ip_output.c revision 15869
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 1.38 1996/05/21 20:47:31 peter Exp $ 35 */ 36 37#define _IP_VHL 38 39#include <sys/param.h> 40#include <sys/queue.h> 41#include <sys/systm.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/errno.h> 45#include <sys/protosw.h> 46#include <sys/socket.h> 47#include <sys/socketvar.h> 48 49#include <net/if.h> 50#include <net/route.h> 51 52#include <netinet/in.h> 53#include <netinet/in_systm.h> 54#include <netinet/ip.h> 55#include <netinet/in_pcb.h> 56#include <netinet/in_var.h> 57#include <netinet/ip_var.h> 58 59#ifdef vax 60#include <machine/mtpr.h> 61#endif 62#include <machine/in_cksum.h> 63 64#if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1 65#undef COMPAT_IPFW 66#define COMPAT_IPFW 1 67#else 68#undef COMPAT_IPFW 69#endif 70 71u_short ip_id; 72 73static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *)); 74static void ip_mloopback 75 __P((struct ifnet *, struct mbuf *, struct sockaddr_in *)); 76static int ip_getmoptions 77 __P((int, struct ip_moptions *, struct mbuf **)); 78static int ip_optcopy __P((struct ip *, struct ip *)); 79static int ip_pcbopts __P((struct mbuf **, struct mbuf *)); 80static int ip_setmoptions 81 __P((int, struct ip_moptions **, struct mbuf *)); 82 83/* 84 * IP output. The packet in mbuf chain m contains a skeletal IP 85 * header (with len, off, ttl, proto, tos, src, dst). 86 * The mbuf chain containing the packet will be freed. 87 * The mbuf opt, if present, will not be freed. 88 */ 89int 90ip_output(m0, opt, ro, flags, imo) 91 struct mbuf *m0; 92 struct mbuf *opt; 93 struct route *ro; 94 int flags; 95 struct ip_moptions *imo; 96{ 97 struct ip *ip, *mhip; 98 struct ifnet *ifp; 99 struct mbuf *m = m0; 100 int hlen = sizeof (struct ip); 101 int len, off, error = 0; 102 struct sockaddr_in *dst; 103 struct in_ifaddr *ia; 104 int isbroadcast; 105 106#ifdef DIAGNOSTIC 107 if ((m->m_flags & M_PKTHDR) == 0) 108 panic("ip_output no HDR"); 109 if (!ro) 110 panic("ip_output no route, proto = %d", 111 mtod(m, struct ip *)->ip_p); 112#endif 113 if (opt) { 114 m = ip_insertoptions(m, opt, &len); 115 hlen = len; 116 } 117 ip = mtod(m, struct ip *); 118 /* 119 * Fill in IP header. 120 */ 121 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 122 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2); 123 ip->ip_off &= IP_DF; 124 ip->ip_id = htons(ip_id++); 125 ipstat.ips_localout++; 126 } else { 127 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 128 } 129 130 dst = (struct sockaddr_in *)&ro->ro_dst; 131 /* 132 * If there is a cached route, 133 * check that it is to the same destination 134 * and is still up. If not, free it and try again. 135 */ 136 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 137 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 138 RTFREE(ro->ro_rt); 139 ro->ro_rt = (struct rtentry *)0; 140 } 141 if (ro->ro_rt == 0) { 142 dst->sin_family = AF_INET; 143 dst->sin_len = sizeof(*dst); 144 dst->sin_addr = ip->ip_dst; 145 } 146 /* 147 * If routing to interface only, 148 * short circuit routing lookup. 149 */ 150#define ifatoia(ifa) ((struct in_ifaddr *)(ifa)) 151#define sintosa(sin) ((struct sockaddr *)(sin)) 152 if (flags & IP_ROUTETOIF) { 153 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 && 154 (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) { 155 ipstat.ips_noroute++; 156 error = ENETUNREACH; 157 goto bad; 158 } 159 ifp = ia->ia_ifp; 160 ip->ip_ttl = 1; 161 isbroadcast = in_broadcast(dst->sin_addr, ifp); 162 } else { 163 /* 164 * If this is the case, we probably don't want to allocate 165 * a protocol-cloned route since we didn't get one from the 166 * ULP. This lets TCP do its thing, while not burdening 167 * forwarding or ICMP with the overhead of cloning a route. 168 * Of course, we still want to do any cloning requested by 169 * the link layer, as this is probably required in all cases 170 * for correct operation (as it is for ARP). 171 */ 172 if (ro->ro_rt == 0) 173 rtalloc_ign(ro, RTF_PRCLONING); 174 if (ro->ro_rt == 0) { 175 ipstat.ips_noroute++; 176 error = EHOSTUNREACH; 177 goto bad; 178 } 179 ia = ifatoia(ro->ro_rt->rt_ifa); 180 ifp = ro->ro_rt->rt_ifp; 181 ro->ro_rt->rt_use++; 182 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 183 dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway; 184 if (ro->ro_rt->rt_flags & RTF_HOST) 185 isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST); 186 else 187 isbroadcast = in_broadcast(dst->sin_addr, ifp); 188 } 189 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 190 struct in_multi *inm; 191 192 m->m_flags |= M_MCAST; 193 /* 194 * IP destination address is multicast. Make sure "dst" 195 * still points to the address in "ro". (It may have been 196 * changed to point to a gateway address, above.) 197 */ 198 dst = (struct sockaddr_in *)&ro->ro_dst; 199 /* 200 * See if the caller provided any multicast options 201 */ 202 if (imo != NULL) { 203 ip->ip_ttl = imo->imo_multicast_ttl; 204 if (imo->imo_multicast_ifp != NULL) 205 ifp = imo->imo_multicast_ifp; 206 if (imo->imo_multicast_vif != -1) 207 ip->ip_src.s_addr = 208 ip_mcast_src(imo->imo_multicast_vif); 209 } else 210 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 211 /* 212 * Confirm that the outgoing interface supports multicast. 213 */ 214 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 215 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 216 ipstat.ips_noroute++; 217 error = ENETUNREACH; 218 goto bad; 219 } 220 } 221 /* 222 * If source address not specified yet, use address 223 * of outgoing interface. 224 */ 225 if (ip->ip_src.s_addr == INADDR_ANY) { 226 register struct in_ifaddr *ia; 227 228 for (ia = in_ifaddr; ia; ia = ia->ia_next) 229 if (ia->ia_ifp == ifp) { 230 ip->ip_src = IA_SIN(ia)->sin_addr; 231 break; 232 } 233 } 234 235 IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm); 236 if (inm != NULL && 237 (imo == NULL || imo->imo_multicast_loop)) { 238 /* 239 * If we belong to the destination multicast group 240 * on the outgoing interface, and the caller did not 241 * forbid loopback, loop back a copy. 242 */ 243 ip_mloopback(ifp, m, dst); 244 } 245 else { 246 /* 247 * If we are acting as a multicast router, perform 248 * multicast forwarding as if the packet had just 249 * arrived on the interface to which we are about 250 * to send. The multicast forwarding function 251 * recursively calls this function, using the 252 * IP_FORWARDING flag to prevent infinite recursion. 253 * 254 * Multicasts that are looped back by ip_mloopback(), 255 * above, will be forwarded by the ip_input() routine, 256 * if necessary. 257 */ 258 if (ip_mrouter && (flags & IP_FORWARDING) == 0) { 259 /* 260 * Check if rsvp daemon is running. If not, don't 261 * set ip_moptions. This ensures that the packet 262 * is multicast and not just sent down one link 263 * as prescribed by rsvpd. 264 */ 265 if (!rsvp_on) 266 imo = NULL; 267 if (ip_mforward(ip, ifp, m, imo) != 0) { 268 m_freem(m); 269 goto done; 270 } 271 } 272 } 273 274 /* 275 * Multicasts with a time-to-live of zero may be looped- 276 * back, above, but must not be transmitted on a network. 277 * Also, multicasts addressed to the loopback interface 278 * are not sent -- the above call to ip_mloopback() will 279 * loop back a copy if this host actually belongs to the 280 * destination group on the loopback interface. 281 */ 282 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 283 m_freem(m); 284 goto done; 285 } 286 287 goto sendit; 288 } 289#ifndef notdef 290 /* 291 * If source address not specified yet, use address 292 * of outgoing interface. 293 */ 294 if (ip->ip_src.s_addr == INADDR_ANY) 295 ip->ip_src = IA_SIN(ia)->sin_addr; 296#endif 297 /* 298 * Verify that we have any chance at all of being able to queue 299 * the packet or packet fragments 300 */ 301 if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >= 302 ifp->if_snd.ifq_maxlen) { 303 error = ENOBUFS; 304 goto bad; 305 } 306 307 /* 308 * Look for broadcast address and 309 * and verify user is allowed to send 310 * such a packet. 311 */ 312 if (isbroadcast) { 313 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 314 error = EADDRNOTAVAIL; 315 goto bad; 316 } 317 if ((flags & IP_ALLOWBROADCAST) == 0) { 318 error = EACCES; 319 goto bad; 320 } 321 /* don't allow broadcast messages to be fragmented */ 322 if ((u_short)ip->ip_len > ifp->if_mtu) { 323 error = EMSGSIZE; 324 goto bad; 325 } 326 m->m_flags |= M_BCAST; 327 } else { 328 m->m_flags &= ~M_BCAST; 329 } 330 331sendit: 332 /* 333 * Check with the firewall... 334 */ 335#ifdef COMPAT_IPFW 336 if (ip_fw_chk_ptr && !(*ip_fw_chk_ptr)(&ip, hlen, ifp, 1, &m)) { 337 error = EACCES; 338 goto done; 339 } 340#endif 341 342 /* 343 * If small enough for interface, can just send directly. 344 */ 345 if ((u_short)ip->ip_len <= ifp->if_mtu) { 346 ip->ip_len = htons((u_short)ip->ip_len); 347 ip->ip_off = htons((u_short)ip->ip_off); 348 ip->ip_sum = 0; 349 if (ip->ip_vhl == IP_VHL_BORING) { 350 ip->ip_sum = in_cksum_hdr(ip); 351 } else { 352 ip->ip_sum = in_cksum(m, hlen); 353 } 354 error = (*ifp->if_output)(ifp, m, 355 (struct sockaddr *)dst, ro->ro_rt); 356 goto done; 357 } 358 /* 359 * Too large for interface; fragment if possible. 360 * Must be able to put at least 8 bytes per fragment. 361 */ 362 if (ip->ip_off & IP_DF) { 363 error = EMSGSIZE; 364 /* 365 * This case can happen if the user changed the MTU 366 * of an interface after enabling IP on it. Because 367 * most netifs don't keep track of routes pointing to 368 * them, there is no way for one to update all its 369 * routes when the MTU is changed. 370 */ 371 if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) 372 && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) 373 && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) { 374 ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu; 375 } 376 ipstat.ips_cantfrag++; 377 goto bad; 378 } 379 len = (ifp->if_mtu - hlen) &~ 7; 380 if (len < 8) { 381 error = EMSGSIZE; 382 goto bad; 383 } 384 385 { 386 int mhlen, firstlen = len; 387 struct mbuf **mnext = &m->m_nextpkt; 388 389 /* 390 * Loop through length of segment after first fragment, 391 * make new header and copy data of each part and link onto chain. 392 */ 393 m0 = m; 394 mhlen = sizeof (struct ip); 395 for (off = hlen + len; off < (u_short)ip->ip_len; off += len) { 396 MGETHDR(m, M_DONTWAIT, MT_HEADER); 397 if (m == 0) { 398 error = ENOBUFS; 399 ipstat.ips_odropped++; 400 goto sendorfree; 401 } 402 m->m_data += max_linkhdr; 403 mhip = mtod(m, struct ip *); 404 *mhip = *ip; 405 if (hlen > sizeof (struct ip)) { 406 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 407 mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2); 408 } 409 m->m_len = mhlen; 410 mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF); 411 if (ip->ip_off & IP_MF) 412 mhip->ip_off |= IP_MF; 413 if (off + len >= (u_short)ip->ip_len) 414 len = (u_short)ip->ip_len - off; 415 else 416 mhip->ip_off |= IP_MF; 417 mhip->ip_len = htons((u_short)(len + mhlen)); 418 m->m_next = m_copy(m0, off, len); 419 if (m->m_next == 0) { 420 (void) m_free(m); 421 error = ENOBUFS; /* ??? */ 422 ipstat.ips_odropped++; 423 goto sendorfree; 424 } 425 m->m_pkthdr.len = mhlen + len; 426 m->m_pkthdr.rcvif = (struct ifnet *)0; 427 mhip->ip_off = htons((u_short)mhip->ip_off); 428 mhip->ip_sum = 0; 429 if (mhip->ip_vhl == IP_VHL_BORING) { 430 mhip->ip_sum = in_cksum_hdr(mhip); 431 } else { 432 mhip->ip_sum = in_cksum(m, mhlen); 433 } 434 *mnext = m; 435 mnext = &m->m_nextpkt; 436 ipstat.ips_ofragments++; 437 } 438 /* 439 * Update first fragment by trimming what's been copied out 440 * and updating header, then send each fragment (in order). 441 */ 442 m = m0; 443 m_adj(m, hlen + firstlen - (u_short)ip->ip_len); 444 m->m_pkthdr.len = hlen + firstlen; 445 ip->ip_len = htons((u_short)m->m_pkthdr.len); 446 ip->ip_off = htons((u_short)(ip->ip_off | IP_MF)); 447 ip->ip_sum = 0; 448 if (ip->ip_vhl == IP_VHL_BORING) { 449 ip->ip_sum = in_cksum_hdr(ip); 450 } else { 451 ip->ip_sum = in_cksum(m, hlen); 452 } 453sendorfree: 454 for (m = m0; m; m = m0) { 455 m0 = m->m_nextpkt; 456 m->m_nextpkt = 0; 457 if (error == 0) 458 error = (*ifp->if_output)(ifp, m, 459 (struct sockaddr *)dst, ro->ro_rt); 460 else 461 m_freem(m); 462 } 463 464 if (error == 0) 465 ipstat.ips_fragmented++; 466 } 467done: 468 return (error); 469bad: 470 m_freem(m0); 471 goto done; 472} 473 474/* 475 * Insert IP options into preformed packet. 476 * Adjust IP destination as required for IP source routing, 477 * as indicated by a non-zero in_addr at the start of the options. 478 * 479 * XXX This routine assumes that the packet has no options in place. 480 */ 481static struct mbuf * 482ip_insertoptions(m, opt, phlen) 483 register struct mbuf *m; 484 struct mbuf *opt; 485 int *phlen; 486{ 487 register struct ipoption *p = mtod(opt, struct ipoption *); 488 struct mbuf *n; 489 register struct ip *ip = mtod(m, struct ip *); 490 unsigned optlen; 491 492 optlen = opt->m_len - sizeof(p->ipopt_dst); 493 if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) 494 return (m); /* XXX should fail */ 495 if (p->ipopt_dst.s_addr) 496 ip->ip_dst = p->ipopt_dst; 497 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) { 498 MGETHDR(n, M_DONTWAIT, MT_HEADER); 499 if (n == 0) 500 return (m); 501 n->m_pkthdr.len = m->m_pkthdr.len + optlen; 502 m->m_len -= sizeof(struct ip); 503 m->m_data += sizeof(struct ip); 504 n->m_next = m; 505 m = n; 506 m->m_len = optlen + sizeof(struct ip); 507 m->m_data += max_linkhdr; 508 (void)memcpy(mtod(m, void *), ip, sizeof(struct ip)); 509 } else { 510 m->m_data -= optlen; 511 m->m_len += optlen; 512 m->m_pkthdr.len += optlen; 513 ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); 514 } 515 ip = mtod(m, struct ip *); 516 (void)memcpy(ip + 1, p->ipopt_list, (unsigned)optlen); 517 *phlen = sizeof(struct ip) + optlen; 518 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2); 519 ip->ip_len += optlen; 520 return (m); 521} 522 523/* 524 * Copy options from ip to jp, 525 * omitting those not copied during fragmentation. 526 */ 527static int 528ip_optcopy(ip, jp) 529 struct ip *ip, *jp; 530{ 531 register u_char *cp, *dp; 532 int opt, optlen, cnt; 533 534 cp = (u_char *)(ip + 1); 535 dp = (u_char *)(jp + 1); 536 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); 537 for (; cnt > 0; cnt -= optlen, cp += optlen) { 538 opt = cp[0]; 539 if (opt == IPOPT_EOL) 540 break; 541 if (opt == IPOPT_NOP) { 542 /* Preserve for IP mcast tunnel's LSRR alignment. */ 543 *dp++ = IPOPT_NOP; 544 optlen = 1; 545 continue; 546 } else 547 optlen = cp[IPOPT_OLEN]; 548 /* bogus lengths should have been caught by ip_dooptions */ 549 if (optlen > cnt) 550 optlen = cnt; 551 if (IPOPT_COPIED(opt)) { 552 (void)memcpy(dp, cp, (unsigned)optlen); 553 dp += optlen; 554 } 555 } 556 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++) 557 *dp++ = IPOPT_EOL; 558 return (optlen); 559} 560 561/* 562 * IP socket option processing. 563 */ 564int 565ip_ctloutput(op, so, level, optname, mp) 566 int op; 567 struct socket *so; 568 int level, optname; 569 struct mbuf **mp; 570{ 571 register struct inpcb *inp = sotoinpcb(so); 572 register struct mbuf *m = *mp; 573 register int optval = 0; 574 int error = 0; 575 576 if (level != IPPROTO_IP) { 577 error = EINVAL; 578 if (op == PRCO_SETOPT && *mp) 579 (void) m_free(*mp); 580 } else switch (op) { 581 582 case PRCO_SETOPT: 583 switch (optname) { 584 case IP_OPTIONS: 585#ifdef notyet 586 case IP_RETOPTS: 587 return (ip_pcbopts(optname, &inp->inp_options, m)); 588#else 589 return (ip_pcbopts(&inp->inp_options, m)); 590#endif 591 592 case IP_TOS: 593 case IP_TTL: 594 case IP_RECVOPTS: 595 case IP_RECVRETOPTS: 596 case IP_RECVDSTADDR: 597 if (m == 0 || m->m_len != sizeof(int)) 598 error = EINVAL; 599 else { 600 optval = *mtod(m, int *); 601 switch (optname) { 602 603 case IP_TOS: 604 inp->inp_ip.ip_tos = optval; 605 break; 606 607 case IP_TTL: 608 inp->inp_ip.ip_ttl = optval; 609 break; 610#define OPTSET(bit) \ 611 if (optval) \ 612 inp->inp_flags |= bit; \ 613 else \ 614 inp->inp_flags &= ~bit; 615 616 case IP_RECVOPTS: 617 OPTSET(INP_RECVOPTS); 618 break; 619 620 case IP_RECVRETOPTS: 621 OPTSET(INP_RECVRETOPTS); 622 break; 623 624 case IP_RECVDSTADDR: 625 OPTSET(INP_RECVDSTADDR); 626 break; 627 } 628 } 629 break; 630#undef OPTSET 631 632 case IP_MULTICAST_IF: 633 case IP_MULTICAST_VIF: 634 case IP_MULTICAST_TTL: 635 case IP_MULTICAST_LOOP: 636 case IP_ADD_MEMBERSHIP: 637 case IP_DROP_MEMBERSHIP: 638 error = ip_setmoptions(optname, &inp->inp_moptions, m); 639 break; 640 641 case IP_PORTRANGE: 642 if (m == 0 || m->m_len != sizeof(int)) 643 error = EINVAL; 644 else { 645 optval = *mtod(m, int *); 646 647 switch (optval) { 648 649 case IP_PORTRANGE_DEFAULT: 650 inp->inp_flags &= ~(INP_LOWPORT); 651 inp->inp_flags &= ~(INP_HIGHPORT); 652 break; 653 654 case IP_PORTRANGE_HIGH: 655 inp->inp_flags &= ~(INP_LOWPORT); 656 inp->inp_flags |= INP_HIGHPORT; 657 break; 658 659 case IP_PORTRANGE_LOW: 660 inp->inp_flags &= ~(INP_HIGHPORT); 661 inp->inp_flags |= INP_LOWPORT; 662 break; 663 664 default: 665 error = EINVAL; 666 break; 667 } 668 } 669 break; 670 671 default: 672 error = ENOPROTOOPT; 673 break; 674 } 675 if (m) 676 (void)m_free(m); 677 break; 678 679 case PRCO_GETOPT: 680 switch (optname) { 681 case IP_OPTIONS: 682 case IP_RETOPTS: 683 *mp = m = m_get(M_WAIT, MT_SOOPTS); 684 if (inp->inp_options) { 685 m->m_len = inp->inp_options->m_len; 686 (void)memcpy(mtod(m, void *), 687 mtod(inp->inp_options, void *), (unsigned)m->m_len); 688 } else 689 m->m_len = 0; 690 break; 691 692 case IP_TOS: 693 case IP_TTL: 694 case IP_RECVOPTS: 695 case IP_RECVRETOPTS: 696 case IP_RECVDSTADDR: 697 *mp = m = m_get(M_WAIT, MT_SOOPTS); 698 m->m_len = sizeof(int); 699 switch (optname) { 700 701 case IP_TOS: 702 optval = inp->inp_ip.ip_tos; 703 break; 704 705 case IP_TTL: 706 optval = inp->inp_ip.ip_ttl; 707 break; 708 709#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 710 711 case IP_RECVOPTS: 712 optval = OPTBIT(INP_RECVOPTS); 713 break; 714 715 case IP_RECVRETOPTS: 716 optval = OPTBIT(INP_RECVRETOPTS); 717 break; 718 719 case IP_RECVDSTADDR: 720 optval = OPTBIT(INP_RECVDSTADDR); 721 break; 722 } 723 *mtod(m, int *) = optval; 724 break; 725 726 case IP_MULTICAST_IF: 727 case IP_MULTICAST_VIF: 728 case IP_MULTICAST_TTL: 729 case IP_MULTICAST_LOOP: 730 case IP_ADD_MEMBERSHIP: 731 case IP_DROP_MEMBERSHIP: 732 error = ip_getmoptions(optname, inp->inp_moptions, mp); 733 break; 734 735 case IP_PORTRANGE: 736 *mp = m = m_get(M_WAIT, MT_SOOPTS); 737 m->m_len = sizeof(int); 738 739 if (inp->inp_flags & INP_HIGHPORT) 740 optval = IP_PORTRANGE_HIGH; 741 else if (inp->inp_flags & INP_LOWPORT) 742 optval = IP_PORTRANGE_LOW; 743 else 744 optval = 0; 745 746 *mtod(m, int *) = optval; 747 break; 748 749 default: 750 error = ENOPROTOOPT; 751 break; 752 } 753 break; 754 } 755 return (error); 756} 757 758/* 759 * Set up IP options in pcb for insertion in output packets. 760 * Store in mbuf with pointer in pcbopt, adding pseudo-option 761 * with destination address if source routed. 762 */ 763static int 764#ifdef notyet 765ip_pcbopts(optname, pcbopt, m) 766 int optname; 767#else 768ip_pcbopts(pcbopt, m) 769#endif 770 struct mbuf **pcbopt; 771 register struct mbuf *m; 772{ 773 register cnt, optlen; 774 register u_char *cp; 775 u_char opt; 776 777 /* turn off any old options */ 778 if (*pcbopt) 779 (void)m_free(*pcbopt); 780 *pcbopt = 0; 781 if (m == (struct mbuf *)0 || m->m_len == 0) { 782 /* 783 * Only turning off any previous options. 784 */ 785 if (m) 786 (void)m_free(m); 787 return (0); 788 } 789 790#ifndef vax 791 if (m->m_len % sizeof(long)) 792 goto bad; 793#endif 794 /* 795 * IP first-hop destination address will be stored before 796 * actual options; move other options back 797 * and clear it when none present. 798 */ 799 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN]) 800 goto bad; 801 cnt = m->m_len; 802 m->m_len += sizeof(struct in_addr); 803 cp = mtod(m, u_char *) + sizeof(struct in_addr); 804 ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt); 805 bzero(mtod(m, caddr_t), sizeof(struct in_addr)); 806 807 for (; cnt > 0; cnt -= optlen, cp += optlen) { 808 opt = cp[IPOPT_OPTVAL]; 809 if (opt == IPOPT_EOL) 810 break; 811 if (opt == IPOPT_NOP) 812 optlen = 1; 813 else { 814 optlen = cp[IPOPT_OLEN]; 815 if (optlen <= IPOPT_OLEN || optlen > cnt) 816 goto bad; 817 } 818 switch (opt) { 819 820 default: 821 break; 822 823 case IPOPT_LSRR: 824 case IPOPT_SSRR: 825 /* 826 * user process specifies route as: 827 * ->A->B->C->D 828 * D must be our final destination (but we can't 829 * check that since we may not have connected yet). 830 * A is first hop destination, which doesn't appear in 831 * actual IP option, but is stored before the options. 832 */ 833 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr)) 834 goto bad; 835 m->m_len -= sizeof(struct in_addr); 836 cnt -= sizeof(struct in_addr); 837 optlen -= sizeof(struct in_addr); 838 cp[IPOPT_OLEN] = optlen; 839 /* 840 * Move first hop before start of options. 841 */ 842 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t), 843 sizeof(struct in_addr)); 844 /* 845 * Then copy rest of options back 846 * to close up the deleted entry. 847 */ 848 ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] + 849 sizeof(struct in_addr)), 850 (caddr_t)&cp[IPOPT_OFFSET+1], 851 (unsigned)cnt + sizeof(struct in_addr)); 852 break; 853 } 854 } 855 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr)) 856 goto bad; 857 *pcbopt = m; 858 return (0); 859 860bad: 861 (void)m_free(m); 862 return (EINVAL); 863} 864 865/* 866 * Set the IP multicast options in response to user setsockopt(). 867 */ 868static int 869ip_setmoptions(optname, imop, m) 870 int optname; 871 struct ip_moptions **imop; 872 struct mbuf *m; 873{ 874 register int error = 0; 875 u_char loop; 876 register int i; 877 struct in_addr addr; 878 register struct ip_mreq *mreq; 879 register struct ifnet *ifp; 880 register struct ip_moptions *imo = *imop; 881 struct route ro; 882 register struct sockaddr_in *dst; 883 int s; 884 885 if (imo == NULL) { 886 /* 887 * No multicast option buffer attached to the pcb; 888 * allocate one and initialize to default values. 889 */ 890 imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS, 891 M_WAITOK); 892 893 if (imo == NULL) 894 return (ENOBUFS); 895 *imop = imo; 896 imo->imo_multicast_ifp = NULL; 897 imo->imo_multicast_vif = -1; 898 imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL; 899 imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP; 900 imo->imo_num_memberships = 0; 901 } 902 903 switch (optname) { 904 /* store an index number for the vif you wanna use in the send */ 905 case IP_MULTICAST_VIF: 906 if (!legal_vif_num) { 907 error = EOPNOTSUPP; 908 break; 909 } 910 if (m == NULL || m->m_len != sizeof(int)) { 911 error = EINVAL; 912 break; 913 } 914 i = *(mtod(m, int *)); 915 if (!legal_vif_num(i) && (i != -1)) { 916 error = EINVAL; 917 break; 918 } 919 imo->imo_multicast_vif = i; 920 break; 921 922 case IP_MULTICAST_IF: 923 /* 924 * Select the interface for outgoing multicast packets. 925 */ 926 if (m == NULL || m->m_len != sizeof(struct in_addr)) { 927 error = EINVAL; 928 break; 929 } 930 addr = *(mtod(m, struct in_addr *)); 931 /* 932 * INADDR_ANY is used to remove a previous selection. 933 * When no interface is selected, a default one is 934 * chosen every time a multicast packet is sent. 935 */ 936 if (addr.s_addr == INADDR_ANY) { 937 imo->imo_multicast_ifp = NULL; 938 break; 939 } 940 /* 941 * The selected interface is identified by its local 942 * IP address. Find the interface and confirm that 943 * it supports multicasting. 944 */ 945 s = splimp(); 946 INADDR_TO_IFP(addr, ifp); 947 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 948 splx(s); 949 error = EADDRNOTAVAIL; 950 break; 951 } 952 imo->imo_multicast_ifp = ifp; 953 splx(s); 954 break; 955 956 case IP_MULTICAST_TTL: 957 /* 958 * Set the IP time-to-live for outgoing multicast packets. 959 */ 960 if (m == NULL || m->m_len != 1) { 961 error = EINVAL; 962 break; 963 } 964 imo->imo_multicast_ttl = *(mtod(m, u_char *)); 965 break; 966 967 case IP_MULTICAST_LOOP: 968 /* 969 * Set the loopback flag for outgoing multicast packets. 970 * Must be zero or one. 971 */ 972 if (m == NULL || m->m_len != 1 || 973 (loop = *(mtod(m, u_char *))) > 1) { 974 error = EINVAL; 975 break; 976 } 977 imo->imo_multicast_loop = loop; 978 break; 979 980 case IP_ADD_MEMBERSHIP: 981 /* 982 * Add a multicast group membership. 983 * Group must be a valid IP multicast address. 984 */ 985 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 986 error = EINVAL; 987 break; 988 } 989 mreq = mtod(m, struct ip_mreq *); 990 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 991 error = EINVAL; 992 break; 993 } 994 s = splimp(); 995 /* 996 * If no interface address was provided, use the interface of 997 * the route to the given multicast address. 998 */ 999 if (mreq->imr_interface.s_addr == INADDR_ANY) { 1000 bzero((caddr_t)&ro, sizeof(ro)); 1001 dst = (struct sockaddr_in *)&ro.ro_dst; 1002 dst->sin_len = sizeof(*dst); 1003 dst->sin_family = AF_INET; 1004 dst->sin_addr = mreq->imr_multiaddr; 1005 rtalloc(&ro); 1006 if (ro.ro_rt == NULL) { 1007 error = EADDRNOTAVAIL; 1008 splx(s); 1009 break; 1010 } 1011 ifp = ro.ro_rt->rt_ifp; 1012 rtfree(ro.ro_rt); 1013 } 1014 else { 1015 INADDR_TO_IFP(mreq->imr_interface, ifp); 1016 } 1017 1018 /* 1019 * See if we found an interface, and confirm that it 1020 * supports multicast. 1021 */ 1022 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1023 error = EADDRNOTAVAIL; 1024 splx(s); 1025 break; 1026 } 1027 /* 1028 * See if the membership already exists or if all the 1029 * membership slots are full. 1030 */ 1031 for (i = 0; i < imo->imo_num_memberships; ++i) { 1032 if (imo->imo_membership[i]->inm_ifp == ifp && 1033 imo->imo_membership[i]->inm_addr.s_addr 1034 == mreq->imr_multiaddr.s_addr) 1035 break; 1036 } 1037 if (i < imo->imo_num_memberships) { 1038 error = EADDRINUSE; 1039 splx(s); 1040 break; 1041 } 1042 if (i == IP_MAX_MEMBERSHIPS) { 1043 error = ETOOMANYREFS; 1044 splx(s); 1045 break; 1046 } 1047 /* 1048 * Everything looks good; add a new record to the multicast 1049 * address list for the given interface. 1050 */ 1051 if ((imo->imo_membership[i] = 1052 in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) { 1053 error = ENOBUFS; 1054 splx(s); 1055 break; 1056 } 1057 ++imo->imo_num_memberships; 1058 splx(s); 1059 break; 1060 1061 case IP_DROP_MEMBERSHIP: 1062 /* 1063 * Drop a multicast group membership. 1064 * Group must be a valid IP multicast address. 1065 */ 1066 if (m == NULL || m->m_len != sizeof(struct ip_mreq)) { 1067 error = EINVAL; 1068 break; 1069 } 1070 mreq = mtod(m, struct ip_mreq *); 1071 if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) { 1072 error = EINVAL; 1073 break; 1074 } 1075 1076 s = splimp(); 1077 /* 1078 * If an interface address was specified, get a pointer 1079 * to its ifnet structure. 1080 */ 1081 if (mreq->imr_interface.s_addr == INADDR_ANY) 1082 ifp = NULL; 1083 else { 1084 INADDR_TO_IFP(mreq->imr_interface, ifp); 1085 if (ifp == NULL) { 1086 error = EADDRNOTAVAIL; 1087 splx(s); 1088 break; 1089 } 1090 } 1091 /* 1092 * Find the membership in the membership array. 1093 */ 1094 for (i = 0; i < imo->imo_num_memberships; ++i) { 1095 if ((ifp == NULL || 1096 imo->imo_membership[i]->inm_ifp == ifp) && 1097 imo->imo_membership[i]->inm_addr.s_addr == 1098 mreq->imr_multiaddr.s_addr) 1099 break; 1100 } 1101 if (i == imo->imo_num_memberships) { 1102 error = EADDRNOTAVAIL; 1103 splx(s); 1104 break; 1105 } 1106 /* 1107 * Give up the multicast address record to which the 1108 * membership points. 1109 */ 1110 in_delmulti(imo->imo_membership[i]); 1111 /* 1112 * Remove the gap in the membership array. 1113 */ 1114 for (++i; i < imo->imo_num_memberships; ++i) 1115 imo->imo_membership[i-1] = imo->imo_membership[i]; 1116 --imo->imo_num_memberships; 1117 splx(s); 1118 break; 1119 1120 default: 1121 error = EOPNOTSUPP; 1122 break; 1123 } 1124 1125 /* 1126 * If all options have default values, no need to keep the mbuf. 1127 */ 1128 if (imo->imo_multicast_ifp == NULL && 1129 imo->imo_multicast_vif == -1 && 1130 imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL && 1131 imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP && 1132 imo->imo_num_memberships == 0) { 1133 free(*imop, M_IPMOPTS); 1134 *imop = NULL; 1135 } 1136 1137 return (error); 1138} 1139 1140/* 1141 * Return the IP multicast options in response to user getsockopt(). 1142 */ 1143static int 1144ip_getmoptions(optname, imo, mp) 1145 int optname; 1146 register struct ip_moptions *imo; 1147 register struct mbuf **mp; 1148{ 1149 u_char *ttl; 1150 u_char *loop; 1151 struct in_addr *addr; 1152 struct in_ifaddr *ia; 1153 1154 *mp = m_get(M_WAIT, MT_SOOPTS); 1155 1156 switch (optname) { 1157 1158 case IP_MULTICAST_VIF: 1159 if (imo != NULL) 1160 *(mtod(*mp, int *)) = imo->imo_multicast_vif; 1161 else 1162 *(mtod(*mp, int *)) = -1; 1163 (*mp)->m_len = sizeof(int); 1164 return(0); 1165 1166 case IP_MULTICAST_IF: 1167 addr = mtod(*mp, struct in_addr *); 1168 (*mp)->m_len = sizeof(struct in_addr); 1169 if (imo == NULL || imo->imo_multicast_ifp == NULL) 1170 addr->s_addr = INADDR_ANY; 1171 else { 1172 IFP_TO_IA(imo->imo_multicast_ifp, ia); 1173 addr->s_addr = (ia == NULL) ? INADDR_ANY 1174 : IA_SIN(ia)->sin_addr.s_addr; 1175 } 1176 return (0); 1177 1178 case IP_MULTICAST_TTL: 1179 ttl = mtod(*mp, u_char *); 1180 (*mp)->m_len = 1; 1181 *ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL 1182 : imo->imo_multicast_ttl; 1183 return (0); 1184 1185 case IP_MULTICAST_LOOP: 1186 loop = mtod(*mp, u_char *); 1187 (*mp)->m_len = 1; 1188 *loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP 1189 : imo->imo_multicast_loop; 1190 return (0); 1191 1192 default: 1193 return (EOPNOTSUPP); 1194 } 1195} 1196 1197/* 1198 * Discard the IP multicast options. 1199 */ 1200void 1201ip_freemoptions(imo) 1202 register struct ip_moptions *imo; 1203{ 1204 register int i; 1205 1206 if (imo != NULL) { 1207 for (i = 0; i < imo->imo_num_memberships; ++i) 1208 in_delmulti(imo->imo_membership[i]); 1209 free(imo, M_IPMOPTS); 1210 } 1211} 1212 1213/* 1214 * Routine called from ip_output() to loop back a copy of an IP multicast 1215 * packet to the input queue of a specified interface. Note that this 1216 * calls the output routine of the loopback "driver", but with an interface 1217 * pointer that might NOT be a loopback interface -- evil, but easier than 1218 * replicating that code here. 1219 */ 1220static void 1221ip_mloopback(ifp, m, dst) 1222 struct ifnet *ifp; 1223 register struct mbuf *m; 1224 register struct sockaddr_in *dst; 1225{ 1226 register struct ip *ip; 1227 struct mbuf *copym; 1228 1229 copym = m_copy(m, 0, M_COPYALL); 1230 if (copym != NULL) { 1231 /* 1232 * We don't bother to fragment if the IP length is greater 1233 * than the interface's MTU. Can this possibly matter? 1234 */ 1235 ip = mtod(copym, struct ip *); 1236 ip->ip_len = htons((u_short)ip->ip_len); 1237 ip->ip_off = htons((u_short)ip->ip_off); 1238 ip->ip_sum = 0; 1239 if (ip->ip_vhl == IP_VHL_BORING) { 1240 ip->ip_sum = in_cksum_hdr(ip); 1241 } else { 1242 ip->ip_sum = in_cksum(copym, 1243 IP_VHL_HL(ip->ip_vhl) << 2); 1244 } 1245 /* 1246 * NB: 1247 * We can't simply call ip_input() directly because 1248 * the ip_mforward() depends on the `input interface' 1249 * being set to something unreasonable so that we don't 1250 * attempt to forward the looped-back copy. 1251 * It's also not clear whether there are any lingering 1252 * reentrancy problems in other areas which might be 1253 * exposed by this code. For the moment, we'll err 1254 * on the side of safety by continuing to abuse 1255 * loinput(). 1256 */ 1257#ifdef notdef 1258 copym->m_pkthdr.rcvif = &loif[0]; 1259 ip_input(copym) 1260#else 1261 (void) looutput(ifp, copym, (struct sockaddr *)dst, NULL); 1262#endif 1263 } 1264} 1265