ip_output.c revision 25201
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 1.54 1997/04/03 10:47:12 darrenr 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_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); 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 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 int action; 376 377#ifdef IPDIVERT 378 action = (*ip_fw_chk_ptr)(&ip, 379 hlen, ifp, (~0 << 16) | ip_divert_ignore, &m); 380 ip_divert_ignore = 0; 381#else 382 action = (*ip_fw_chk_ptr)(&ip, hlen, ifp, (~0 << 16), &m); 383#endif 384 if (action == -1) { 385 error = EACCES; /* XXX is this appropriate? */ 386 goto done; 387 } else if (action != 0) { 388#ifdef IPDIVERT 389 ip_divert_port = action; /* divert to port */ 390 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 391 goto done; 392#else 393 m_freem(m); /* ipfw says divert, but we can't */ 394 goto done; 395#endif 396 } 397 } 398#endif /* COMPAT_IPFW */ 399 400 /* 401 * If small enough for interface, can just send directly. 402 */ 403 if ((u_short)ip->ip_len <= ifp->if_mtu) { 404 ip->ip_len = htons((u_short)ip->ip_len); 405 ip->ip_off = htons((u_short)ip->ip_off); 406 ip->ip_sum = 0; 407 if (ip->ip_vhl == IP_VHL_BORING) { 408 ip->ip_sum = in_cksum_hdr(ip); 409 } else { 410 ip->ip_sum = in_cksum(m, hlen); 411 } 412 error = (*ifp->if_output)(ifp, m, 413 (struct sockaddr *)dst, ro->ro_rt); 414 goto done; 415 } 416 /* 417 * Too large for interface; fragment if possible. 418 * Must be able to put at least 8 bytes per fragment. 419 */ 420 if (ip->ip_off & IP_DF) { 421 error = EMSGSIZE; 422 /* 423 * This case can happen if the user changed the MTU 424 * of an interface after enabling IP on it. Because 425 * most netifs don't keep track of routes pointing to 426 * them, there is no way for one to update all its 427 * routes when the MTU is changed. 428 */ 429 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 430 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 431 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 432 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 433 } 434 ipstat.ips_cantfrag++; 435 goto bad; 436 } 437 len = (ifp->if_mtu - hlen) &~ 7; 438 if (len < 8) { 439 error = EMSGSIZE; 440 goto bad; 441 } 442 443 { 444 int mhlen, firstlen = len; 445 struct mbuf **mnext = &m->m_nextpkt; 446 447 /* 448 * Loop through length of segment after first fragment, 449 * make new header and copy data of each part and link onto chain. 450 */ 451 m0 = m; 452 mhlen = sizeof (struct ip); 453 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 454 MGETHDR(m, M_DONTWAIT, MT_HEADER); 455 if (m == 0) { 456 error = ENOBUFS; 457 ipstat.ips_odropped++; 458 goto sendorfree; 459 } 460 m->m_data += max_linkhdr; 461 mhip = mtod(m, struct ip *); 462 *mhip = *ip; 463 if (hlen > sizeof (struct ip)) { 464 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 465 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 466 } 467 m->m_len = mhlen; 468 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 469 if (ip->ip_off & IP_MF) 470 mhip->ip_off |= IP_MF; 471 if (off + len >= (u_short)ip->ip_len) 472 len = (u_short)ip->ip_len - off; 473 else 474 mhip->ip_off |= IP_MF; 475 mhip->ip_len = htons((u_short)(len + mhlen)); 476 m->m_next = m_copy(m0, off, len); 477 if (m->m_next == 0) { 478 (void) m_free(m); 479 error = ENOBUFS; /* ??? */ 480 ipstat.ips_odropped++; 481 goto sendorfree; 482 } 483 m->m_pkthdr.len = mhlen + len; 484 m->m_pkthdr.rcvif = (struct ifnet *)0; 485 mhip->ip_off = htons((u_short)mhip->ip_off); 486 mhip->ip_sum = 0; 487 if (mhip->ip_vhl == IP_VHL_BORING) { 488 mhip->ip_sum = in_cksum_hdr(mhip); 489 } else { 490 mhip->ip_sum = in_cksum(m, mhlen); 491 } 492 *mnext = m; 493 mnext = &m->m_nextpkt; 494 ipstat.ips_ofragments++; 495 } 496 /* 497 * Update first fragment by trimming what's been copied out 498 * and updating header, then send each fragment (in order). 499 */ 500 m = m0; 501 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 502 m->m_pkthdr.len = hlen + firstlen; 503 ip->ip_len = htons((u_short)m->m_pkthdr.len); 504 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 505 ip->ip_sum = 0; 506 if (ip->ip_vhl == IP_VHL_BORING) { 507 ip->ip_sum = in_cksum_hdr(ip); 508 } else { 509 ip->ip_sum = in_cksum(m, hlen); 510 } 511sendorfree: 512 for (m = m0; m; m = m0) { 513 m0 = m->m_nextpkt; 514 m->m_nextpkt = 0; 515 if (error == 0) 516 error = (*ifp->if_output)(ifp, m, 517 (struct sockaddr *)dst, ro->ro_rt); 518 else 519 m_freem(m); 520 } 521 522 if (error == 0) 523 ipstat.ips_fragmented++; 524 } 525done: 526 return (error); 527bad: 528 m_freem(m0); 529 goto done; 530} 531 532/* 533 * Insert IP options into preformed packet. 534 * Adjust IP destination as required for IP source routing, 535 * as indicated by a non-zero in_addr at the start of the options. 536 * 537 * XXX This routine assumes that the packet has no options in place. 538 */ 539static struct mbuf * 540ip_insertoptions(m, opt, phlen) 541 register struct mbuf *m; 542 struct mbuf *opt; 543 int *phlen; 544{ 545 register struct ipoption *p = mtod(opt, struct ipoption *); 546 struct mbuf *n; 547 register struct ip *ip = mtod(m, struct ip *); 548 unsigned optlen; 549 550 optlen = opt->m_len - sizeof(p->ipopt_dst); 551 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 552 return (m); /* XXX should fail */ 553 if (p->ipopt_dst.s_addr) 554 ip->ip_dst = p->ipopt_dst; 555 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 556 MGETHDR(n, M_DONTWAIT, MT_HEADER); 557 if (n == 0) 558 return (m); 559 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 560 m->m_len -= sizeof(struct ip); 561 m->m_data += sizeof(struct ip); 562 n->m_next = m; 563 m = n; 564 m->m_len = optlen + sizeof(struct ip); 565 m->m_data += max_linkhdr; 566 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 567 } else { 568 m->m_data -= optlen; 569 m->m_len += optlen; 570 m->m_pkthdr.len += optlen; 571 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 572 } 573 ip = mtod(m, struct ip *); 574 bcopy(p->ipopt_list, ip + 1, optlen); 575 *phlen = sizeof(struct ip) + optlen; 576 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 577 ip->ip_len += optlen; 578 return (m); 579} 580 581/* 582 * Copy options from ip to jp, 583 * omitting those not copied during fragmentation. 584 */ 585#if !defined(IPFILTER) && !defined(IPFILTER_LKM) 586static 587#endif 588int 589ip_optcopy(ip, jp) 590 struct ip *ip, *jp; 591{ 592 register u_char *cp, *dp; 593 int opt, optlen, cnt; 594 595 cp = (u_char *)(ip + 1); 596 dp = (u_char *)(jp + 1); 597 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 598 for (; cnt > 0; cnt -= optlen, cp += optlen) { 599 opt = cp[0]; 600 if (opt == IPOPT_EOL) 601 break; 602 if (opt == IPOPT_NOP) { 603 /* Preserve for IP mcast tunnel's LSRR alignment. */ 604 *dp++ = IPOPT_NOP; 605 optlen = 1; 606 continue; 607 } else 608 optlen = cp[IPOPT_OLEN]; 609 /* bogus lengths should have been caught by ip_dooptions */ 610 if (optlen > cnt) 611 optlen = cnt; 612 if (IPOPT_COPIED(opt)) { 613 bcopy(cp, dp, optlen); 614 dp += optlen; 615 } 616 } 617 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 618 *dp++ = IPOPT_EOL; 619 return (optlen); 620} 621 622/* 623 * IP socket option processing. 624 */ 625int 626ip_ctloutput(op, so, level, optname, mp, p) 627 int op; 628 struct socket *so; 629 int level, optname; 630 struct mbuf **mp; 631 struct proc *p; 632{ 633 register struct inpcb *inp = sotoinpcb(so); 634 register struct mbuf *m = *mp; 635 register int optval = 0; 636 int error = 0; 637 638 if (level != IPPROTO_IP) { 639 error = EINVAL; 640 if (op == PRCO_SETOPT && *mp) 641 (void) m_free(*mp); 642 } else switch (op) { 643 644 case PRCO_SETOPT: 645 switch (optname) { 646 case IP_OPTIONS: 647#ifdef notyet 648 case IP_RETOPTS: 649 return (ip_pcbopts(optname, &inp->inp_options, m)); 650#else 651 return (ip_pcbopts(&inp->inp_options, m)); 652#endif 653 654 case IP_TOS: 655 case IP_TTL: 656 case IP_RECVOPTS: 657 case IP_RECVRETOPTS: 658 case IP_RECVDSTADDR: 659 case IP_RECVIF: 660 if (m == 0 || m->m_len != sizeof(int)) 661 error = EINVAL; 662 else { 663 optval = *mtod(m, int *); 664 switch (optname) { 665 666 case IP_TOS: 667 inp->inp_ip_tos = optval; 668 break; 669 670 case IP_TTL: 671 inp->inp_ip_ttl = optval; 672 break; 673#define OPTSET(bit) \ 674 if (optval) \ 675 inp->inp_flags |= bit; \ 676 else \ 677 inp->inp_flags &= ~bit; 678 679 case IP_RECVOPTS: 680 OPTSET(INP_RECVOPTS); 681 break; 682 683 case IP_RECVRETOPTS: 684 OPTSET(INP_RECVRETOPTS); 685 break; 686 687 case IP_RECVDSTADDR: 688 OPTSET(INP_RECVDSTADDR); 689 break; 690 691 case IP_RECVIF: 692 OPTSET(INP_RECVIF); 693 break; 694 } 695 } 696 break; 697#undef OPTSET 698 699 case IP_MULTICAST_IF: 700 case IP_MULTICAST_VIF: 701 case IP_MULTICAST_TTL: 702 case IP_MULTICAST_LOOP: 703 case IP_ADD_MEMBERSHIP: 704 case IP_DROP_MEMBERSHIP: 705 error = ip_setmoptions(optname, &inp->inp_moptions, m); 706 break; 707 708 case IP_PORTRANGE: 709 if (m == 0 || m->m_len != sizeof(int)) 710 error = EINVAL; 711 else { 712 optval = *mtod(m, int *); 713 714 switch (optval) { 715 716 case IP_PORTRANGE_DEFAULT: 717 inp->inp_flags &= ~(INP_LOWPORT); 718 inp->inp_flags &= ~(INP_HIGHPORT); 719 break; 720 721 case IP_PORTRANGE_HIGH: 722 inp->inp_flags &= ~(INP_LOWPORT); 723 inp->inp_flags |= INP_HIGHPORT; 724 break; 725 726 case IP_PORTRANGE_LOW: 727 inp->inp_flags &= ~(INP_HIGHPORT); 728 inp->inp_flags |= INP_LOWPORT; 729 break; 730 731 default: 732 error = EINVAL; 733 break; 734 } 735 } 736 break; 737 738 default: 739 error = ENOPROTOOPT; 740 break; 741 } 742 if (m) 743 (void)m_free(m); 744 break; 745 746 case PRCO_GETOPT: 747 switch (optname) { 748 case IP_OPTIONS: 749 case IP_RETOPTS: 750 *mp = m = m_get(M_WAIT, MT_SOOPTS); 751 if (inp->inp_options) { 752 m->m_len = inp->inp_options->m_len; 753 bcopy(mtod(inp->inp_options, void *), 754 mtod(m, void *), m->m_len); 755 } else 756 m->m_len = 0; 757 break; 758 759 case IP_TOS: 760 case IP_TTL: 761 case IP_RECVOPTS: 762 case IP_RECVRETOPTS: 763 case IP_RECVDSTADDR: 764 case IP_RECVIF: 765 *mp = m = m_get(M_WAIT, MT_SOOPTS); 766 m->m_len = sizeof(int); 767 switch (optname) { 768 769 case IP_TOS: 770 optval = inp->inp_ip_tos; 771 break; 772 773 case IP_TTL: 774 optval = inp->inp_ip_ttl; 775 break; 776 777#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 778 779 case IP_RECVOPTS: 780 optval = OPTBIT(INP_RECVOPTS); 781 break; 782 783 case IP_RECVRETOPTS: 784 optval = OPTBIT(INP_RECVRETOPTS); 785 break; 786 787 case IP_RECVDSTADDR: 788 optval = OPTBIT(INP_RECVDSTADDR); 789 break; 790 791 case IP_RECVIF: 792 optval = OPTBIT(INP_RECVIF); 793 break; 794 } 795 *mtod(m, int *) = optval; 796 break; 797 798 case IP_MULTICAST_IF: 799 case IP_MULTICAST_VIF: 800 case IP_MULTICAST_TTL: 801 case IP_MULTICAST_LOOP: 802 case IP_ADD_MEMBERSHIP: 803 case IP_DROP_MEMBERSHIP: 804 error = ip_getmoptions(optname, inp->inp_moptions, mp); 805 break; 806 807 case IP_PORTRANGE: 808 *mp = m = m_get(M_WAIT, MT_SOOPTS); 809 m->m_len = sizeof(int); 810 811 if (inp->inp_flags & INP_HIGHPORT) 812 optval = IP_PORTRANGE_HIGH; 813 else if (inp->inp_flags & INP_LOWPORT) 814 optval = IP_PORTRANGE_LOW; 815 else 816 optval = 0; 817 818 *mtod(m, int *) = optval; 819 break; 820 821 default: 822 error = ENOPROTOOPT; 823 break; 824 } 825 break; 826 } 827 return (error); 828} 829 830/* 831 * Set up IP options in pcb for insertion in output packets. 832 * Store in mbuf with pointer in pcbopt, adding pseudo-option 833 * with destination address if source routed. 834 */ 835static int 836#ifdef notyet 837ip_pcbopts(optname, pcbopt, m) 838 int optname; 839#else 840ip_pcbopts(pcbopt, m) 841#endif 842 struct mbuf **pcbopt; 843 register struct mbuf *m; 844{ 845 register cnt, optlen; 846 register u_char *cp; 847 u_char opt; 848 849 /* turn off any old options */ 850 if (*pcbopt) 851 (void)m_free(*pcbopt); 852 *pcbopt = 0; 853 if (m == (struct mbuf *)0 || m->m_len == 0) { 854 /* 855 * Only turning off any previous options. 856 */ 857 if (m) 858 (void)m_free(m); 859 return (0); 860 } 861 862#ifndef vax 863 if (m->m_len % sizeof(long)) 864 goto bad; 865#endif 866 /* 867 * IP first-hop destination address will be stored before 868 * actual options; move other options back 869 * and clear it when none present. 870 */ 871 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 872 goto bad; 873 cnt = m->m_len; 874 m->m_len += sizeof(struct in_addr); 875 cp = mtod(m, u_char *) + sizeof(struct in_addr); 876 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 877 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 878 879 for (; cnt > 0; cnt -= optlen, cp += optlen) { 880 opt = cp[IPOPT_OPTVAL]; 881 if (opt == IPOPT_EOL) 882 break; 883 if (opt == IPOPT_NOP) 884 optlen = 1; 885 else { 886 optlen = cp[IPOPT_OLEN]; 887 if (optlen <= IPOPT_OLEN || optlen > cnt) 888 goto bad; 889 } 890 switch (opt) { 891 892 default: 893 break; 894 895 case IPOPT_LSRR: 896 case IPOPT_SSRR: 897 /* 898 * user process specifies route as: 899 * ->A->B->C->D 900 * D must be our final destination (but we can't 901 * check that since we may not have connected yet). 902 * A is first hop destination, which doesn't appear in 903 * actual IP option, but is stored before the options. 904 */ 905 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 906 goto bad; 907 m->m_len -= sizeof(struct in_addr); 908 cnt -= sizeof(struct in_addr); 909 optlen -= sizeof(struct in_addr); 910 cp[IPOPT_OLEN] = optlen; 911 /* 912 * Move first hop before start of options. 913 */ 914 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 915 sizeof(struct in_addr)); 916 /* 917 * Then copy rest of options back 918 * to close up the deleted entry. 919 */ 920 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 921 sizeof(struct in_addr)), 922 (caddr_t)&cp[IPOPT_OFFSET+1], 923 (unsigned)cnt + sizeof(struct in_addr)); 924 break; 925 } 926 } 927 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 928 goto bad; 929 *pcbopt = m; 930 return (0); 931 932bad: 933 (void)m_free(m); 934 return (EINVAL); 935} 936 937/* 938 * Set the IP multicast options in response to user setsockopt(). 939 */ 940static int 941ip_setmoptions(optname, imop, m) 942 int optname; 943 struct ip_moptions **imop; 944 struct mbuf *m; 945{ 946 register int error = 0; 947 u_char loop; 948 register int i; 949 struct in_addr addr; 950 register struct ip_mreq *mreq; 951 register struct ifnet *ifp; 952 register struct ip_moptions *imo = *imop; 953 struct route ro; 954 register struct sockaddr_in *dst; 955 int s; 956 957 if (imo == NULL) { 958 /* 959 * No multicast option buffer attached to the pcb; 960 * allocate one and initialize to default values. 961 */ 962 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 963 M_WAITOK); 964 965 if (imo == NULL) 966 return (ENOBUFS); 967 *imop = imo; 968 imo->imo_multicast_ifp = NULL; 969 imo->imo_multicast_vif = -1; 970 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 971 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 972 imo->imo_num_memberships = 0; 973 } 974 975 switch (optname) { 976 /* store an index number for the vif you wanna use in the send */ 977 case IP_MULTICAST_VIF: 978 if (!legal_vif_num) { 979 error = EOPNOTSUPP; 980 break; 981 } 982 if (m == NULL || m->m_len != sizeof(int)) { 983 error = EINVAL; 984 break; 985 } 986 i = *(mtod(m, int *)); 987 if (!legal_vif_num(i) && (i != -1)) { 988 error = EINVAL; 989 break; 990 } 991 imo->imo_multicast_vif = i; 992 break; 993 994 case IP_MULTICAST_IF: 995 /* 996 * Select the interface for outgoing multicast packets. 997 */ 998 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 999 error = EINVAL; 1000 break; 1001 } 1002 addr = *(mtod(m, struct in_addr *)); 1003 /* 1004 * INADDR_ANY is used to remove a previous selection. 1005 * When no interface is selected, a default one is 1006 * chosen every time a multicast packet is sent. 1007 */ 1008 if (addr.s_addr == INADDR_ANY) { 1009 imo->imo_multicast_ifp = NULL; 1010 break; 1011 } 1012 /* 1013 * The selected interface is identified by its local 1014 * IP address. Find the interface and confirm that 1015 * it supports multicasting. 1016 */ 1017 s = splimp(); 1018 INADDR_TO_IFP(addr, ifp); 1019 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1020 splx(s); 1021 error = EADDRNOTAVAIL; 1022 break; 1023 } 1024 imo->imo_multicast_ifp = ifp; 1025 splx(s); 1026 break; 1027 1028 case IP_MULTICAST_TTL: 1029 /* 1030 * Set the IP time-to-live for outgoing multicast packets. 1031 */ 1032 if (m == NULL || m->m_len != 1) { 1033 error = EINVAL; 1034 break; 1035 } 1036 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1037 break; 1038 1039 case IP_MULTICAST_LOOP: 1040 /* 1041 * Set the loopback flag for outgoing multicast packets. 1042 * Must be zero or one. 1043 */ 1044 if (m == NULL || m->m_len != 1 || 1045 (loop = *(mtod(m, u_char *))) > 1) { 1046 error = EINVAL; 1047 break; 1048 } 1049 imo->imo_multicast_loop = loop; 1050 break; 1051 1052 case IP_ADD_MEMBERSHIP: 1053 /* 1054 * Add a multicast group membership. 1055 * Group must be a valid IP multicast address. 1056 */ 1057 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1058 error = EINVAL; 1059 break; 1060 } 1061 mreq = mtod(m, struct ip_mreq *); 1062 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1063 error = EINVAL; 1064 break; 1065 } 1066 s = splimp(); 1067 /* 1068 * If no interface address was provided, use the interface of 1069 * the route to the given multicast address. 1070 */ 1071 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1072 bzero((caddr_t)&ro, sizeof(ro)); 1073 dst = (struct sockaddr_in *)&ro.ro_dst; 1074 dst->sin_len = sizeof(*dst); 1075 dst->sin_family = AF_INET; 1076 dst->sin_addr = mreq->imr_multiaddr; 1077 rtalloc(&ro); 1078 if (ro.ro_rt == NULL) { 1079 error = EADDRNOTAVAIL; 1080 splx(s); 1081 break; 1082 } 1083 ifp = ro.ro_rt->rt_ifp; 1084 rtfree(ro.ro_rt); 1085 } 1086 else { 1087 INADDR_TO_IFP(mreq->imr_interface, ifp); 1088 } 1089 1090 /* 1091 * See if we found an interface, and confirm that it 1092 * supports multicast. 1093 */ 1094 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1095 error = EADDRNOTAVAIL; 1096 splx(s); 1097 break; 1098 } 1099 /* 1100 * See if the membership already exists or if all the 1101 * membership slots are full. 1102 */ 1103 for (i = 0; i < imo->imo_num_memberships; ++i) { 1104 if (imo->imo_membership[i]->inm_ifp == ifp && 1105 imo->imo_membership[i]->inm_addr.s_addr 1106 == mreq->imr_multiaddr.s_addr) 1107 break; 1108 } 1109 if (i < imo->imo_num_memberships) { 1110 error = EADDRINUSE; 1111 splx(s); 1112 break; 1113 } 1114 if (i == IP_MAX_MEMBERSHIPS) { 1115 error = ETOOMANYREFS; 1116 splx(s); 1117 break; 1118 } 1119 /* 1120 * Everything looks good; add a new record to the multicast 1121 * address list for the given interface. 1122 */ 1123 if ((imo->imo_membership[i] = 1124 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1125 error = ENOBUFS; 1126 splx(s); 1127 break; 1128 } 1129 ++imo->imo_num_memberships; 1130 splx(s); 1131 break; 1132 1133 case IP_DROP_MEMBERSHIP: 1134 /* 1135 * Drop a multicast group membership. 1136 * Group must be a valid IP multicast address. 1137 */ 1138 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1139 error = EINVAL; 1140 break; 1141 } 1142 mreq = mtod(m, struct ip_mreq *); 1143 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1144 error = EINVAL; 1145 break; 1146 } 1147 1148 s = splimp(); 1149 /* 1150 * If an interface address was specified, get a pointer 1151 * to its ifnet structure. 1152 */ 1153 if (mreq->imr_interface.s_addr == INADDR_ANY) 1154 ifp = NULL; 1155 else { 1156 INADDR_TO_IFP(mreq->imr_interface, ifp); 1157 if (ifp == NULL) { 1158 error = EADDRNOTAVAIL; 1159 splx(s); 1160 break; 1161 } 1162 } 1163 /* 1164 * Find the membership in the membership array. 1165 */ 1166 for (i = 0; i < imo->imo_num_memberships; ++i) { 1167 if ((ifp == NULL || 1168 imo->imo_membership[i]->inm_ifp == ifp) && 1169 imo->imo_membership[i]->inm_addr.s_addr == 1170 mreq->imr_multiaddr.s_addr) 1171 break; 1172 } 1173 if (i == imo->imo_num_memberships) { 1174 error = EADDRNOTAVAIL; 1175 splx(s); 1176 break; 1177 } 1178 /* 1179 * Give up the multicast address record to which the 1180 * membership points. 1181 */ 1182 in_delmulti(imo->imo_membership[i]); 1183 /* 1184 * Remove the gap in the membership array. 1185 */ 1186 for (++i; i < imo->imo_num_memberships; ++i) 1187 imo->imo_membership[i-1] = imo->imo_membership[i]; 1188 --imo->imo_num_memberships; 1189 splx(s); 1190 break; 1191 1192 default: 1193 error = EOPNOTSUPP; 1194 break; 1195 } 1196 1197 /* 1198 * If all options have default values, no need to keep the mbuf. 1199 */ 1200 if (imo->imo_multicast_ifp == NULL && 1201 imo->imo_multicast_vif == -1 && 1202 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1203 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1204 imo->imo_num_memberships == 0) { 1205 free(*imop, M_IPMOPTS); 1206 *imop = NULL; 1207 } 1208 1209 return (error); 1210} 1211 1212/* 1213 * Return the IP multicast options in response to user getsockopt(). 1214 */ 1215static int 1216ip_getmoptions(optname, imo, mp) 1217 int optname; 1218 register struct ip_moptions *imo; 1219 register struct mbuf **mp; 1220{ 1221 u_char *ttl; 1222 u_char *loop; 1223 struct in_addr *addr; 1224 struct in_ifaddr *ia; 1225 1226 *mp = m_get(M_WAIT, MT_SOOPTS); 1227 1228 switch (optname) { 1229 1230 case IP_MULTICAST_VIF: 1231 if (imo != NULL) 1232 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1233 else 1234 *(mtod(*mp, int *)) = -1; 1235 (*mp)->m_len = sizeof(int); 1236 return(0); 1237 1238 case IP_MULTICAST_IF: 1239 addr = mtod(*mp, struct in_addr *); 1240 (*mp)->m_len = sizeof(struct in_addr); 1241 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1242 addr->s_addr = INADDR_ANY; 1243 else { 1244 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1245 addr->s_addr = (ia == NULL) ? INADDR_ANY 1246 : IA_SIN(ia)->sin_addr.s_addr; 1247 } 1248 return (0); 1249 1250 case IP_MULTICAST_TTL: 1251 ttl = mtod(*mp, u_char *); 1252 (*mp)->m_len = 1; 1253 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1254 : imo->imo_multicast_ttl; 1255 return (0); 1256 1257 case IP_MULTICAST_LOOP: 1258 loop = mtod(*mp, u_char *); 1259 (*mp)->m_len = 1; 1260 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1261 : imo->imo_multicast_loop; 1262 return (0); 1263 1264 default: 1265 return (EOPNOTSUPP); 1266 } 1267} 1268 1269/* 1270 * Discard the IP multicast options. 1271 */ 1272void 1273ip_freemoptions(imo) 1274 register struct ip_moptions *imo; 1275{ 1276 register int i; 1277 1278 if (imo != NULL) { 1279 for (i = 0; i < imo->imo_num_memberships; ++i) 1280 in_delmulti(imo->imo_membership[i]); 1281 free(imo, M_IPMOPTS); 1282 } 1283} 1284 1285/* 1286 * Routine called from ip_output() to loop back a copy of an IP multicast 1287 * packet to the input queue of a specified interface. Note that this 1288 * calls the output routine of the loopback "driver", but with an interface 1289 * pointer that might NOT be a loopback interface -- evil, but easier than 1290 * replicating that code here. 1291 */ 1292static void 1293ip_mloopback(ifp, m, dst) 1294 struct ifnet *ifp; 1295 register struct mbuf *m; 1296 register struct sockaddr_in *dst; 1297{ 1298 register struct ip *ip; 1299 struct mbuf *copym; 1300 1301 copym = m_copy(m, 0, M_COPYALL); 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, 1315 IP_VHL_HL(ip->ip_vhl) << 2); 1316 } 1317 /* 1318 * NB: 1319 * It's not clear whether there are any lingering 1320 * reentrancy problems in other areas which might 1321 * be exposed by using ip_input directly (in 1322 * particular, everything which modifies the packet 1323 * in-place). Yet another option is using the 1324 * protosw directly to deliver the looped back 1325 * packet. For the moment, we'll err on the side 1326 * of safety by continuing to abuse looutput(). 1327 */ 1328#ifdef notdef 1329 copym->m_pkthdr.rcvif = ifp; 1330 ip_input(copym) 1331#else 1332 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1333#endif 1334 } 1335} 1336