ip_output.c revision 19622
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 1.44 1996/10/22 22:26:02 sos 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#ifdef COMPAT_IPFW 343 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, ifp, IP_NAT_OUT)) { 344 error = EACCES; 345 goto done; 346 } 347 348 /* 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 case IP_RECVIF: 632 if (m == 0 || m->m_len != sizeof(int)) 633 error = EINVAL; 634 else { 635 optval = *mtod(m, int *); 636 switch (optname) { 637 638 case IP_TOS: 639 inp->inp_ip.ip_tos = optval; 640 break; 641 642 case IP_TTL: 643 inp->inp_ip.ip_ttl = optval; 644 break; 645#define OPTSET(bit) \ 646 if (optval) \ 647 inp->inp_flags |= bit; \ 648 else \ 649 inp->inp_flags &= ~bit; 650 651 case IP_RECVOPTS: 652 OPTSET(INP_RECVOPTS); 653 break; 654 655 case IP_RECVRETOPTS: 656 OPTSET(INP_RECVRETOPTS); 657 break; 658 659 case IP_RECVDSTADDR: 660 OPTSET(INP_RECVDSTADDR); 661 break; 662 663 case IP_RECVIF: 664 OPTSET(INP_RECVIF); 665 break; 666 } 667 } 668 break; 669#undef OPTSET 670 671 case IP_MULTICAST_IF: 672 case IP_MULTICAST_VIF: 673 case IP_MULTICAST_TTL: 674 case IP_MULTICAST_LOOP: 675 case IP_ADD_MEMBERSHIP: 676 case IP_DROP_MEMBERSHIP: 677 error = ip_setmoptions(optname, &inp->inp_moptions, m); 678 break; 679 680 case IP_PORTRANGE: 681 if (m == 0 || m->m_len != sizeof(int)) 682 error = EINVAL; 683 else { 684 optval = *mtod(m, int *); 685 686 switch (optval) { 687 688 case IP_PORTRANGE_DEFAULT: 689 inp->inp_flags &= ~(INP_LOWPORT); 690 inp->inp_flags &= ~(INP_HIGHPORT); 691 break; 692 693 case IP_PORTRANGE_HIGH: 694 inp->inp_flags &= ~(INP_LOWPORT); 695 inp->inp_flags |= INP_HIGHPORT; 696 break; 697 698 case IP_PORTRANGE_LOW: 699 inp->inp_flags &= ~(INP_HIGHPORT); 700 inp->inp_flags |= INP_LOWPORT; 701 break; 702 703 default: 704 error = EINVAL; 705 break; 706 } 707 } 708 break; 709 710 default: 711 error = ENOPROTOOPT; 712 break; 713 } 714 if (m) 715 (void)m_free(m); 716 break; 717 718 case PRCO_GETOPT: 719 switch (optname) { 720 case IP_OPTIONS: 721 case IP_RETOPTS: 722 *mp = m = m_get(M_WAIT, MT_SOOPTS); 723 if (inp->inp_options) { 724 m->m_len = inp->inp_options->m_len; 725 bcopy(mtod(inp->inp_options, void *), 726 mtod(m, void *), m->m_len); 727 } else 728 m->m_len = 0; 729 break; 730 731 case IP_TOS: 732 case IP_TTL: 733 case IP_RECVOPTS: 734 case IP_RECVRETOPTS: 735 case IP_RECVDSTADDR: 736 case IP_RECVIF: 737 *mp = m = m_get(M_WAIT, MT_SOOPTS); 738 m->m_len = sizeof(int); 739 switch (optname) { 740 741 case IP_TOS: 742 optval = inp->inp_ip.ip_tos; 743 break; 744 745 case IP_TTL: 746 optval = inp->inp_ip.ip_ttl; 747 break; 748 749#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 750 751 case IP_RECVOPTS: 752 optval = OPTBIT(INP_RECVOPTS); 753 break; 754 755 case IP_RECVRETOPTS: 756 optval = OPTBIT(INP_RECVRETOPTS); 757 break; 758 759 case IP_RECVDSTADDR: 760 optval = OPTBIT(INP_RECVDSTADDR); 761 break; 762 763 case IP_RECVIF: 764 optval = OPTBIT(INP_RECVIF); 765 break; 766 } 767 *mtod(m, int *) = optval; 768 break; 769 770 case IP_MULTICAST_IF: 771 case IP_MULTICAST_VIF: 772 case IP_MULTICAST_TTL: 773 case IP_MULTICAST_LOOP: 774 case IP_ADD_MEMBERSHIP: 775 case IP_DROP_MEMBERSHIP: 776 error = ip_getmoptions(optname, inp->inp_moptions, mp); 777 break; 778 779 case IP_PORTRANGE: 780 *mp = m = m_get(M_WAIT, MT_SOOPTS); 781 m->m_len = sizeof(int); 782 783 if (inp->inp_flags & INP_HIGHPORT) 784 optval = IP_PORTRANGE_HIGH; 785 else if (inp->inp_flags & INP_LOWPORT) 786 optval = IP_PORTRANGE_LOW; 787 else 788 optval = 0; 789 790 *mtod(m, int *) = optval; 791 break; 792 793 default: 794 error = ENOPROTOOPT; 795 break; 796 } 797 break; 798 } 799 return (error); 800} 801 802/* 803 * Set up IP options in pcb for insertion in output packets. 804 * Store in mbuf with pointer in pcbopt, adding pseudo-option 805 * with destination address if source routed. 806 */ 807static int 808#ifdef notyet 809ip_pcbopts(optname, pcbopt, m) 810 int optname; 811#else 812ip_pcbopts(pcbopt, m) 813#endif 814 struct mbuf **pcbopt; 815 register struct mbuf *m; 816{ 817 register cnt, optlen; 818 register u_char *cp; 819 u_char opt; 820 821 /* turn off any old options */ 822 if (*pcbopt) 823 (void)m_free(*pcbopt); 824 *pcbopt = 0; 825 if (m == (struct mbuf *)0 || m->m_len == 0) { 826 /* 827 * Only turning off any previous options. 828 */ 829 if (m) 830 (void)m_free(m); 831 return (0); 832 } 833 834#ifndef vax 835 if (m->m_len % sizeof(long)) 836 goto bad; 837#endif 838 /* 839 * IP first-hop destination address will be stored before 840 * actual options; move other options back 841 * and clear it when none present. 842 */ 843 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 844 goto bad; 845 cnt = m->m_len; 846 m->m_len += sizeof(struct in_addr); 847 cp = mtod(m, u_char *) + sizeof(struct in_addr); 848 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 849 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 850 851 for (; cnt > 0; cnt -= optlen, cp += optlen) { 852 opt = cp[IPOPT_OPTVAL]; 853 if (opt == IPOPT_EOL) 854 break; 855 if (opt == IPOPT_NOP) 856 optlen = 1; 857 else { 858 optlen = cp[IPOPT_OLEN]; 859 if (optlen <= IPOPT_OLEN || optlen > cnt) 860 goto bad; 861 } 862 switch (opt) { 863 864 default: 865 break; 866 867 case IPOPT_LSRR: 868 case IPOPT_SSRR: 869 /* 870 * user process specifies route as: 871 * ->A->B->C->D 872 * D must be our final destination (but we can't 873 * check that since we may not have connected yet). 874 * A is first hop destination, which doesn't appear in 875 * actual IP option, but is stored before the options. 876 */ 877 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 878 goto bad; 879 m->m_len -= sizeof(struct in_addr); 880 cnt -= sizeof(struct in_addr); 881 optlen -= sizeof(struct in_addr); 882 cp[IPOPT_OLEN] = optlen; 883 /* 884 * Move first hop before start of options. 885 */ 886 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 887 sizeof(struct in_addr)); 888 /* 889 * Then copy rest of options back 890 * to close up the deleted entry. 891 */ 892 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 893 sizeof(struct in_addr)), 894 (caddr_t)&cp[IPOPT_OFFSET+1], 895 (unsigned)cnt + sizeof(struct in_addr)); 896 break; 897 } 898 } 899 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 900 goto bad; 901 *pcbopt = m; 902 return (0); 903 904bad: 905 (void)m_free(m); 906 return (EINVAL); 907} 908 909/* 910 * Set the IP multicast options in response to user setsockopt(). 911 */ 912static int 913ip_setmoptions(optname, imop, m) 914 int optname; 915 struct ip_moptions **imop; 916 struct mbuf *m; 917{ 918 register int error = 0; 919 u_char loop; 920 register int i; 921 struct in_addr addr; 922 register struct ip_mreq *mreq; 923 register struct ifnet *ifp; 924 register struct ip_moptions *imo = *imop; 925 struct route ro; 926 register struct sockaddr_in *dst; 927 int s; 928 929 if (imo == NULL) { 930 /* 931 * No multicast option buffer attached to the pcb; 932 * allocate one and initialize to default values. 933 */ 934 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 935 M_WAITOK); 936 937 if (imo == NULL) 938 return (ENOBUFS); 939 *imop = imo; 940 imo->imo_multicast_ifp = NULL; 941 imo->imo_multicast_vif = -1; 942 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 943 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 944 imo->imo_num_memberships = 0; 945 } 946 947 switch (optname) { 948 /* store an index number for the vif you wanna use in the send */ 949 case IP_MULTICAST_VIF: 950 if (!legal_vif_num) { 951 error = EOPNOTSUPP; 952 break; 953 } 954 if (m == NULL || m->m_len != sizeof(int)) { 955 error = EINVAL; 956 break; 957 } 958 i = *(mtod(m, int *)); 959 if (!legal_vif_num(i) && (i != -1)) { 960 error = EINVAL; 961 break; 962 } 963 imo->imo_multicast_vif = i; 964 break; 965 966 case IP_MULTICAST_IF: 967 /* 968 * Select the interface for outgoing multicast packets. 969 */ 970 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 971 error = EINVAL; 972 break; 973 } 974 addr = *(mtod(m, struct in_addr *)); 975 /* 976 * INADDR_ANY is used to remove a previous selection. 977 * When no interface is selected, a default one is 978 * chosen every time a multicast packet is sent. 979 */ 980 if (addr.s_addr == INADDR_ANY) { 981 imo->imo_multicast_ifp = NULL; 982 break; 983 } 984 /* 985 * The selected interface is identified by its local 986 * IP address. Find the interface and confirm that 987 * it supports multicasting. 988 */ 989 s = splimp(); 990 INADDR_TO_IFP(addr, ifp); 991 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 992 splx(s); 993 error = EADDRNOTAVAIL; 994 break; 995 } 996 imo->imo_multicast_ifp = ifp; 997 splx(s); 998 break; 999 1000 case IP_MULTICAST_TTL: 1001 /* 1002 * Set the IP time-to-live for outgoing multicast packets. 1003 */ 1004 if (m == NULL || m->m_len != 1) { 1005 error = EINVAL; 1006 break; 1007 } 1008 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1009 break; 1010 1011 case IP_MULTICAST_LOOP: 1012 /* 1013 * Set the loopback flag for outgoing multicast packets. 1014 * Must be zero or one. 1015 */ 1016 if (m == NULL || m->m_len != 1 || 1017 (loop = *(mtod(m, u_char *))) > 1) { 1018 error = EINVAL; 1019 break; 1020 } 1021 imo->imo_multicast_loop = loop; 1022 break; 1023 1024 case IP_ADD_MEMBERSHIP: 1025 /* 1026 * Add a multicast group membership. 1027 * Group must be a valid IP multicast address. 1028 */ 1029 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1030 error = EINVAL; 1031 break; 1032 } 1033 mreq = mtod(m, struct ip_mreq *); 1034 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1035 error = EINVAL; 1036 break; 1037 } 1038 s = splimp(); 1039 /* 1040 * If no interface address was provided, use the interface of 1041 * the route to the given multicast address. 1042 */ 1043 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1044 bzero((caddr_t)&ro, sizeof(ro)); 1045 dst = (struct sockaddr_in *)&ro.ro_dst; 1046 dst->sin_len = sizeof(*dst); 1047 dst->sin_family = AF_INET; 1048 dst->sin_addr = mreq->imr_multiaddr; 1049 rtalloc(&ro); 1050 if (ro.ro_rt == NULL) { 1051 error = EADDRNOTAVAIL; 1052 splx(s); 1053 break; 1054 } 1055 ifp = ro.ro_rt->rt_ifp; 1056 rtfree(ro.ro_rt); 1057 } 1058 else { 1059 INADDR_TO_IFP(mreq->imr_interface, ifp); 1060 } 1061 1062 /* 1063 * See if we found an interface, and confirm that it 1064 * supports multicast. 1065 */ 1066 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1067 error = EADDRNOTAVAIL; 1068 splx(s); 1069 break; 1070 } 1071 /* 1072 * See if the membership already exists or if all the 1073 * membership slots are full. 1074 */ 1075 for (i = 0; i < imo->imo_num_memberships; ++i) { 1076 if (imo->imo_membership[i]->inm_ifp == ifp && 1077 imo->imo_membership[i]->inm_addr.s_addr 1078 == mreq->imr_multiaddr.s_addr) 1079 break; 1080 } 1081 if (i < imo->imo_num_memberships) { 1082 error = EADDRINUSE; 1083 splx(s); 1084 break; 1085 } 1086 if (i == IP_MAX_MEMBERSHIPS) { 1087 error = ETOOMANYREFS; 1088 splx(s); 1089 break; 1090 } 1091 /* 1092 * Everything looks good; add a new record to the multicast 1093 * address list for the given interface. 1094 */ 1095 if ((imo->imo_membership[i] = 1096 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1097 error = ENOBUFS; 1098 splx(s); 1099 break; 1100 } 1101 ++imo->imo_num_memberships; 1102 splx(s); 1103 break; 1104 1105 case IP_DROP_MEMBERSHIP: 1106 /* 1107 * Drop a multicast group membership. 1108 * Group must be a valid IP multicast address. 1109 */ 1110 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1111 error = EINVAL; 1112 break; 1113 } 1114 mreq = mtod(m, struct ip_mreq *); 1115 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1116 error = EINVAL; 1117 break; 1118 } 1119 1120 s = splimp(); 1121 /* 1122 * If an interface address was specified, get a pointer 1123 * to its ifnet structure. 1124 */ 1125 if (mreq->imr_interface.s_addr == INADDR_ANY) 1126 ifp = NULL; 1127 else { 1128 INADDR_TO_IFP(mreq->imr_interface, ifp); 1129 if (ifp == NULL) { 1130 error = EADDRNOTAVAIL; 1131 splx(s); 1132 break; 1133 } 1134 } 1135 /* 1136 * Find the membership in the membership array. 1137 */ 1138 for (i = 0; i < imo->imo_num_memberships; ++i) { 1139 if ((ifp == NULL || 1140 imo->imo_membership[i]->inm_ifp == ifp) && 1141 imo->imo_membership[i]->inm_addr.s_addr == 1142 mreq->imr_multiaddr.s_addr) 1143 break; 1144 } 1145 if (i == imo->imo_num_memberships) { 1146 error = EADDRNOTAVAIL; 1147 splx(s); 1148 break; 1149 } 1150 /* 1151 * Give up the multicast address record to which the 1152 * membership points. 1153 */ 1154 in_delmulti(imo->imo_membership[i]); 1155 /* 1156 * Remove the gap in the membership array. 1157 */ 1158 for (++i; i < imo->imo_num_memberships; ++i) 1159 imo->imo_membership[i-1] = imo->imo_membership[i]; 1160 --imo->imo_num_memberships; 1161 splx(s); 1162 break; 1163 1164 default: 1165 error = EOPNOTSUPP; 1166 break; 1167 } 1168 1169 /* 1170 * If all options have default values, no need to keep the mbuf. 1171 */ 1172 if (imo->imo_multicast_ifp == NULL && 1173 imo->imo_multicast_vif == -1 && 1174 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1175 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1176 imo->imo_num_memberships == 0) { 1177 free(*imop, M_IPMOPTS); 1178 *imop = NULL; 1179 } 1180 1181 return (error); 1182} 1183 1184/* 1185 * Return the IP multicast options in response to user getsockopt(). 1186 */ 1187static int 1188ip_getmoptions(optname, imo, mp) 1189 int optname; 1190 register struct ip_moptions *imo; 1191 register struct mbuf **mp; 1192{ 1193 u_char *ttl; 1194 u_char *loop; 1195 struct in_addr *addr; 1196 struct in_ifaddr *ia; 1197 1198 *mp = m_get(M_WAIT, MT_SOOPTS); 1199 1200 switch (optname) { 1201 1202 case IP_MULTICAST_VIF: 1203 if (imo != NULL) 1204 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1205 else 1206 *(mtod(*mp, int *)) = -1; 1207 (*mp)->m_len = sizeof(int); 1208 return(0); 1209 1210 case IP_MULTICAST_IF: 1211 addr = mtod(*mp, struct in_addr *); 1212 (*mp)->m_len = sizeof(struct in_addr); 1213 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1214 addr->s_addr = INADDR_ANY; 1215 else { 1216 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1217 addr->s_addr = (ia == NULL) ? INADDR_ANY 1218 : IA_SIN(ia)->sin_addr.s_addr; 1219 } 1220 return (0); 1221 1222 case IP_MULTICAST_TTL: 1223 ttl = mtod(*mp, u_char *); 1224 (*mp)->m_len = 1; 1225 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1226 : imo->imo_multicast_ttl; 1227 return (0); 1228 1229 case IP_MULTICAST_LOOP: 1230 loop = mtod(*mp, u_char *); 1231 (*mp)->m_len = 1; 1232 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1233 : imo->imo_multicast_loop; 1234 return (0); 1235 1236 default: 1237 return (EOPNOTSUPP); 1238 } 1239} 1240 1241/* 1242 * Discard the IP multicast options. 1243 */ 1244void 1245ip_freemoptions(imo) 1246 register struct ip_moptions *imo; 1247{ 1248 register int i; 1249 1250 if (imo != NULL) { 1251 for (i = 0; i < imo->imo_num_memberships; ++i) 1252 in_delmulti(imo->imo_membership[i]); 1253 free(imo, M_IPMOPTS); 1254 } 1255} 1256 1257/* 1258 * Routine called from ip_output() to loop back a copy of an IP multicast 1259 * packet to the input queue of a specified interface. Note that this 1260 * calls the output routine of the loopback "driver", but with an interface 1261 * pointer that might NOT be a loopback interface -- evil, but easier than 1262 * replicating that code here. 1263 */ 1264static void 1265ip_mloopback(ifp, m, dst) 1266 struct ifnet *ifp; 1267 register struct mbuf *m; 1268 register struct sockaddr_in *dst; 1269{ 1270 register struct ip *ip; 1271 struct mbuf *copym; 1272 1273 copym = m_copy(m, 0, M_COPYALL); 1274 if (copym != NULL) { 1275 /* 1276 * We don't bother to fragment if the IP length is greater 1277 * than the interface's MTU. Can this possibly matter? 1278 */ 1279 ip = mtod(copym, struct ip *); 1280 ip->ip_len = htons((u_short)ip->ip_len); 1281 ip->ip_off = htons((u_short)ip->ip_off); 1282 ip->ip_sum = 0; 1283 if (ip->ip_vhl == IP_VHL_BORING) { 1284 ip->ip_sum = in_cksum_hdr(ip); 1285 } else { 1286 ip->ip_sum = in_cksum(copym, 1287 IP_VHL_HL(ip->ip_vhl) << 2); 1288 } 1289 /* 1290 * NB: 1291 * We can't simply call ip_input() directly because 1292 * the ip_mforward() depends on the `input interface' 1293 * being set to something unreasonable so that we don't 1294 * attempt to forward the looped-back copy. 1295 * It's also not clear whether there are any lingering 1296 * reentrancy problems in other areas which might be 1297 * exposed by this code. For the moment, we'll err 1298 * on the side of safety by continuing to abuse 1299 * loinput(). 1300 */ 1301#ifdef notdef 1302 copym->m_pkthdr.rcvif = &loif[0]; 1303 ip_input(copym) 1304#else 1305 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1306#endif 1307 } 1308} 1309