inp_ip.ip_ttl; 748 break; 749 750#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 751 752 case IP_RECVOPTS: 753 optval = OPTBIT(INP_RECVOPTS); 754 break; 755 756 case IP_RECVRETOPTS: 757 optval = OPTBIT(INP_RECVRETOPTS); 758 break; 759 760 case IP_RECVDSTADDR: 761 optval = OPTBIT(INP_RECVDSTADDR); 762 break; 763 764 case IP_RECVIF: 765 optval = OPTBIT(INP_RECVIF); 766 break; 767 } 768 *mtod(m, int *) = optval; 769 break; 770 771 case IP_MULTICAST_IF: 772 case IP_MULTICAST_VIF: 773 case IP_MULTICAST_TTL: 774 case IP_MULTICAST_LOOP: 775 case IP_ADD_MEMBERSHIP: 776 case IP_DROP_MEMBERSHIP: 777 error = ip_getmoptions(optname, inp->inp_moptions, mp); 778 break; 779 780 case IP_PORTRANGE: 781 *mp = m = m_get(M_WAIT, MT_SOOPTS); 782 m->m_len = sizeof(int); 783 784 if (inp->inp_flags & INP_HIGHPORT) 785 optval = IP_PORTRANGE_HIGH; 786 else if (inp->inp_flags & INP_LOWPORT) 787 optval = IP_PORTRANGE_LOW; 788 else 789 optval = 0; 790 791 *mtod(m, int *) = optval; 792 break; 793 794 default: 795 error = ENOPROTOOPT; 796 break; 797 } 798 break; 799 } 800 return (error); 801} 802 803/* 804 * Set up IP options in pcb for insertion in output packets. 805 * Store in mbuf with pointer in pcbopt, adding pseudo-option 806 * with destination address if source routed. 807 */ 808static int 809#ifdef notyet 810ip_pcbopts(optname, pcbopt, m) 811 int optname; 812#else 813ip_pcbopts(pcbopt, m) 814#endif 815 struct mbuf **pcbopt; 816 register struct mbuf *m; 817{ 818 register cnt, optlen; 819 register u_char *cp; 820 u_char opt; 821 822 /* turn off any old options */ 823 if (*pcbopt) 824 (void)m_free(*pcbopt); 825 *pcbopt = 0; 826 if (m == (struct mbuf *)0 || m->m_len == 0) { 827 /* 828 * Only turning off any previous options. 829 */ 830 if (m) 831 (void)m_free(m); 832 return (0); 833 } 834 835#ifndef vax 836 if (m->m_len % sizeof(long)) 837 goto bad; 838#endif 839 /* 840 * IP first-hop destination address will be stored before 841 * actual options; move other options back 842 * and clear it when none present. 843 */ 844 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 845 goto bad; 846 cnt = m->m_len; 847 m->m_len += sizeof(struct in_addr); 848 cp = mtod(m, u_char *) + sizeof(struct in_addr); 849 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 850 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 851 852 for (; cnt > 0; cnt -= optlen, cp += optlen) { 853 opt = cp[IPOPT_OPTVAL]; 854 if (opt == IPOPT_EOL) 855 break; 856 if (opt == IPOPT_NOP) 857 optlen = 1; 858 else { 859 optlen = cp[IPOPT_OLEN]; 860 if (optlen <= IPOPT_OLEN || optlen > cnt) 861 goto bad; 862 } 863 switch (opt) { 864 865 default: 866 break; 867 868 case IPOPT_LSRR: 869 case IPOPT_SSRR: 870 /* 871 * user process specifies route as: 872 * ->A->B->C->D 873 * D must be our final destination (but we can't 874 * check that since we may not have connected yet). 875 * A is first hop destination, which doesn't appear in 876 * actual IP option, but is stored before the options. 877 */ 878 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 879 goto bad; 880 m->m_len -= sizeof(struct in_addr); 881 cnt -= sizeof(struct in_addr); 882 optlen -= sizeof(struct in_addr); 883 cp[IPOPT_OLEN] = optlen; 884 /* 885 * Move first hop before start of options. 886 */ 887 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 888 sizeof(struct in_addr)); 889 /* 890 * Then copy rest of options back 891 * to close up the deleted entry. 892 */ 893 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 894 sizeof(struct in_addr)), 895 (caddr_t)&cp[IPOPT_OFFSET+1], 896 (unsigned)cnt + sizeof(struct in_addr)); 897 break; 898 } 899 } 900 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 901 goto bad; 902 *pcbopt = m; 903 return (0); 904 905bad: 906 (void)m_free(m); 907 return (EINVAL); 908} 909 910/* 911 * Set the IP multicast options in response to user setsockopt(). 912 */ 913static int 914ip_setmoptions(optname, imop, m) 915 int optname; 916 struct ip_moptions **imop; 917 struct mbuf *m; 918{ 919 register int error = 0; 920 u_char loop; 921 register int i; 922 struct in_addr addr; 923 register struct ip_mreq *mreq; 924 register struct ifnet *ifp; 925 register struct ip_moptions *imo = *imop; 926 struct route ro; 927 register struct sockaddr_in *dst; 928 int s; 929 930 if (imo == NULL) { 931 /* 932 * No multicast option buffer attached to the pcb; 933 * allocate one and initialize to default values. 934 */ 935 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 936 M_WAITOK); 937 938 if (imo == NULL) 939 return (ENOBUFS); 940 *imop = imo; 941 imo->imo_multicast_ifp = NULL; 942 imo->imo_multicast_vif = -1; 943 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 944 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 945 imo->imo_num_memberships = 0; 946 } 947 948 switch (optname) { 949 /* store an index number for the vif you wanna use in the send */ 950 case IP_MULTICAST_VIF: 951 if (!legal_vif_num) { 952 error = EOPNOTSUPP; 953 break; 954 } 955 if (m == NULL || m->m_len != sizeof(int)) { 956 error = EINVAL; 957 break; 958 } 959 i = *(mtod(m, int *)); 960 if (!legal_vif_num(i) && (i != -1)) { 961 error = EINVAL; 962 break; 963 } 964 imo->imo_multicast_vif = i; 965 break; 966 967 case IP_MULTICAST_IF: 968 /* 969 * Select the interface for outgoing multicast packets. 970 */ 971 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 972 error = EINVAL; 973 break; 974 } 975 addr = *(mtod(m, struct in_addr *)); 976 /* 977 * INADDR_ANY is used to remove a previous selection. 978 * When no interface is selected, a default one is 979 * chosen every time a multicast packet is sent. 980 */ 981 if (addr.s_addr == INADDR_ANY) { 982 imo->imo_multicast_ifp = NULL; 983 break; 984 } 985 /* 986 * The selected interface is identified by its local 987 * IP address. Find the interface and confirm that 988 * it supports multicasting. 989 */ 990 s = splimp(); 991 INADDR_TO_IFP(addr, ifp); 992 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 993 splx(s); 994 error = EADDRNOTAVAIL; 995 break; 996 } 997 imo->imo_multicast_ifp = ifp; 998 splx(s); 999 break; 1000 1001 case IP_MULTICAST_TTL: 1002 /* 1003 * Set the IP time-to-live for outgoing multicast packets. 1004 */ 1005 if (m == NULL || m->m_len != 1) { 1006 error = EINVAL; 1007 break; 1008 } 1009 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1010 break; 1011 1012 case IP_MULTICAST_LOOP: 1013 /* 1014 * Set the loopback flag for outgoing multicast packets. 1015 * Must be zero or one. 1016 */ 1017 if (m == NULL || m->m_len != 1 || 1018 (loop = *(mtod(m, u_char *))) > 1) { 1019 error = EINVAL; 1020 break; 1021 } 1022 imo->imo_multicast_loop = loop; 1023 break; 1024 1025 case IP_ADD_MEMBERSHIP: 1026 /* 1027 * Add a multicast group membership. 1028 * Group must be a valid IP multicast address. 1029 */ 1030 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1031 error = EINVAL; 1032 break; 1033 } 1034 mreq = mtod(m, struct ip_mreq *); 1035 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1036 error = EINVAL; 1037 break; 1038 } 1039 s = splimp(); 1040 /* 1041 * If no interface address was provided, use the interface of 1042 * the route to the given multicast address. 1043 */ 1044 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1045 bzero((caddr_t)&ro, sizeof(ro)); 1046 dst = (struct sockaddr_in *)&ro.ro_dst; 1047 dst->sin_len = sizeof(*dst); 1048 dst->sin_family = AF_INET; 1049 dst->sin_addr = mreq->imr_multiaddr; 1050 rtalloc(&ro); 1051 if (ro.ro_rt == NULL) { 1052 error = EADDRNOTAVAIL; 1053 splx(s); 1054 break; 1055 } 1056 ifp = ro.ro_rt->rt_ifp; 1057 rtfree(ro.ro_rt); 1058 } 1059 else { 1060 INADDR_TO_IFP(mreq->imr_interface, ifp); 1061 } 1062 1063 /* 1064 * See if we found an interface, and confirm that it 1065 * supports multicast. 1066 */ 1067 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1068 error = EADDRNOTAVAIL; 1069 splx(s); 1070 break; 1071 } 1072 /* 1073 * See if the membership already exists or if all the 1074 * membership slots are full. 1075 */ 1076 for (i = 0; i < imo->imo_num_memberships; ++i) { 1077 if (imo->imo_membership[i]->inm_ifp == ifp && 1078 imo->imo_membership[i]->inm_addr.s_addr 1079 == mreq->imr_multiaddr.s_addr) 1080 break; 1081 } 1082 if (i < imo->imo_num_memberships) { 1083 error = EADDRINUSE; 1084 splx(s); 1085 break; 1086 } 1087 if (i == IP_MAX_MEMBERSHIPS) { 1088 error = ETOOMANYREFS; 1089 splx(s); 1090 break; 1091 } 1092 /* 1093 * Everything looks good; add a new record to the multicast 1094 * address list for the given interface. 1095 */ 1096 if ((imo->imo_membership[i] = 1097 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1098 error = ENOBUFS; 1099 splx(s); 1100 break; 1101 } 1102 ++imo->imo_num_memberships; 1103 splx(s); 1104 break; 1105 1106 case IP_DROP_MEMBERSHIP: 1107 /* 1108 * Drop a multicast group membership. 1109 * Group must be a valid IP multicast address. 1110 */ 1111 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1112 error = EINVAL; 1113 break; 1114 } 1115 mreq = mtod(m, struct ip_mreq *); 1116 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1117 error = EINVAL; 1118 break; 1119 } 1120 1121 s = splimp(); 1122 /* 1123 * If an interface address was specified, get a pointer 1124 * to its ifnet structure. 1125 */ 1126 if (mreq->imr_interface.s_addr == INADDR_ANY) 1127 ifp = NULL; 1128 else { 1129 INADDR_TO_IFP(mreq->imr_interface, ifp); 1130 if (ifp == NULL) { 1131 error = EADDRNOTAVAIL; 1132 splx(s); 1133 break; 1134 } 1135 } 1136 /* 1137 * Find the membership in the membership array. 1138 */ 1139 for (i = 0; i < imo->imo_num_memberships; ++i) { 1140 if ((ifp == NULL || 1141 imo->imo_membership[i]->inm_ifp == ifp) && 1142 imo->imo_membership[i]->inm_addr.s_addr == 1143 mreq->imr_multiaddr.s_addr) 1144 break; 1145 } 1146 if (i == imo->imo_num_memberships) { 1147 error = EADDRNOTAVAIL; 1148 splx(s); 1149 break; 1150 } 1151 /* 1152 * Give up the multicast address record to which the 1153 * membership points. 1154 */ 1155 in_delmulti(imo->imo_membership[i]); 1156 /* 1157 * Remove the gap in the membership array. 1158 */ 1159 for (++i; i < imo->imo_num_memberships; ++i) 1160 imo->imo_membership[i-1] = imo->imo_membership[i]; 1161 --imo->imo_num_memberships; 1162 splx(s); 1163 break; 1164 1165 default: 1166 error = EOPNOTSUPP; 1167 break; 1168 } 1169 1170 /* 1171 * If all options have default values, no need to keep the mbuf. 1172 */ 1173 if (imo->imo_multicast_ifp == NULL && 1174 imo->imo_multicast_vif == -1 && 1175 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1176 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1177 imo->imo_num_memberships == 0) { 1178 free(*imop, M_IPMOPTS); 1179 *imop = NULL; 1180 } 1181 1182 return (error); 1183} 1184 1185/* 1186 * Return the IP multicast options in response to user getsockopt(). 1187 */ 1188static int 1189ip_getmoptions(optname, imo, mp) 1190 int optname; 1191 register struct ip_moptions *imo; 1192 register struct mbuf **mp; 1193{ 1194 u_char *ttl; 1195 u_char *loop; 1196 struct in_addr *addr; 1197 struct in_ifaddr *ia; 1198 1199 *mp = m_get(M_WAIT, MT_SOOPTS); 1200 1201 switch (optname) { 1202 1203 case IP_MULTICAST_VIF: 1204 if (imo != NULL) 1205 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1206 else 1207 *(mtod(*mp, int *)) = -1; 1208 (*mp)->m_len = sizeof(int); 1209 return(0); 1210 1211 case IP_MULTICAST_IF: 1212 addr = mtod(*mp, struct in_addr *); 1213 (*mp)->m_len = sizeof(struct in_addr); 1214 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1215 addr->s_addr = INADDR_ANY; 1216 else { 1217 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1218 addr->s_addr = (ia == NULL) ? INADDR_ANY 1219 : IA_SIN(ia)->sin_addr.s_addr; 1220 } 1221 return (0); 1222 1223 case IP_MULTICAST_TTL: 1224 ttl = mtod(*mp, u_char *); 1225 (*mp)->m_len = 1; 1226 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1227 : imo->imo_multicast_ttl; 1228 return (0); 1229 1230 case IP_MULTICAST_LOOP: 1231 loop = mtod(*mp, u_char *); 1232 (*mp)->m_len = 1; 1233 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1234 : imo->imo_multicast_loop; 1235 return (0); 1236 1237 default: 1238 return (EOPNOTSUPP); 1239 } 1240} 1241 1242/* 1243 * Discard the IP multicast options. 1244 */ 1245void 1246ip_freemoptions(imo) 1247 register struct ip_moptions *imo; 1248{ 1249 register int i; 1250 1251 if (imo != NULL) { 1252 for (i = 0; i < imo->imo_num_memberships; ++i) 1253 in_delmulti(imo->imo_membership[i]); 1254 free(imo, M_IPMOPTS); 1255 } 1256} 1257 1258/* 1259 * Routine called from ip_output() to loop back a copy of an IP multicast 1260 * packet to the input queue of a specified interface. Note that this 1261 * calls the output routine of the loopback "driver", but with an interface 1262 * pointer that might NOT be a loopback interface -- evil, but easier than 1263 * replicating that code here. 1264 */ 1265static void 1266ip_mloopback(ifp, m, dst) 1267 struct ifnet *ifp; 1268 register struct mbuf *m; 1269 register struct sockaddr_in *dst; 1270{ 1271 register struct ip *ip; 1272 struct mbuf *copym; 1273 1274 copym = m_copy(m, 0, M_COPYALL); 1275 if (copym != NULL) { 1276 /* 1277 * We don't bother to fragment if the IP length is greater 1278 * than the interface's MTU. Can this possibly matter? 1279 */ 1280 ip = mtod(copym, struct ip *); 1281 ip->ip_len = htons((u_short)ip->ip_len); 1282 ip->ip_off = htons((u_short)ip->ip_off); 1283 ip->ip_sum = 0; 1284 if (ip->ip_vhl == IP_VHL_BORING) { 1285 ip->ip_sum = in_cksum_hdr(ip); 1286 } else { 1287 ip->ip_sum = in_cksum(copym, 1288 IP_VHL_HL(ip->ip_vhl) << 2); 1289 } 1290 /* 1291 * NB: 1292 * We can't simply call ip_input() directly because 1293 * the ip_mforward() depends on the `input interface' 1294 * being set to something unreasonable so that we don't 1295 * attempt to forward the looped-back copy. 1296 * It's also not clear whether there are any lingering 1297 * reentrancy problems in other areas which might be 1298 * exposed by this code. For the moment, we'll err 1299 * on the side of safety by continuing to abuse 1300 * loinput(). 1301 */ 1302#ifdef notdef 1303 copym->m_pkthdr.rcvif = &loif[0]; 1304 ip_input(copym) 1305#else 1306 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1307#endif 1308 } 1309} 1310