ip_output.c revision 33134
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 1.63 1998/02/04 22:33:09 eivind Exp $ 35 */ 36 37#define _IP_VHL 38 39#include "opt_ipdivert.h" 40 41#include <sys/param.h> 42#include <sys/systm.h> 43#include <sys/malloc.h> 44#include <sys/mbuf.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 64static MALLOC_DEFINE(M_IPMOPTS, "ip_moptions", "internet multicast options"); 65 66#if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1 67#undef COMPAT_IPFW 68#define COMPAT_IPFW 1 69#else 70#undef COMPAT_IPFW 71#endif 72 73u_short ip_id; 74 75static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 76static void ip_mloopback 77 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *, int)); 78static int ip_getmoptions 79 __P((int, struct ip_moptions *, struct mbuf **)); 80static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 81static int ip_setmoptions 82 __P((int, struct ip_moptions **, struct mbuf *)); 83 84#if defined(IPFILTER_LKM) || defined(IPFILTER) 85int ip_optcopy __P((struct ip *, struct ip *)); 86extern int fr_check __P((struct ip *, int, struct ifnet *, int, struct mbuf **)); 87extern int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **)); 88#else 89static int ip_optcopy __P((struct ip *, struct ip *)); 90#endif 91 92 93extern struct protosw inetsw[]; 94 95/* 96 * IP output. The packet in mbuf chain m contains a skeletal IP 97 * header (with len, off, ttl, proto, tos, src, dst). 98 * The mbuf chain containing the packet will be freed. 99 * The mbuf opt, if present, will not be freed. 100 */ 101int 102ip_output(m0, opt, ro, flags, imo) 103 struct mbuf *m0; 104 struct mbuf *opt; 105 struct route *ro; 106 int flags; 107 struct ip_moptions *imo; 108{ 109 struct ip *ip, *mhip; 110 struct ifnet *ifp; 111 struct mbuf *m = m0; 112 int hlen = sizeof (struct ip); 113 int len, off, error = 0; 114 struct sockaddr_in *dst; 115 struct in_ifaddr *ia; 116 int isbroadcast; 117 118#ifdef DIAGNOSTIC 119 if ((m->m_flags & M_PKTHDR) == 0) 120 panic("ip_output no HDR"); 121 if (!ro) 122 panic("ip_output no route, proto = %d", 123 mtod(m, struct ip *)->ip_p); 124#endif 125 if (opt) { 126 m = ip_insertoptions(m, opt, &len); 127 hlen = len; 128 } 129 ip = mtod(m, struct ip *); 130 /* 131 * Fill in IP header. 132 */ 133 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 134 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 135 ip->ip_off &= IP_DF; 136 ip->ip_id = htons(ip_id++); 137 ipstat.ips_localout++; 138 } else { 139 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 140 } 141 142 dst = (struct sockaddr_in *)&ro->ro_dst; 143 /* 144 * If there is a cached route, 145 * check that it is to the same destination 146 * and is still up. If not, free it and try again. 147 */ 148 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 149 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 150 RTFREE(ro->ro_rt); 151 ro->ro_rt = (struct rtentry *)0; 152 } 153 if (ro->ro_rt == 0) { 154 dst->sin_family = AF_INET; 155 dst->sin_len = sizeof(*dst); 156 dst->sin_addr = ip->ip_dst; 157 } 158 /* 159 * If routing to interface only, 160 * short circuit routing lookup. 161 */ 162#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 163#define sintosa(sin) ((struct sockaddr *)(sin)) 164 if (flags & IP_ROUTETOIF) { 165 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 166 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 167 ipstat.ips_noroute++; 168 error = ENETUNREACH; 169 goto bad; 170 } 171 ifp = ia->ia_ifp; 172 ip->ip_ttl = 1; 173 isbroadcast = in_broadcast(dst->sin_addr, ifp); 174 } else { 175 /* 176 * If this is the case, we probably don't want to allocate 177 * a protocol-cloned route since we didn't get one from the 178 * ULP. This lets TCP do its thing, while not burdening 179 * forwarding or ICMP with the overhead of cloning a route. 180 * Of course, we still want to do any cloning requested by 181 * the link layer, as this is probably required in all cases 182 * for correct operation (as it is for ARP). 183 */ 184 if (ro->ro_rt == 0) 185 rtalloc_ign(ro, RTF_PRCLONING); 186 if (ro->ro_rt == 0) { 187 ipstat.ips_noroute++; 188 error = EHOSTUNREACH; 189 goto bad; 190 } 191 ia = ifatoia(ro->ro_rt->rt_ifa); 192 ifp = ro->ro_rt->rt_ifp; 193 ro->ro_rt->rt_use++; 194 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 195 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 196 if (ro->ro_rt->rt_flags & RTF_HOST) 197 isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST); 198 else 199 isbroadcast = in_broadcast(dst->sin_addr, ifp); 200 } 201 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 202 struct in_multi *inm; 203 204 m->m_flags |= M_MCAST; 205 /* 206 * IP destination address is multicast. Make sure "dst" 207 * still points to the address in "ro". (It may have been 208 * changed to point to a gateway address, above.) 209 */ 210 dst = (struct sockaddr_in *)&ro->ro_dst; 211 /* 212 * See if the caller provided any multicast options 213 */ 214 if (imo != NULL) { 215 ip->ip_ttl = imo->imo_multicast_ttl; 216 if (imo->imo_multicast_ifp != NULL) 217 ifp = imo->imo_multicast_ifp; 218 if (imo->imo_multicast_vif != -1) 219 ip->ip_src.s_addr = 220 ip_mcast_src(imo->imo_multicast_vif); 221 } else 222 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 223 /* 224 * Confirm that the outgoing interface supports multicast. 225 */ 226 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 227 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 228 ipstat.ips_noroute++; 229 error = ENETUNREACH; 230 goto bad; 231 } 232 } 233 /* 234 * If source address not specified yet, use address 235 * of outgoing interface. 236 */ 237 if (ip->ip_src.s_addr == INADDR_ANY) { 238 register struct in_ifaddr *ia1; 239 240 for (ia1 = in_ifaddrhead.tqh_first; ia1; 241 ia1 = ia1->ia_link.tqe_next) 242 if (ia1->ia_ifp == ifp) { 243 ip->ip_src = IA_SIN(ia1)->sin_addr; 244 break; 245 } 246 } 247 248 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 249 if (inm != NULL && 250 (imo == NULL || imo->imo_multicast_loop)) { 251 /* 252 * If we belong to the destination multicast group 253 * on the outgoing interface, and the caller did not 254 * forbid loopback, loop back a copy. 255 */ 256 ip_mloopback(ifp, m, dst, hlen); 257 } 258 else { 259 /* 260 * If we are acting as a multicast router, perform 261 * multicast forwarding as if the packet had just 262 * arrived on the interface to which we are about 263 * to send. The multicast forwarding function 264 * recursively calls this function, using the 265 * IP_FORWARDING flag to prevent infinite recursion. 266 * 267 * Multicasts that are looped back by ip_mloopback(), 268 * above, will be forwarded by the ip_input() routine, 269 * if necessary. 270 */ 271 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 272 /* 273 * Check if rsvp daemon is running. If not, don't 274 * set ip_moptions. This ensures that the packet 275 * is multicast and not just sent down one link 276 * as prescribed by rsvpd. 277 */ 278 if (!rsvp_on) 279 imo = NULL; 280 if (ip_mforward(ip, ifp, m, imo) != 0) { 281 m_freem(m); 282 goto done; 283 } 284 } 285 } 286 287 /* 288 * Multicasts with a time-to-live of zero may be looped- 289 * back, above, but must not be transmitted on a network. 290 * Also, multicasts addressed to the loopback interface 291 * are not sent -- the above call to ip_mloopback() will 292 * loop back a copy if this host actually belongs to the 293 * destination group on the loopback interface. 294 */ 295 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 296 m_freem(m); 297 goto done; 298 } 299 300 goto sendit; 301 } 302#ifndef notdef 303 /* 304 * If source address not specified yet, use address 305 * of outgoing interface. 306 */ 307 if (ip->ip_src.s_addr == INADDR_ANY) 308 ip->ip_src = IA_SIN(ia)->sin_addr; 309#endif 310 /* 311 * Verify that we have any chance at all of being able to queue 312 * the packet or packet fragments 313 */ 314 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 315 ifp->if_snd.ifq_maxlen) { 316 error = ENOBUFS; 317 goto bad; 318 } 319 320 /* 321 * Look for broadcast address and 322 * and verify user is allowed to send 323 * such a packet. 324 */ 325 if (isbroadcast) { 326 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 327 error = EADDRNOTAVAIL; 328 goto bad; 329 } 330 if ((flags & IP_ALLOWBROADCAST) == 0) { 331 error = EACCES; 332 goto bad; 333 } 334 /* don't allow broadcast messages to be fragmented */ 335 if ((u_short)ip->ip_len > ifp->if_mtu) { 336 error = EMSGSIZE; 337 goto bad; 338 } 339 m->m_flags |= M_BCAST; 340 } else { 341 m->m_flags &= ~M_BCAST; 342 } 343 344sendit: 345#if defined(IPFILTER) || defined(IPFILTER_LKM) 346 /* 347 * looks like most checking has been done now...do a filter check 348 */ 349 if (fr_checkp) { 350 struct mbuf *m1 = m; 351 352 if ((*fr_checkp)(ip, hlen, ifp, 1, &m1)) 353 error = EHOSTUNREACH; 354 if (error || !m1) 355 goto done; 356 ip = mtod(m = m1, struct ip *); 357 } 358#endif 359 /* 360 * IpHack's section. 361 * - Xlate: translate packet's addr/port (NAT). 362 * - Firewall: deny/allow/etc. 363 * - Wrap: fake packet's addr/port <unimpl.> 364 * - Encapsulate: put it in another IP and send out. <unimp.> 365 */ 366 367#ifdef COMPAT_IPFW 368 if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, ifp, IP_NAT_OUT)) { 369 error = EACCES; 370 goto done; 371 } 372 373 /* 374 * Check with the firewall... 375 */ 376 if (ip_fw_chk_ptr) { 377#ifdef IPDIVERT 378 ip_divert_port = (*ip_fw_chk_ptr)(&ip, 379 hlen, ifp, ip_divert_ignore, &m); 380 ip_divert_ignore = 0; 381 if (ip_divert_port) { /* Divert packet */ 382 (*inetsw[ip_protox[IPPROTO_DIVERT]].pr_input)(m, 0); 383 goto done; 384 } 385#else 386 /* If ipfw says divert, we have to just drop packet */ 387 if ((*ip_fw_chk_ptr)(&ip, hlen, ifp, 0, &m)) { 388 m_freem(m); 389 goto done; 390 } 391#endif 392 if (!m) { 393 error = EACCES; 394 goto done; 395 } 396 } 397#endif /* COMPAT_IPFW */ 398 399 /* 400 * If small enough for interface, can just send directly. 401 */ 402 if ((u_short)ip->ip_len <= ifp->if_mtu) { 403 ip->ip_len = htons((u_short)ip->ip_len); 404 ip->ip_off = htons((u_short)ip->ip_off); 405 ip->ip_sum = 0; 406 if (ip->ip_vhl == IP_VHL_BORING) { 407 ip->ip_sum = in_cksum_hdr(ip); 408 } else { 409 ip->ip_sum = in_cksum(m, hlen); 410 } 411 error = (*ifp->if_output)(ifp, m, 412 (struct sockaddr *)dst, ro->ro_rt); 413 goto done; 414 } 415 /* 416 * Too large for interface; fragment if possible. 417 * Must be able to put at least 8 bytes per fragment. 418 */ 419 if (ip->ip_off & IP_DF) { 420 error = EMSGSIZE; 421 /* 422 * This case can happen if the user changed the MTU 423 * of an interface after enabling IP on it. Because 424 * most netifs don't keep track of routes pointing to 425 * them, there is no way for one to update all its 426 * routes when the MTU is changed. 427 */ 428 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 429 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 430 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 431 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 432 } 433 ipstat.ips_cantfrag++; 434 goto bad; 435 } 436 len = (ifp->if_mtu - hlen) &~ 7; 437 if (len < 8) { 438 error = EMSGSIZE; 439 goto bad; 440 } 441 442 { 443 int mhlen, firstlen = len; 444 struct mbuf **mnext = &m->m_nextpkt; 445 446 /* 447 * Loop through length of segment after first fragment, 448 * make new header and copy data of each part and link onto chain. 449 */ 450 m0 = m; 451 mhlen = sizeof (struct ip); 452 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 453 MGETHDR(m, M_DONTWAIT, MT_HEADER); 454 if (m == 0) { 455 error = ENOBUFS; 456 ipstat.ips_odropped++; 457 goto sendorfree; 458 } 459 m->m_data += max_linkhdr; 460 mhip = mtod(m, struct ip *); 461 *mhip = *ip; 462 if (hlen > sizeof (struct ip)) { 463 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 464 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 465 } 466 m->m_len = mhlen; 467 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 468 if (ip->ip_off & IP_MF) 469 mhip->ip_off |= IP_MF; 470 if (off + len >= (u_short)ip->ip_len) 471 len = (u_short)ip->ip_len - off; 472 else 473 mhip->ip_off |= IP_MF; 474 mhip->ip_len = htons((u_short)(len + mhlen)); 475 m->m_next = m_copy(m0, off, len); 476 if (m->m_next == 0) { 477 (void) m_free(m); 478 error = ENOBUFS; /* ??? */ 479 ipstat.ips_odropped++; 480 goto sendorfree; 481 } 482 m->m_pkthdr.len = mhlen + len; 483 m->m_pkthdr.rcvif = (struct ifnet *)0; 484 mhip->ip_off = htons((u_short)mhip->ip_off); 485 mhip->ip_sum = 0; 486 if (mhip->ip_vhl == IP_VHL_BORING) { 487 mhip->ip_sum = in_cksum_hdr(mhip); 488 } else { 489 mhip->ip_sum = in_cksum(m, mhlen); 490 } 491 *mnext = m; 492 mnext = &m->m_nextpkt; 493 ipstat.ips_ofragments++; 494 } 495 /* 496 * Update first fragment by trimming what's been copied out 497 * and updating header, then send each fragment (in order). 498 */ 499 m = m0; 500 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 501 m->m_pkthdr.len = hlen + firstlen; 502 ip->ip_len = htons((u_short)m->m_pkthdr.len); 503 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 504 ip->ip_sum = 0; 505 if (ip->ip_vhl == IP_VHL_BORING) { 506 ip->ip_sum = in_cksum_hdr(ip); 507 } else { 508 ip->ip_sum = in_cksum(m, hlen); 509 } 510sendorfree: 511 for (m = m0; m; m = m0) { 512 m0 = m->m_nextpkt; 513 m->m_nextpkt = 0; 514 if (error == 0) 515 error = (*ifp->if_output)(ifp, m, 516 (struct sockaddr *)dst, ro->ro_rt); 517 else 518 m_freem(m); 519 } 520 521 if (error == 0) 522 ipstat.ips_fragmented++; 523 } 524done: 525 return (error); 526bad: 527 m_freem(m0); 528 goto done; 529} 530 531/* 532 * Insert IP options into preformed packet. 533 * Adjust IP destination as required for IP source routing, 534 * as indicated by a non-zero in_addr at the start of the options. 535 * 536 * XXX This routine assumes that the packet has no options in place. 537 */ 538static struct mbuf * 539ip_insertoptions(m, opt, phlen) 540 register struct mbuf *m; 541 struct mbuf *opt; 542 int *phlen; 543{ 544 register struct ipoption *p = mtod(opt, struct ipoption *); 545 struct mbuf *n; 546 register struct ip *ip = mtod(m, struct ip *); 547 unsigned optlen; 548 549 optlen = opt->m_len - sizeof(p->ipopt_dst); 550 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 551 return (m); /* XXX should fail */ 552 if (p->ipopt_dst.s_addr) 553 ip->ip_dst = p->ipopt_dst; 554 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 555 MGETHDR(n, M_DONTWAIT, MT_HEADER); 556 if (n == 0) 557 return (m); 558 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 559 m->m_len -= sizeof(struct ip); 560 m->m_data += sizeof(struct ip); 561 n->m_next = m; 562 m = n; 563 m->m_len = optlen + sizeof(struct ip); 564 m->m_data += max_linkhdr; 565 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 566 } else { 567 m->m_data -= optlen; 568 m->m_len += optlen; 569 m->m_pkthdr.len += optlen; 570 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 571 } 572 ip = mtod(m, struct ip *); 573 bcopy(p->ipopt_list, ip + 1, optlen); 574 *phlen = sizeof(struct ip) + optlen; 575 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 576 ip->ip_len += optlen; 577 return (m); 578} 579 580/* 581 * Copy options from ip to jp, 582 * omitting those not copied during fragmentation. 583 */ 584#if !defined(IPFILTER) && !defined(IPFILTER_LKM) 585static 586#endif 587int 588ip_optcopy(ip, jp) 589 struct ip *ip, *jp; 590{ 591 register u_char *cp, *dp; 592 int opt, optlen, cnt; 593 594 cp = (u_char *)(ip + 1); 595 dp = (u_char *)(jp + 1); 596 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 597 for (; cnt > 0; cnt -= optlen, cp += optlen) { 598 opt = cp[0]; 599 if (opt == IPOPT_EOL) 600 break; 601 if (opt == IPOPT_NOP) { 602 /* Preserve for IP mcast tunnel's LSRR alignment. */ 603 *dp++ = IPOPT_NOP; 604 optlen = 1; 605 continue; 606 } else 607 optlen = cp[IPOPT_OLEN]; 608 /* bogus lengths should have been caught by ip_dooptions */ 609 if (optlen > cnt) 610 optlen = cnt; 611 if (IPOPT_COPIED(opt)) { 612 bcopy(cp, dp, optlen); 613 dp += optlen; 614 } 615 } 616 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 617 *dp++ = IPOPT_EOL; 618 return (optlen); 619} 620 621/* 622 * IP socket option processing. 623 */ 624int 625ip_ctloutput(op, so, level, optname, mp, p) 626 int op; 627 struct socket *so; 628 int level, optname; 629 struct mbuf **mp; 630 struct proc *p; 631{ 632 register struct inpcb *inp = sotoinpcb(so); 633 register struct mbuf *m = *mp; 634 register int optval = 0; 635 int error = 0; 636 637 if (level != IPPROTO_IP) { 638 error = EINVAL; 639 if (op == PRCO_SETOPT && *mp) 640 (void) m_free(*mp); 641 } else switch (op) { 642 643 case PRCO_SETOPT: 644 switch (optname) { 645 case IP_OPTIONS: 646#ifdef notyet 647 case IP_RETOPTS: 648 return (ip_pcbopts(optname, &inp->inp_options, m)); 649#else 650 return (ip_pcbopts(&inp->inp_options, m)); 651#endif 652 653 case IP_TOS: 654 case IP_TTL: 655 case IP_RECVOPTS: 656 case IP_RECVRETOPTS: 657 case IP_RECVDSTADDR: 658 case IP_RECVIF: 659 if (m == 0 || m->m_len != sizeof(int)) 660 error = EINVAL; 661 else { 662 optval = *mtod(m, int *); 663 switch (optname) { 664 665 case IP_TOS: 666 inp->inp_ip_tos = optval; 667 break; 668 669 case IP_TTL: 670 inp->inp_ip_ttl = optval; 671 break; 672#define OPTSET(bit) \ 673 if (optval) \ 674 inp->inp_flags |= bit; \ 675 else \ 676 inp->inp_flags &= ~bit; 677 678 case IP_RECVOPTS: 679 OPTSET(INP_RECVOPTS); 680 break; 681 682 case IP_RECVRETOPTS: 683 OPTSET(INP_RECVRETOPTS); 684 break; 685 686 case IP_RECVDSTADDR: 687 OPTSET(INP_RECVDSTADDR); 688 break; 689 690 case IP_RECVIF: 691 OPTSET(INP_RECVIF); 692 break; 693 } 694 } 695 break; 696#undef OPTSET 697 698 case IP_MULTICAST_IF: 699 case IP_MULTICAST_VIF: 700 case IP_MULTICAST_TTL: 701 case IP_MULTICAST_LOOP: 702 case IP_ADD_MEMBERSHIP: 703 case IP_DROP_MEMBERSHIP: 704 error = ip_setmoptions(optname, &inp->inp_moptions, m); 705 break; 706 707 case IP_PORTRANGE: 708 if (m == 0 || m->m_len != sizeof(int)) 709 error = EINVAL; 710 else { 711 optval = *mtod(m, int *); 712 713 switch (optval) { 714 715 case IP_PORTRANGE_DEFAULT: 716 inp->inp_flags &= ~(INP_LOWPORT); 717 inp->inp_flags &= ~(INP_HIGHPORT); 718 break; 719 720 case IP_PORTRANGE_HIGH: 721 inp->inp_flags &= ~(INP_LOWPORT); 722 inp->inp_flags |= INP_HIGHPORT; 723 break; 724 725 case IP_PORTRANGE_LOW: 726 inp->inp_flags &= ~(INP_HIGHPORT); 727 inp->inp_flags |= INP_LOWPORT; 728 break; 729 730 default: 731 error = EINVAL; 732 break; 733 } 734 } 735 break; 736 737 default: 738 error = ENOPROTOOPT; 739 break; 740 } 741 if (m) 742 (void)m_free(m); 743 break; 744 745 case PRCO_GETOPT: 746 switch (optname) { 747 case IP_OPTIONS: 748 case IP_RETOPTS: 749 *mp = m = m_get(M_WAIT, MT_SOOPTS); 750 if (inp->inp_options) { 751 m->m_len = inp->inp_options->m_len; 752 bcopy(mtod(inp->inp_options, void *), 753 mtod(m, void *), m->m_len); 754 } else 755 m->m_len = 0; 756 break; 757 758 case IP_TOS: 759 case IP_TTL: 760 case IP_RECVOPTS: 761 case IP_RECVRETOPTS: 762 case IP_RECVDSTADDR: 763 case IP_RECVIF: 764 *mp = m = m_get(M_WAIT, MT_SOOPTS); 765 m->m_len = sizeof(int); 766 switch (optname) { 767 768 case IP_TOS: 769 optval = inp->inp_ip_tos; 770 break; 771 772 case IP_TTL: 773 optval = inp->inp_ip_ttl; 774 break; 775 776#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 777 778 case IP_RECVOPTS: 779 optval = OPTBIT(INP_RECVOPTS); 780 break; 781 782 case IP_RECVRETOPTS: 783 optval = OPTBIT(INP_RECVRETOPTS); 784 break; 785 786 case IP_RECVDSTADDR: 787 optval = OPTBIT(INP_RECVDSTADDR); 788 break; 789 790 case IP_RECVIF: 791 optval = OPTBIT(INP_RECVIF); 792 break; 793 } 794 *mtod(m, int *) = optval; 795 break; 796 797 case IP_MULTICAST_IF: 798 case IP_MULTICAST_VIF: 799 case IP_MULTICAST_TTL: 800 case IP_MULTICAST_LOOP: 801 case IP_ADD_MEMBERSHIP: 802 case IP_DROP_MEMBERSHIP: 803 error = ip_getmoptions(optname, inp->inp_moptions, mp); 804 break; 805 806 case IP_PORTRANGE: 807 *mp = m = m_get(M_WAIT, MT_SOOPTS); 808 m->m_len = sizeof(int); 809 810 if (inp->inp_flags & INP_HIGHPORT) 811 optval = IP_PORTRANGE_HIGH; 812 else if (inp->inp_flags & INP_LOWPORT) 813 optval = IP_PORTRANGE_LOW; 814 else 815 optval = 0; 816 817 *mtod(m, int *) = optval; 818 break; 819 820 default: 821 error = ENOPROTOOPT; 822 break; 823 } 824 break; 825 } 826 return (error); 827} 828 829/* 830 * Set up IP options in pcb for insertion in output packets. 831 * Store in mbuf with pointer in pcbopt, adding pseudo-option 832 * with destination address if source routed. 833 */ 834static int 835#ifdef notyet 836ip_pcbopts(optname, pcbopt, m) 837 int optname; 838#else 839ip_pcbopts(pcbopt, m) 840#endif 841 struct mbuf **pcbopt; 842 register struct mbuf *m; 843{ 844 register cnt, optlen; 845 register u_char *cp; 846 u_char opt; 847 848 /* turn off any old options */ 849 if (*pcbopt) 850 (void)m_free(*pcbopt); 851 *pcbopt = 0; 852 if (m == (struct mbuf *)0 || m->m_len == 0) { 853 /* 854 * Only turning off any previous options. 855 */ 856 if (m) 857 (void)m_free(m); 858 return (0); 859 } 860 861#ifndef vax 862 if (m->m_len % sizeof(long)) 863 goto bad; 864#endif 865 /* 866 * IP first-hop destination address will be stored before 867 * actual options; move other options back 868 * and clear it when none present. 869 */ 870 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 871 goto bad; 872 cnt = m->m_len; 873 m->m_len += sizeof(struct in_addr); 874 cp = mtod(m, u_char *) + sizeof(struct in_addr); 875 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 876 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 877 878 for (; cnt > 0; cnt -= optlen, cp += optlen) { 879 opt = cp[IPOPT_OPTVAL]; 880 if (opt == IPOPT_EOL) 881 break; 882 if (opt == IPOPT_NOP) 883 optlen = 1; 884 else { 885 optlen = cp[IPOPT_OLEN]; 886 if (optlen <= IPOPT_OLEN || optlen > cnt) 887 goto bad; 888 } 889 switch (opt) { 890 891 default: 892 break; 893 894 case IPOPT_LSRR: 895 case IPOPT_SSRR: 896 /* 897 * user process specifies route as: 898 * ->A->B->C->D 899 * D must be our final destination (but we can't 900 * check that since we may not have connected yet). 901 * A is first hop destination, which doesn't appear in 902 * actual IP option, but is stored before the options. 903 */ 904 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 905 goto bad; 906 m->m_len -= sizeof(struct in_addr); 907 cnt -= sizeof(struct in_addr); 908 optlen -= sizeof(struct in_addr); 909 cp[IPOPT_OLEN] = optlen; 910 /* 911 * Move first hop before start of options. 912 */ 913 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 914 sizeof(struct in_addr)); 915 /* 916 * Then copy rest of options back 917 * to close up the deleted entry. 918 */ 919 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 920 sizeof(struct in_addr)), 921 (caddr_t)&cp[IPOPT_OFFSET+1], 922 (unsigned)cnt + sizeof(struct in_addr)); 923 break; 924 } 925 } 926 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 927 goto bad; 928 *pcbopt = m; 929 return (0); 930 931bad: 932 (void)m_free(m); 933 return (EINVAL); 934} 935 936/* 937 * Set the IP multicast options in response to user setsockopt(). 938 */ 939static int 940ip_setmoptions(optname, imop, m) 941 int optname; 942 struct ip_moptions **imop; 943 struct mbuf *m; 944{ 945 register int error = 0; 946 u_char loop; 947 register int i; 948 struct in_addr addr; 949 register struct ip_mreq *mreq; 950 register struct ifnet *ifp; 951 register struct ip_moptions *imo = *imop; 952 struct route ro; 953 register struct sockaddr_in *dst; 954 int s; 955 956 if (imo == NULL) { 957 /* 958 * No multicast option buffer attached to the pcb; 959 * allocate one and initialize to default values. 960 */ 961 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 962 M_WAITOK); 963 964 if (imo == NULL) 965 return (ENOBUFS); 966 *imop = imo; 967 imo->imo_multicast_ifp = NULL; 968 imo->imo_multicast_vif = -1; 969 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 970 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 971 imo->imo_num_memberships = 0; 972 } 973 974 switch (optname) { 975 /* store an index number for the vif you wanna use in the send */ 976 case IP_MULTICAST_VIF: 977 if (!legal_vif_num) { 978 error = EOPNOTSUPP; 979 break; 980 } 981 if (m == NULL || m->m_len != sizeof(int)) { 982 error = EINVAL; 983 break; 984 } 985 i = *(mtod(m, int *)); 986 if (!legal_vif_num(i) && (i != -1)) { 987 error = EINVAL; 988 break; 989 } 990 imo->imo_multicast_vif = i; 991 break; 992 993 case IP_MULTICAST_IF: 994 /* 995 * Select the interface for outgoing multicast packets. 996 */ 997 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 998 error = EINVAL; 999 break; 1000 } 1001 addr = *(mtod(m, struct in_addr *)); 1002 /* 1003 * INADDR_ANY is used to remove a previous selection. 1004 * When no interface is selected, a default one is 1005 * chosen every time a multicast packet is sent. 1006 */ 1007 if (addr.s_addr == INADDR_ANY) { 1008 imo->imo_multicast_ifp = NULL; 1009 break; 1010 } 1011 /* 1012 * The selected interface is identified by its local 1013 * IP address. Find the interface and confirm that 1014 * it supports multicasting. 1015 */ 1016 s = splimp(); 1017 INADDR_TO_IFP(addr, ifp); 1018 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1019 splx(s); 1020 error = EADDRNOTAVAIL; 1021 break; 1022 } 1023 imo->imo_multicast_ifp = ifp; 1024 splx(s); 1025 break; 1026 1027 case IP_MULTICAST_TTL: 1028 /* 1029 * Set the IP time-to-live for outgoing multicast packets. 1030 */ 1031 if (m == NULL || m->m_len != 1) { 1032 error = EINVAL; 1033 break; 1034 } 1035 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 1036 break; 1037 1038 case IP_MULTICAST_LOOP: 1039 /* 1040 * Set the loopback flag for outgoing multicast packets. 1041 * Must be zero or one. 1042 */ 1043 if (m == NULL || m->m_len != 1 || 1044 (loop = *(mtod(m, u_char *))) > 1) { 1045 error = EINVAL; 1046 break; 1047 } 1048 imo->imo_multicast_loop = loop; 1049 break; 1050 1051 case IP_ADD_MEMBERSHIP: 1052 /* 1053 * Add a multicast group membership. 1054 * Group must be a valid IP multicast address. 1055 */ 1056 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1057 error = EINVAL; 1058 break; 1059 } 1060 mreq = mtod(m, struct ip_mreq *); 1061 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1062 error = EINVAL; 1063 break; 1064 } 1065 s = splimp(); 1066 /* 1067 * If no interface address was provided, use the interface of 1068 * the route to the given multicast address. 1069 */ 1070 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1071 bzero((caddr_t)&ro, sizeof(ro)); 1072 dst = (struct sockaddr_in *)&ro.ro_dst; 1073 dst->sin_len = sizeof(*dst); 1074 dst->sin_family = AF_INET; 1075 dst->sin_addr = mreq->imr_multiaddr; 1076 rtalloc(&ro); 1077 if (ro.ro_rt == NULL) { 1078 error = EADDRNOTAVAIL; 1079 splx(s); 1080 break; 1081 } 1082 ifp = ro.ro_rt->rt_ifp; 1083 rtfree(ro.ro_rt); 1084 } 1085 else { 1086 INADDR_TO_IFP(mreq->imr_interface, ifp); 1087 } 1088 1089 /* 1090 * See if we found an interface, and confirm that it 1091 * supports multicast. 1092 */ 1093 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1094 error = EADDRNOTAVAIL; 1095 splx(s); 1096 break; 1097 } 1098 /* 1099 * See if the membership already exists or if all the 1100 * membership slots are full. 1101 */ 1102 for (i = 0; i < imo->imo_num_memberships; ++i) { 1103 if (imo->imo_membership[i]->inm_ifp == ifp && 1104 imo->imo_membership[i]->inm_addr.s_addr 1105 == mreq->imr_multiaddr.s_addr) 1106 break; 1107 } 1108 if (i < imo->imo_num_memberships) { 1109 error = EADDRINUSE; 1110 splx(s); 1111 break; 1112 } 1113 if (i == IP_MAX_MEMBERSHIPS) { 1114 error = ETOOMANYREFS; 1115 splx(s); 1116 break; 1117 } 1118 /* 1119 * Everything looks good; add a new record to the multicast 1120 * address list for the given interface. 1121 */ 1122 if ((imo->imo_membership[i] = 1123 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1124 error = ENOBUFS; 1125 splx(s); 1126 break; 1127 } 1128 ++imo->imo_num_memberships; 1129 splx(s); 1130 break; 1131 1132 case IP_DROP_MEMBERSHIP: 1133 /* 1134 * Drop a multicast group membership. 1135 * Group must be a valid IP multicast address. 1136 */ 1137 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1138 error = EINVAL; 1139 break; 1140 } 1141 mreq = mtod(m, struct ip_mreq *); 1142 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1143 error = EINVAL; 1144 break; 1145 } 1146 1147 s = splimp(); 1148 /* 1149 * If an interface address was specified, get a pointer 1150 * to its ifnet structure. 1151 */ 1152 if (mreq->imr_interface.s_addr == INADDR_ANY) 1153 ifp = NULL; 1154 else { 1155 INADDR_TO_IFP(mreq->imr_interface, ifp); 1156 if (ifp == NULL) { 1157 error = EADDRNOTAVAIL; 1158 splx(s); 1159 break; 1160 } 1161 } 1162 /* 1163 * Find the membership in the membership array. 1164 */ 1165 for (i = 0; i < imo->imo_num_memberships; ++i) { 1166 if ((ifp == NULL || 1167 imo->imo_membership[i]->inm_ifp == ifp) && 1168 imo->imo_membership[i]->inm_addr.s_addr == 1169 mreq->imr_multiaddr.s_addr) 1170 break; 1171 } 1172 if (i == imo->imo_num_memberships) { 1173 error = EADDRNOTAVAIL; 1174 splx(s); 1175 break; 1176 } 1177 /* 1178 * Give up the multicast address record to which the 1179 * membership points. 1180 */ 1181 in_delmulti(imo->imo_membership[i]); 1182 /* 1183 * Remove the gap in the membership array. 1184 */ 1185 for (++i; i < imo->imo_num_memberships; ++i) 1186 imo->imo_membership[i-1] = imo->imo_membership[i]; 1187 --imo->imo_num_memberships; 1188 splx(s); 1189 break; 1190 1191 default: 1192 error = EOPNOTSUPP; 1193 break; 1194 } 1195 1196 /* 1197 * If all options have default values, no need to keep the mbuf. 1198 */ 1199 if (imo->imo_multicast_ifp == NULL && 1200 imo->imo_multicast_vif == -1 && 1201 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1202 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1203 imo->imo_num_memberships == 0) { 1204 free(*imop, M_IPMOPTS); 1205 *imop = NULL; 1206 } 1207 1208 return (error); 1209} 1210 1211/* 1212 * Return the IP multicast options in response to user getsockopt(). 1213 */ 1214static int 1215ip_getmoptions(optname, imo, mp) 1216 int optname; 1217 register struct ip_moptions *imo; 1218 register struct mbuf **mp; 1219{ 1220 u_char *ttl; 1221 u_char *loop; 1222 struct in_addr *addr; 1223 struct in_ifaddr *ia; 1224 1225 *mp = m_get(M_WAIT, MT_SOOPTS); 1226 1227 switch (optname) { 1228 1229 case IP_MULTICAST_VIF: 1230 if (imo != NULL) 1231 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1232 else 1233 *(mtod(*mp, int *)) = -1; 1234 (*mp)->m_len = sizeof(int); 1235 return(0); 1236 1237 case IP_MULTICAST_IF: 1238 addr = mtod(*mp, struct in_addr *); 1239 (*mp)->m_len = sizeof(struct in_addr); 1240 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1241 addr->s_addr = INADDR_ANY; 1242 else { 1243 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1244 addr->s_addr = (ia == NULL) ? INADDR_ANY 1245 : IA_SIN(ia)->sin_addr.s_addr; 1246 } 1247 return (0); 1248 1249 case IP_MULTICAST_TTL: 1250 ttl = mtod(*mp, u_char *); 1251 (*mp)->m_len = 1; 1252 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1253 : imo->imo_multicast_ttl; 1254 return (0); 1255 1256 case IP_MULTICAST_LOOP: 1257 loop = mtod(*mp, u_char *); 1258 (*mp)->m_len = 1; 1259 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1260 : imo->imo_multicast_loop; 1261 return (0); 1262 1263 default: 1264 return (EOPNOTSUPP); 1265 } 1266} 1267 1268/* 1269 * Discard the IP multicast options. 1270 */ 1271void 1272ip_freemoptions(imo) 1273 register struct ip_moptions *imo; 1274{ 1275 register int i; 1276 1277 if (imo != NULL) { 1278 for (i = 0; i < imo->imo_num_memberships; ++i) 1279 in_delmulti(imo->imo_membership[i]); 1280 free(imo, M_IPMOPTS); 1281 } 1282} 1283 1284/* 1285 * Routine called from ip_output() to loop back a copy of an IP multicast 1286 * packet to the input queue of a specified interface. Note that this 1287 * calls the output routine of the loopback "driver", but with an interface 1288 * pointer that might NOT be a loopback interface -- evil, but easier than 1289 * replicating that code here. 1290 */ 1291static void 1292ip_mloopback(ifp, m, dst, hlen) 1293 struct ifnet *ifp; 1294 register struct mbuf *m; 1295 register struct sockaddr_in *dst; 1296 int hlen; 1297{ 1298 register struct ip *ip; 1299 struct mbuf *copym; 1300 1301 copym = m_copy(m, 0, M_COPYALL); 1302 if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen)) 1303 copym = m_pullup(copym, hlen); 1304 if (copym != NULL) { 1305 /* 1306 * We don't bother to fragment if the IP length is greater 1307 * than the interface's MTU. Can this possibly matter? 1308 */ 1309 ip = mtod(copym, struct ip *); 1310 ip->ip_len = htons((u_short)ip->ip_len); 1311 ip->ip_off = htons((u_short)ip->ip_off); 1312 ip->ip_sum = 0; 1313 if (ip->ip_vhl == IP_VHL_BORING) { 1314 ip->ip_sum = in_cksum_hdr(ip); 1315 } else { 1316 ip->ip_sum = in_cksum(copym, hlen); 1317 } 1318 /* 1319 * NB: 1320 * It's not clear whether there are any lingering 1321 * reentrancy problems in other areas which might 1322 * be exposed by using ip_input directly (in 1323 * particular, everything which modifies the packet 1324 * in-place). Yet another option is using the 1325 * protosw directly to deliver the looped back 1326 * packet. For the moment, we'll err on the side 1327 * of safety by continuing to abuse looutput(). 1328 */ 1329#ifdef notdef 1330 copym->m_pkthdr.rcvif = ifp; 1331 ip_input(copym) 1332#else 1333 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1334#endif 1335 } 1336} 1337