ip_output.c revision 30309
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 1.58 1997/08/02 14:32:53 bde Exp $ 35 */ 36 37#define _IP_VHL 38 39#include <sys/param.h> 40#include <sys/systm.h> 41#include <sys/malloc.h> 42#include <sys/mbuf.h> 43#include <sys/protosw.h> 44#include <sys/socket.h> 45#include <sys/socketvar.h> 46 47#include <net/if.h> 48#include <net/route.h> 49 50#include <netinet/in.h> 51#include <netinet/in_systm.h> 52#include <netinet/ip.h> 53#include <netinet/in_pcb.h> 54#include <netinet/in_var.h> 55#include <netinet/ip_var.h> 56 57#ifdef vax 58#include <machine/mtpr.h> 59#endif 60#include <machine/in_cksum.h> 61 62MALLOC_DEFINE(M_IPMOPTS, "ip_moptions", "internet multicast options"); 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 *, int)); 76static int ip_getmoptions 77 __P((int, struct ip_moptions *, struct mbuf **)); 78static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 79static int ip_setmoptions 80 __P((int, struct ip_moptions **, struct mbuf *)); 81 82#if defined(IPFILTER_LKM) || defined(IPFILTER) 83int ip_optcopy __P((struct ip *, struct ip *)); 84extern int fr_check __P((struct ip *, int, struct ifnet *, int, struct mbuf **)); 85extern int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **)); 86#else 87static int ip_optcopy __P((struct ip *, struct ip *)); 88#endif 89 90 91extern struct protosw inetsw[]; 92 93/* 94 * IP output. The packet in mbuf chain m contains a skeletal IP 95 * header (with len, off, ttl, proto, tos, src, dst). 96 * The mbuf chain containing the packet will be freed. 97 * The mbuf opt, if present, will not be freed. 98 */ 99int 100ip_output(m0, opt, ro, flags, imo) 101 struct mbuf *m0; 102 struct mbuf *opt; 103 struct route *ro; 104 int flags; 105 struct ip_moptions *imo; 106{ 107 struct ip *ip, *mhip; 108 struct ifnet *ifp; 109 struct mbuf *m = m0; 110 int hlen = sizeof (struct ip); 111 int len, off, error = 0; 112 struct sockaddr_in *dst; 113 struct in_ifaddr *ia; 114 int isbroadcast; 115 116#ifdef DIAGNOSTIC 117 if ((m->m_flags & M_PKTHDR) == 0) 118 panic("ip_output no HDR"); 119 if (!ro) 120 panic("ip_output no route, proto = %d", 121 mtod(m, struct ip *)->ip_p); 122#endif 123 if (opt) { 124 m = ip_insertoptions(m, opt, &len); 125 hlen = len; 126 } 127 ip = mtod(m, struct ip *); 128 /* 129 * Fill in IP header. 130 */ 131 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 132 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 133 ip->ip_off &= IP_DF; 134 ip->ip_id = htons(ip_id++); 135 ipstat.ips_localout++; 136 } else { 137 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 138 } 139 140 dst = (struct sockaddr_in *)&ro->ro_dst; 141 /* 142 * If there is a cached route, 143 * check that it is to the same destination 144 * and is still up. If not, free it and try again. 145 */ 146 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 147 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 148 RTFREE(ro->ro_rt); 149 ro->ro_rt = (struct rtentry *)0; 150 } 151 if (ro->ro_rt == 0) { 152 dst->sin_family = AF_INET; 153 dst->sin_len = sizeof(*dst); 154 dst->sin_addr = ip->ip_dst; 155 } 156 /* 157 * If routing to interface only, 158 * short circuit routing lookup. 159 */ 160#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 161#define sintosa(sin) ((struct sockaddr *)(sin)) 162 if (flags & IP_ROUTETOIF) { 163 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 164 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 165 ipstat.ips_noroute++; 166 error = ENETUNREACH; 167 goto bad; 168 } 169 ifp = ia->ia_ifp; 170 ip->ip_ttl = 1; 171 isbroadcast = in_broadcast(dst->sin_addr, ifp); 172 } else { 173 /* 174 * If this is the case, we probably don't want to allocate 175 * a protocol-cloned route since we didn't get one from the 176 * ULP. This lets TCP do its thing, while not burdening 177 * forwarding or ICMP with the overhead of cloning a route. 178 * Of course, we still want to do any cloning requested by 179 * the link layer, as this is probably required in all cases 180 * for correct operation (as it is for ARP). 181 */ 182 if (ro->ro_rt == 0) 183 rtalloc_ign(ro, RTF_PRCLONING); 184 if (ro->ro_rt == 0) { 185 ipstat.ips_noroute++; 186 error = EHOSTUNREACH; 187 goto bad; 188 } 189 ia = ifatoia(ro->ro_rt->rt_ifa); 190 ifp = ro->ro_rt->rt_ifp; 191 ro->ro_rt->rt_use++; 192 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 193 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 194 if (ro->ro_rt->rt_flags & RTF_HOST) 195 isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST); 196 else 197 isbroadcast = in_broadcast(dst->sin_addr, ifp); 198 } 199 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 200 struct in_multi *inm; 201 202 m->m_flags |= M_MCAST; 203 /* 204 * IP destination address is multicast. Make sure "dst" 205 * still points to the address in "ro". (It may have been 206 * changed to point to a gateway address, above.) 207 */ 208 dst = (struct sockaddr_in *)&ro->ro_dst; 209 /* 210 * See if the caller provided any multicast options 211 */ 212 if (imo != NULL) { 213 ip->ip_ttl = imo->imo_multicast_ttl; 214 if (imo->imo_multicast_ifp != NULL) 215 ifp = imo->imo_multicast_ifp; 216 if (imo->imo_multicast_vif != -1) 217 ip->ip_src.s_addr = 218 ip_mcast_src(imo->imo_multicast_vif); 219 } else 220 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 221 /* 222 * Confirm that the outgoing interface supports multicast. 223 */ 224 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 225 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 226 ipstat.ips_noroute++; 227 error = ENETUNREACH; 228 goto bad; 229 } 230 } 231 /* 232 * If source address not specified yet, use address 233 * of outgoing interface. 234 */ 235 if (ip->ip_src.s_addr == INADDR_ANY) { 236 register struct in_ifaddr *ia; 237 238 for (ia = in_ifaddrhead.tqh_first; ia; 239 ia = ia->ia_link.tqe_next) 240 if (ia->ia_ifp == ifp) { 241 ip->ip_src = IA_SIN(ia)->sin_addr; 242 break; 243 } 244 } 245 246 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 247 if (inm != NULL && 248 (imo == NULL || imo->imo_multicast_loop)) { 249 /* 250 * If we belong to the destination multicast group 251 * on the outgoing interface, and the caller did not 252 * forbid loopback, loop back a copy. 253 */ 254 ip_mloopback(ifp, m, dst, hlen); 255 } 256 else { 257 /* 258 * If we are acting as a multicast router, perform 259 * multicast forwarding as if the packet had just 260 * arrived on the interface to which we are about 261 * to send. The multicast forwarding function 262 * recursively calls this function, using the 263 * IP_FORWARDING flag to prevent infinite recursion. 264 * 265 * Multicasts that are looped back by ip_mloopback(), 266 * above, will be forwarded by the ip_input() routine, 267 * if necessary. 268 */ 269 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 270 /* 271 * Check if rsvp daemon is running. If not, don't 272 * set ip_moptions. This ensures that the packet 273 * is multicast and not just sent down one link 274 * as prescribed by rsvpd. 275 */ 276 if (!rsvp_on) 277 imo = NULL; 278 if (ip_mforward(ip, ifp, m, imo) != 0) { 279 m_freem(m); 280 goto done; 281 } 282 } 283 } 284 285 /* 286 * Multicasts with a time-to-live of zero may be looped- 287 * back, above, but must not be transmitted on a network. 288 * Also, multicasts addressed to the loopback interface 289 * are not sent -- the above call to ip_mloopback() will 290 * loop back a copy if this host actually belongs to the 291 * destination group on the loopback interface. 292 */ 293 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 294 m_freem(m); 295 goto done; 296 } 297 298 goto sendit; 299 } 300#ifndef notdef 301 /* 302 * If source address not specified yet, use address 303 * of outgoing interface. 304 */ 305 if (ip->ip_src.s_addr == INADDR_ANY) 306 ip->ip_src = IA_SIN(ia)->sin_addr; 307#endif 308 /* 309 * Verify that we have any chance at all of being able to queue 310 * the packet or packet fragments 311 */ 312 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 313 ifp->if_snd.ifq_maxlen) { 314 error = ENOBUFS; 315 goto bad; 316 } 317 318 /* 319 * Look for broadcast address and 320 * and verify user is allowed to send 321 * such a packet. 322 */ 323 if (isbroadcast) { 324 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 325 error = EADDRNOTAVAIL; 326 goto bad; 327 } 328 if ((flags & IP_ALLOWBROADCAST) == 0) { 329 error = EACCES; 330 goto bad; 331 } 332 /* don't allow broadcast messages to be fragmented */ 333 if ((u_short)ip->ip_len > ifp->if_mtu) { 334 error = EMSGSIZE; 335 goto bad; 336 } 337 m->m_flags |= M_BCAST; 338 } else { 339 m->m_flags &= ~M_BCAST; 340 } 341 342sendit: 343#if defined(IPFILTER) || defined(IPFILTER_LKM) 344 /* 345 * looks like most checking has been done now...do a filter check 346 */ 347 if (fr_checkp) { 348 struct mbuf *m1 = m; 349 350 if ((*fr_checkp)(ip, hlen, ifp, 1, &m1)) 351 error = EHOSTUNREACH; 352 if (error || !m1) 353 goto done; 354 ip = mtod(m = m1, struct ip *); 355 } 356#endif 357 /* 358 * IpHack's section. 359 * - Xlate: translate packet's addr/port (NAT). 360 * - Firewall: deny/allow/etc. 361 * - Wrap: fake packet's addr/port <unimpl.> 362 * - Encapsulate: put it in another IP and send out. <unimp.> 363 */ 364 365#ifdef COMPAT_IPFW 366 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, ifp, IP_NAT_OUT)) { 367 error = EACCES; 368 goto done; 369 } 370 371 /* 372 * Check with the firewall... 373 */ 374 if (ip_fw_chk_ptr) { 375#ifdef IPDIVERT 376 ip_divert_port = (*ip_fw_chk_ptr)(&ip, 377 hlen, ifp, ip_divert_ignore, &m); 378 ip_divert_ignore = 0; 379 if (ip_divert_port) { /* Divert packet */ 380 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 381 goto done; 382 } 383#else 384 /* If ipfw says divert, we have to just drop packet */ 385 if ((*ip_fw_chk_ptr)(&ip, hlen, ifp, 0, &m)) { 386 m_freem(m); 387 goto done; 388 } 389#endif 390 if (!m) { 391 error = EACCES; 392 goto done; 393 } 394 } 395#endif /* COMPAT_IPFW */ 396 397 /* 398 * If small enough for interface, can just send directly. 399 */ 400 if ((u_short)ip->ip_len <= ifp->if_mtu) { 401 ip->ip_len = htons((u_short)ip->ip_len); 402 ip->ip_off = htons((u_short)ip->ip_off); 403 ip->ip_sum = 0; 404 if (ip->ip_vhl == IP_VHL_BORING) { 405 ip->ip_sum = in_cksum_hdr(ip); 406 } else { 407 ip->ip_sum = in_cksum(m, hlen); 408 } 409 error = (*ifp->if_output)(ifp, m, 410 (struct sockaddr *)dst, ro->ro_rt); 411 goto done; 412 } 413 /* 414 * Too large for interface; fragment if possible. 415 * Must be able to put at least 8 bytes per fragment. 416 */ 417 if (ip->ip_off & IP_DF) { 418 error = EMSGSIZE; 419 /* 420 * This case can happen if the user changed the MTU 421 * of an interface after enabling IP on it. Because 422 * most netifs don't keep track of routes pointing to 423 * them, there is no way for one to update all its 424 * routes when the MTU is changed. 425 */ 426 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 427 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 428 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 429 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 430 } 431 ipstat.ips_cantfrag++; 432 goto bad; 433 } 434 len = (ifp->if_mtu - hlen) &~ 7; 435 if (len < 8) { 436 error = EMSGSIZE; 437 goto bad; 438 } 439 440 { 441 int mhlen, firstlen = len; 442 struct mbuf **mnext = &m->m_nextpkt; 443 444 /* 445 * Loop through length of segment after first fragment, 446 * make new header and copy data of each part and link onto chain. 447 */ 448 m0 = m; 449 mhlen = sizeof (struct ip); 450 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 451 MGETHDR(m, M_DONTWAIT, MT_HEADER); 452 if (m == 0) { 453 error = ENOBUFS; 454 ipstat.ips_odropped++; 455 goto sendorfree; 456 } 457 m->m_data += max_linkhdr; 458 mhip = mtod(m, struct ip *); 459 *mhip = *ip; 460 if (hlen > sizeof (struct ip)) { 461 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 462 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 463 } 464 m->m_len = mhlen; 465 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 466 if (ip->ip_off & IP_MF) 467 mhip->ip_off |= IP_MF; 468 if (off + len >= (u_short)ip->ip_len) 469 len = (u_short)ip->ip_len - off; 470 else 471 mhip->ip_off |= IP_MF; 472 mhip->ip_len = htons((u_short)(len + mhlen)); 473 m->m_next = m_copy(m0, off, len); 474 if (m->m_next == 0) { 475 (void) m_free(m); 476 error = ENOBUFS; /* ??? */ 477 ipstat.ips_odropped++; 478 goto sendorfree; 479 } 480 m->m_pkthdr.len = mhlen + len; 481 m->m_pkthdr.rcvif = (struct ifnet *)0; 482 mhip->ip_off = htons((u_short)mhip->ip_off); 483 mhip->ip_sum = 0; 484 if (mhip->ip_vhl == IP_VHL_BORING) { 485 mhip->ip_sum = in_cksum_hdr(mhip); 486 } else { 487 mhip->ip_sum = in_cksum(m, mhlen); 488 } 489 *mnext = m; 490 mnext = &m->m_nextpkt; 491 ipstat.ips_ofragments++; 492 } 493 /* 494 * Update first fragment by trimming what's been copied out 495 * and updating header, then send each fragment (in order). 496 */ 497 m = m0; 498 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 499 m->m_pkthdr.len = hlen + firstlen; 500 ip->ip_len = htons((u_short)m->m_pkthdr.len); 501 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 502 ip->ip_sum = 0; 503 if (ip->ip_vhl == IP_VHL_BORING) { 504 ip->ip_sum = in_cksum_hdr(ip); 505 } else { 506 ip->ip_sum = in_cksum(m, hlen); 507 } 508sendorfree: 509 for (m = m0; m; m = m0) { 510 m0 = m->m_nextpkt; 511 m->m_nextpkt = 0; 512 if (error == 0) 513 error = (*ifp->if_output)(ifp, m, 514 (struct sockaddr *)dst, ro->ro_rt); 515 else 516 m_freem(m); 517 } 518 519 if (error == 0) 520 ipstat.ips_fragmented++; 521 } 522done: 523 return (error); 524bad: 525 m_freem(m0); 526 goto done; 527} 528 529/* 530 * Insert IP options into preformed packet. 531 * Adjust IP destination as required for IP source routing, 532 * as indicated by a non-zero in_addr at the start of the options. 533 * 534 * XXX This routine assumes that the packet has no options in place. 535 */ 536static struct mbuf * 537ip_insertoptions(m, opt, phlen) 538 register struct mbuf *m; 539 struct mbuf *opt; 540 int *phlen; 541{ 542 register struct ipoption *p = mtod(opt, struct ipoption *); 543 struct mbuf *n; 544 register struct ip *ip = mtod(m, struct ip *); 545 unsigned optlen; 546 547 optlen = opt->m_len - sizeof(p->ipopt_dst); 548 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 549 return (m); /* XXX should fail */ 550 if (p->ipopt_dst.s_addr) 551 ip->ip_dst = p->ipopt_dst; 552 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 553 MGETHDR(n, M_DONTWAIT, MT_HEADER); 554 if (n == 0) 555 return (m); 556 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 557 m->m_len -= sizeof(struct ip); 558 m->m_data += sizeof(struct ip); 559 n->m_next = m; 560 m = n; 561 m->m_len = optlen + sizeof(struct ip); 562 m->m_data += max_linkhdr; 563 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 564 } else { 565 m->m_data -= optlen; 566 m->m_len += optlen; 567 m->m_pkthdr.len += optlen; 568 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 569 } 570 ip = mtod(m, struct ip *); 571 bcopy(p->ipopt_list, ip + 1, optlen); 572 *phlen = sizeof(struct ip) + optlen; 573 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 574 ip->ip_len += optlen; 575 return (m); 576} 577 578/* 579 * Copy options from ip to jp, 580 * omitting those not copied during fragmentation. 581 */ 582#if !defined(IPFILTER) && !defined(IPFILTER_LKM) 583static 584#endif 585int 586ip_optcopy(ip, jp) 587 struct ip *ip, *jp; 588{ 589 register u_char *cp, *dp; 590 int opt, optlen, cnt; 591 592 cp = (u_char *)(ip + 1); 593 dp = (u_char *)(jp + 1); 594 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 595 for (; cnt > 0; cnt -= optlen, cp += optlen) { 596 opt = cp[0]; 597 if (opt == IPOPT_EOL) 598 break; 599 if (opt == IPOPT_NOP) { 600 /* Preserve for IP mcast tunnel's LSRR alignment. */ 601 *dp++ = IPOPT_NOP; 602 optlen = 1; 603 continue; 604 } else 605 optlen = cp[IPOPT_OLEN]; 606 /* bogus lengths should have been caught by ip_dooptions */ 607 if (optlen > cnt) 608 optlen = cnt; 609 if (IPOPT_COPIED(opt)) { 610 bcopy(cp, dp, optlen); 611 dp += optlen; 612 } 613 } 614 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 615 *dp++ = IPOPT_EOL; 616 return (optlen); 617} 618 619/* 620 * IP socket option processing. 621 */ 622int 623ip_ctloutput(op, so, level, optname, mp, p) 624 int op; 625 struct socket *so; 626 int level, optname; 627 struct mbuf **mp; 628 struct proc *p; 629{ 630 register struct inpcb *inp = sotoinpcb(so); 631 register struct mbuf *m = *mp; 632 register int optval = 0; 633 int error = 0; 634 635 if (level != IPPROTO_IP) { 636 error = EINVAL; 637 if (op == PRCO_SETOPT && *mp) 638 (void) m_free(*mp); 639 } else switch (op) { 640 641 case PRCO_SETOPT: 642 switch (optname) { 643 case IP_OPTIONS: 644#ifdef notyet 645 case IP_RETOPTS: 646 return (ip_pcbopts(optname, &inp->inp_options, m)); 647#else 648 return (ip_pcbopts(&inp->inp_options, m)); 649#endif 650 651 case IP_TOS: 652 case IP_TTL: 653 case IP_RECVOPTS: 654 case IP_RECVRETOPTS: 655 case IP_RECVDSTADDR: 656 case IP_RECVIF: 657 if (m == 0 || m->m_len != sizeof(int)) 658 error = EINVAL; 659 else { 660 optval = *mtod(m, int *); 661 switch (optname) { 662 663 case IP_TOS: 664 inp->inp_ip_tos = optval; 665 break; 666 667 case IP_TTL: 668 inp->inp_ip_ttl = optval; 669 break; 670#define OPTSET(bit) \ 671 if (optval) \ 672 inp->inp_flags |= bit; \ 673 else \ 674 inp->inp_flags &= ~bit; 675 676 case IP_RECVOPTS: 677 OPTSET(INP_RECVOPTS); 678 break; 679 680 case IP_RECVRETOPTS: 681 OPTSET(INP_RECVRETOPTS); 682 break; 683 684 case IP_RECVDSTADDR: 685 OPTSET(INP_RECVDSTADDR); 686 break; 687 688 case IP_RECVIF: 689 OPTSET(INP_RECVIF); 690 break; 691 } 692 } 693 break; 694#undef OPTSET 695 696 case IP_MULTICAST_IF: 697 case IP_MULTICAST_VIF: 698 case IP_MULTICAST_TTL: 699 case IP_MULTICAST_LOOP: 700 case IP_ADD_MEMBERSHIP: 701 case IP_DROP_MEMBERSHIP: 702 error = ip_setmoptions(optname, &inp->inp_moptions, m); 703 break; 704 705 case IP_PORTRANGE: 706 if (m == 0 || m->m_len != sizeof(int)) 707 error = EINVAL; 708 else { 709 optval = *mtod(m, int *); 710 711 switch (optval) { 712 713 case IP_PORTRANGE_DEFAULT: 714 inp->inp_flags &= ~(INP_LOWPORT); 715 inp->inp_flags &= ~(INP_HIGHPORT); 716 break; 717 718 case IP_PORTRANGE_HIGH: 719 inp->inp_flags &= ~(INP_LOWPORT); 720 inp->inp_flags |= INP_HIGHPORT; 721 break; 722 723 case IP_PORTRANGE_LOW: 724 inp->inp_flags &= ~(INP_HIGHPORT); 725 inp->inp_flags |= INP_LOWPORT; 726 break; 727 728 default: 729 error = EINVAL; 730 break; 731 } 732 } 733 break; 734 735 default: 736 error = ENOPROTOOPT; 737 break; 738 } 739 if (m) 740 (void)m_free(m); 741 break; 742 743 case PRCO_GETOPT: 744 switch (optname) { 745 case IP_OPTIONS: 746 case IP_RETOPTS: 747 *mp = m = m_get(M_WAIT, MT_SOOPTS); 748 if (inp->inp_options) { 749 m->m_len = inp->inp_options->m_len; 750 bcopy(mtod(inp->inp_options, void *), 751 mtod(m, void *), m->m_len); 752 } else 753 m->m_len = 0; 754 break; 755 756 case IP_TOS: 757 case IP_TTL: 758 case IP_RECVOPTS: 759 case IP_RECVRETOPTS: 760 case IP_RECVDSTADDR: 761 case IP_RECVIF: 762 *mp = m = m_get(M_WAIT, MT_SOOPTS); 763 m->m_len = sizeof(int); 764 switch (optname) { 765 766 case IP_TOS: 767 optval = inp->inp_ip_tos; 768 break; 769 770 case IP_TTL: 771 optval = inp->inp_ip_ttl; 772 break; 773 774#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 775 776 case IP_RECVOPTS: 777 optval = OPTBIT(INP_RECVOPTS); 778 break; 779 780 case IP_RECVRETOPTS: 781 optval = OPTBIT(INP_RECVRETOPTS); 782 break; 783 784 case IP_RECVDSTADDR: 785 optval = OPTBIT(INP_RECVDSTADDR); 786 break; 787 788 case IP_RECVIF: 789 optval = OPTBIT(INP_RECVIF); 790 break; 791 } 792 *mtod(m, int *) = optval; 793 break; 794 795 case IP_MULTICAST_IF: 796 case IP_MULTICAST_VIF: 797 case IP_MULTICAST_TTL: 798 case IP_MULTICAST_LOOP: 799 case IP_ADD_MEMBERSHIP: 800 case IP_DROP_MEMBERSHIP: 801 error = ip_getmoptions(optname, inp->inp_moptions, mp); 802 break; 803 804 case IP_PORTRANGE: 805 *mp = m = m_get(M_WAIT, MT_SOOPTS); 806 m->m_len = sizeof(int); 807 808 if (inp->inp_flags & INP_HIGHPORT) 809 optval = IP_PORTRANGE_HIGH; 810 else if (inp->inp_flags & INP_LOWPORT) 811 optval = IP_PORTRANGE_LOW; 812 else 813 optval = 0; 814 815 *mtod(m, int *) = optval; 816 break; 817 818 default: 819 error = ENOPROTOOPT; 820 break; 821 } 822 break; 823 } 824 return (error); 825} 826 827/* 828 * Set up IP options in pcb for insertion in output packets. 829 * Store in mbuf with pointer in pcbopt, adding pseudo-option 830 * with destination address if source routed. 831 */ 832static int 833#ifdef notyet 834ip_pcbopts(optname, pcbopt, m) 835 int optname; 836#else 837ip_pcbopts(pcbopt, m) 838#endif 839 struct mbuf **pcbopt; 840 register struct mbuf *m; 841{ 842 register cnt, optlen; 843 register u_char *cp; 844 u_char opt; 845 846 /* turn off any old options */ 847 if (*pcbopt) 848 (void)m_free(*pcbopt); 849 *pcbopt = 0; 850 if (m == (struct mbuf *)0 || m->m_len == 0) { 851 /* 852 * Only turning off any previous options. 853 */ 854 if (m) 855 (void)m_free(m); 856 return (0); 857 } 858 859#ifndef vax 860 if (m->m_len % sizeof(long)) 861 goto bad; 862#endif 863 /* 864 * IP first-hop destination address will be stored before 865 * actual options; move other options back 866 * and clear it when none present. 867 */ 868 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 869 goto bad; 870 cnt = m->m_len; 871 m->m_len += sizeof(struct in_addr); 872 cp = mtod(m, u_char *) + sizeof(struct in_addr); 873 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 874 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 875 876 for (; cnt > 0; cnt -= optlen, cp += optlen) { 877 opt = cp[IPOPT_OPTVAL]; 878 if (opt == IPOPT_EOL) 879 break; 880 if (opt == IPOPT_NOP) 881 optlen = 1; 882 else { 883 optlen = cp[IPOPT_OLEN]; 884 if (optlen <= IPOPT_OLEN || optlen > cnt) 885 goto bad; 886 } 887 switch (opt) { 888 889 default: 890 break; 891 892 case IPOPT_LSRR: 893 case IPOPT_SSRR: 894 /* 895 * user process specifies route as: 896 * ->A->B->C->D 897 * D must be our final destination (but we can't 898 * check that since we may not have connected yet). 899 * A is first hop destination, which doesn't appear in 900 * actual IP option, but is stored before the options. 901 */ 902 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 903 goto bad; 904 m->m_len -= sizeof(struct in_addr); 905 cnt -= sizeof(struct in_addr); 906 optlen -= sizeof(struct in_addr); 907 cp[IPOPT_OLEN] = optlen; 908 /* 909 * Move first hop before start of options. 910 */ 911 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 912 sizeof(struct in_addr)); 913 /* 914 * Then copy rest of options back 915 * to close up the deleted entry. 916 */ 917 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 918 sizeof(struct in_addr)), 919 (caddr_t)&cp[IPOPT_OFFSET+1], 920 (unsigned)cnt + sizeof(struct in_addr)); 921 break; 922 } 923 } 924 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 925 goto bad; 926 *pcbopt = m; 927 return (0); 928 929bad: 930 (void)m_free(m); 931 return (EINVAL); 932} 933 934/* 935 * Set the IP multicast options in response to user setsockopt(). 936 */ 937static int 938ip_setmoptions(optname, imop, m) 939 int optname; 940 struct ip_moptions **imop; 941 struct mbuf *m; 942{ 943 register int error = 0; 944 u_char loop; 945 register int i; 946 struct in_addr addr; 947 register struct ip_mreq *mreq; 948 register struct ifnet *ifp; 949 register struct ip_moptions *imo = *imop; 950 struct route ro; 951 register struct sockaddr_in *dst; 952 int s; 953 954 if (imo == NULL) { 955 /* 956 * No multicast option buffer attached to the pcb; 957 * allocate one and initialize to default values. 958 */ 959 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 960 M_WAITOK); 961 962 if (imo == NULL) 963 return (ENOBUFS); 964 *imop = imo; 965 imo->imo_multicast_ifp = NULL; 966 imo->imo_multicast_vif = -1; 967 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 968 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 969 imo->imo_num_memberships = 0; 970 } 971 972 switch (optname) { 973 /* store an index number for the vif you wanna use in the send */ 974 case IP_MULTICAST_VIF: 975 if (!legal_vif_num) { 976 error = EOPNOTSUPP; 977 break; 978 } 979 if (m == NULL || m->m_len != sizeof(int)) { 980 error = EINVAL; 981 break; 982 } 983 i = *(mtod(m, int *)); 984 if (!legal_vif_num(i) && (i != -1)) { 985 error = EINVAL; 986 break; 987 } 988 imo->imo_multicast_vif = i; 989 break; 990 991 case IP_MULTICAST_IF: 992 /* 993 * Select the interface for outgoing multicast packets. 994 */ 995 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 996 error = EINVAL; 997 break; 998 } 999 addr = *(mtod(m, struct in_addr *)); 1000 /* 1001 * INADDR_ANY is used to remove a previous selection. 1002 * When no interface is selected, a default one is 1003 * chosen every time a multicast packet is sent. 1004 */ 1005 if (addr.s_addr == INADDR_ANY) { 1006 imo->imo_multicast_ifp = NULL; 1007 break; 1008 } 1009 /* 1010 * The selected interface is identified by its local 1011 * IP address. Find the interface and confirm that 1012 * it supports multicasting. 1013 */ 1014 s = splimp(); 1015 INADDR_TO_IFP(addr, ifp); 1016 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1017 splx(s); 1018 error = EADDRNOTAVAIL; 1019 break; 1020 } 1021 imo->imo_multicast_ifp = ifp; 1022 splx(s); 1023 break; 1024 1025 case IP_MULTICAST_TTL: 1026 /* 1027 * Set the IP time-to-live for outgoing multicast packets. 1028 */ 1029 if (m == NULL || m->m_len != 1) { 1030 error = EINVAL; 1031 break; 1032 } 1033 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1034 break; 1035 1036 case IP_MULTICAST_LOOP: 1037 /* 1038 * Set the loopback flag for outgoing multicast packets. 1039 * Must be zero or one. 1040 */ 1041 if (m == NULL || m->m_len != 1 || 1042 (loop = *(mtod(m, u_char *))) > 1) { 1043 error = EINVAL; 1044 break; 1045 } 1046 imo->imo_multicast_loop = loop; 1047 break; 1048 1049 case IP_ADD_MEMBERSHIP: 1050 /* 1051 * Add a multicast group membership. 1052 * Group must be a valid IP multicast address. 1053 */ 1054 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1055 error = EINVAL; 1056 break; 1057 } 1058 mreq = mtod(m, struct ip_mreq *); 1059 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1060 error = EINVAL; 1061 break; 1062 } 1063 s = splimp(); 1064 /* 1065 * If no interface address was provided, use the interface of 1066 * the route to the given multicast address. 1067 */ 1068 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1069 bzero((caddr_t)&ro, sizeof(ro)); 1070 dst = (struct sockaddr_in *)&ro.ro_dst; 1071 dst->sin_len = sizeof(*dst); 1072 dst->sin_family = AF_INET; 1073 dst->sin_addr = mreq->imr_multiaddr; 1074 rtalloc(&ro); 1075 if (ro.ro_rt == NULL) { 1076 error = EADDRNOTAVAIL; 1077 splx(s); 1078 break; 1079 } 1080 ifp = ro.ro_rt->rt_ifp; 1081 rtfree(ro.ro_rt); 1082 } 1083 else { 1084 INADDR_TO_IFP(mreq->imr_interface, ifp); 1085 } 1086 1087 /* 1088 * See if we found an interface, and confirm that it 1089 * supports multicast. 1090 */ 1091 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1092 error = EADDRNOTAVAIL; 1093 splx(s); 1094 break; 1095 } 1096 /* 1097 * See if the membership already exists or if all the 1098 * membership slots are full. 1099 */ 1100 for (i = 0; i < imo->imo_num_memberships; ++i) { 1101 if (imo->imo_membership[i]->inm_ifp == ifp && 1102 imo->imo_membership[i]->inm_addr.s_addr 1103 == mreq->imr_multiaddr.s_addr) 1104 break; 1105 } 1106 if (i < imo->imo_num_memberships) { 1107 error = EADDRINUSE; 1108 splx(s); 1109 break; 1110 } 1111 if (i == IP_MAX_MEMBERSHIPS) { 1112 error = ETOOMANYREFS; 1113 splx(s); 1114 break; 1115 } 1116 /* 1117 * Everything looks good; add a new record to the multicast 1118 * address list for the given interface. 1119 */ 1120 if ((imo->imo_membership[i] = 1121 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1122 error = ENOBUFS; 1123 splx(s); 1124 break; 1125 } 1126 ++imo->imo_num_memberships; 1127 splx(s); 1128 break; 1129 1130 case IP_DROP_MEMBERSHIP: 1131 /* 1132 * Drop a multicast group membership. 1133 * Group must be a valid IP multicast address. 1134 */ 1135 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1136 error = EINVAL; 1137 break; 1138 } 1139 mreq = mtod(m, struct ip_mreq *); 1140 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1141 error = EINVAL; 1142 break; 1143 } 1144 1145 s = splimp(); 1146 /* 1147 * If an interface address was specified, get a pointer 1148 * to its ifnet structure. 1149 */ 1150 if (mreq->imr_interface.s_addr == INADDR_ANY) 1151 ifp = NULL; 1152 else { 1153 INADDR_TO_IFP(mreq->imr_interface, ifp); 1154 if (ifp == NULL) { 1155 error = EADDRNOTAVAIL; 1156 splx(s); 1157 break; 1158 } 1159 } 1160 /* 1161 * Find the membership in the membership array. 1162 */ 1163 for (i = 0; i < imo->imo_num_memberships; ++i) { 1164 if ((ifp == NULL || 1165 imo->imo_membership[i]->inm_ifp == ifp) && 1166 imo->imo_membership[i]->inm_addr.s_addr == 1167 mreq->imr_multiaddr.s_addr) 1168 break; 1169 } 1170 if (i == imo->imo_num_memberships) { 1171 error = EADDRNOTAVAIL; 1172 splx(s); 1173 break; 1174 } 1175 /* 1176 * Give up the multicast address record to which the 1177 * membership points. 1178 */ 1179 in_delmulti(imo->imo_membership[i]); 1180 /* 1181 * Remove the gap in the membership array. 1182 */ 1183 for (++i; i < imo->imo_num_memberships; ++i) 1184 imo->imo_membership[i-1] = imo->imo_membership[i]; 1185 --imo->imo_num_memberships; 1186 splx(s); 1187 break; 1188 1189 default: 1190 error = EOPNOTSUPP; 1191 break; 1192 } 1193 1194 /* 1195 * If all options have default values, no need to keep the mbuf. 1196 */ 1197 if (imo->imo_multicast_ifp == NULL && 1198 imo->imo_multicast_vif == -1 && 1199 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1200 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1201 imo->imo_num_memberships == 0) { 1202 free(*imop, M_IPMOPTS); 1203 *imop = NULL; 1204 } 1205 1206 return (error); 1207} 1208 1209/* 1210 * Return the IP multicast options in response to user getsockopt(). 1211 */ 1212static int 1213ip_getmoptions(optname, imo, mp) 1214 int optname; 1215 register struct ip_moptions *imo; 1216 register struct mbuf **mp; 1217{ 1218 u_char *ttl; 1219 u_char *loop; 1220 struct in_addr *addr; 1221 struct in_ifaddr *ia; 1222 1223 *mp = m_get(M_WAIT, MT_SOOPTS); 1224 1225 switch (optname) { 1226 1227 case IP_MULTICAST_VIF: 1228 if (imo != NULL) 1229 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1230 else 1231 *(mtod(*mp, int *)) = -1; 1232 (*mp)->m_len = sizeof(int); 1233 return(0); 1234 1235 case IP_MULTICAST_IF: 1236 addr = mtod(*mp, struct in_addr *); 1237 (*mp)->m_len = sizeof(struct in_addr); 1238 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1239 addr->s_addr = INADDR_ANY; 1240 else { 1241 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1242 addr->s_addr = (ia == NULL) ? INADDR_ANY 1243 : IA_SIN(ia)->sin_addr.s_addr; 1244 } 1245 return (0); 1246 1247 case IP_MULTICAST_TTL: 1248 ttl = mtod(*mp, u_char *); 1249 (*mp)->m_len = 1; 1250 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1251 : imo->imo_multicast_ttl; 1252 return (0); 1253 1254 case IP_MULTICAST_LOOP: 1255 loop = mtod(*mp, u_char *); 1256 (*mp)->m_len = 1; 1257 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1258 : imo->imo_multicast_loop; 1259 return (0); 1260 1261 default: 1262 return (EOPNOTSUPP); 1263 } 1264} 1265 1266/* 1267 * Discard the IP multicast options. 1268 */ 1269void 1270ip_freemoptions(imo) 1271 register struct ip_moptions *imo; 1272{ 1273 register int i; 1274 1275 if (imo != NULL) { 1276 for (i = 0; i < imo->imo_num_memberships; ++i) 1277 in_delmulti(imo->imo_membership[i]); 1278 free(imo, M_IPMOPTS); 1279 } 1280} 1281 1282/* 1283 * Routine called from ip_output() to loop back a copy of an IP multicast 1284 * packet to the input queue of a specified interface. Note that this 1285 * calls the output routine of the loopback "driver", but with an interface 1286 * pointer that might NOT be a loopback interface -- evil, but easier than 1287 * replicating that code here. 1288 */ 1289static void 1290ip_mloopback(ifp, m, dst, hlen) 1291 struct ifnet *ifp; 1292 register struct mbuf *m; 1293 register struct sockaddr_in *dst; 1294 int hlen; 1295{ 1296 register struct ip *ip; 1297 struct mbuf *copym; 1298 1299 copym = m_copy(m, 0, M_COPYALL); 1300 if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen)) 1301 copym = m_pullup(copym, hlen); 1302 if (copym != NULL) { 1303 /* 1304 * We don't bother to fragment if the IP length is greater 1305 * than the interface's MTU. Can this possibly matter? 1306 */ 1307 ip = mtod(copym, struct ip *); 1308 ip->ip_len = htons((u_short)ip->ip_len); 1309 ip->ip_off = htons((u_short)ip->ip_off); 1310 ip->ip_sum = 0; 1311 if (ip->ip_vhl == IP_VHL_BORING) { 1312 ip->ip_sum = in_cksum_hdr(ip); 1313 } else { 1314 ip->ip_sum = in_cksum(copym, hlen); 1315 } 1316 /* 1317 * NB: 1318 * It's not clear whether there are any lingering 1319 * reentrancy problems in other areas which might 1320 * be exposed by using ip_input directly (in 1321 * particular, everything which modifies the packet 1322 * in-place). Yet another option is using the 1323 * protosw directly to deliver the looped back 1324 * packet. For the moment, we'll err on the side 1325 * of safety by continuing to abuse looutput(). 1326 */ 1327#ifdef notdef 1328 copym->m_pkthdr.rcvif = ifp; 1329 ip_input(copym) 1330#else 1331 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1332#endif 1333 } 1334} 1335