ip6_output.c revision 1.75
1 /* $NetBSD: ip6_output.c,v 1.75 2003/12/10 22:35:35 itojun Exp $ */ 2 /* $KAME: ip6_output.c,v 1.172 2001/03/25 09:55:56 itojun Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1988, 1990, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)ip_output.c 8.3 (Berkeley) 1/21/94 62 */ 63 64 #include <sys/cdefs.h> 65 __KERNEL_RCSID(0, "$NetBSD: ip6_output.c,v 1.75 2003/12/10 22:35:35 itojun Exp $"); 66 67 #include "opt_inet.h" 68 #include "opt_ipsec.h" 69 #include "opt_pfil_hooks.h" 70 71 #include <sys/param.h> 72 #include <sys/malloc.h> 73 #include <sys/mbuf.h> 74 #include <sys/errno.h> 75 #include <sys/protosw.h> 76 #include <sys/socket.h> 77 #include <sys/socketvar.h> 78 #include <sys/systm.h> 79 #include <sys/proc.h> 80 81 #include <net/if.h> 82 #include <net/route.h> 83 #ifdef PFIL_HOOKS 84 #include <net/pfil.h> 85 #endif 86 87 #include <netinet/in.h> 88 #include <netinet/in_var.h> 89 #include <netinet/ip6.h> 90 #include <netinet/icmp6.h> 91 #include <netinet6/ip6_var.h> 92 #include <netinet6/in6_pcb.h> 93 #include <netinet6/nd6.h> 94 95 #ifdef IPSEC 96 #include <netinet6/ipsec.h> 97 #include <netkey/key.h> 98 #endif /* IPSEC */ 99 100 #include "loop.h" 101 102 #include <net/net_osdep.h> 103 104 #ifdef PFIL_HOOKS 105 extern struct pfil_head inet6_pfil_hook; /* XXX */ 106 #endif 107 108 struct ip6_exthdrs { 109 struct mbuf *ip6e_ip6; 110 struct mbuf *ip6e_hbh; 111 struct mbuf *ip6e_dest1; 112 struct mbuf *ip6e_rthdr; 113 struct mbuf *ip6e_dest2; 114 }; 115 116 static int ip6_pcbopts __P((struct ip6_pktopts **, struct mbuf *, 117 struct socket *)); 118 static int ip6_setmoptions __P((int, struct ip6_moptions **, struct mbuf *)); 119 static int ip6_getmoptions __P((int, struct ip6_moptions *, struct mbuf **)); 120 static int ip6_copyexthdr __P((struct mbuf **, caddr_t, int)); 121 static int ip6_insertfraghdr __P((struct mbuf *, struct mbuf *, int, 122 struct ip6_frag **)); 123 static int ip6_insert_jumboopt __P((struct ip6_exthdrs *, u_int32_t)); 124 static int ip6_splithdr __P((struct mbuf *, struct ip6_exthdrs *)); 125 static int ip6_getpmtu __P((struct route_in6 *, struct route_in6 *, 126 struct ifnet *, struct in6_addr *, u_long *)); 127 128 extern struct ifnet loif[NLOOP]; 129 130 /* 131 * IP6 output. The packet in mbuf chain m contains a skeletal IP6 132 * header (with pri, len, nxt, hlim, src, dst). 133 * This function may modify ver and hlim only. 134 * The mbuf chain containing the packet will be freed. 135 * The mbuf opt, if present, will not be freed. 136 * 137 * type of "mtu": rt_rmx.rmx_mtu is u_long, ifnet.ifr_mtu is int, and 138 * nd_ifinfo.linkmtu is u_int32_t. so we use u_long to hold largest one, 139 * which is rt_rmx.rmx_mtu. 140 */ 141 int 142 ip6_output(m0, opt, ro, flags, im6o, so, ifpp) 143 struct mbuf *m0; 144 struct ip6_pktopts *opt; 145 struct route_in6 *ro; 146 int flags; 147 struct ip6_moptions *im6o; 148 struct socket *so; 149 struct ifnet **ifpp; /* XXX: just for statistics */ 150 { 151 struct ip6_hdr *ip6, *mhip6; 152 struct ifnet *ifp, *origifp; 153 struct mbuf *m = m0; 154 int hlen, tlen, len, off; 155 struct route_in6 ip6route; 156 struct sockaddr_in6 *dst; 157 int error = 0; 158 u_long mtu; 159 u_int32_t optlen = 0, plen = 0, unfragpartlen = 0; 160 struct ip6_exthdrs exthdrs; 161 struct in6_addr finaldst; 162 struct route_in6 *ro_pmtu = NULL; 163 int hdrsplit = 0; 164 int needipsec = 0; 165 #ifdef IPSEC 166 int needipsectun = 0; 167 struct secpolicy *sp = NULL; 168 169 ip6 = mtod(m, struct ip6_hdr *); 170 #endif /* IPSEC */ 171 172 #define MAKE_EXTHDR(hp, mp) \ 173 do { \ 174 if (hp) { \ 175 struct ip6_ext *eh = (struct ip6_ext *)(hp); \ 176 error = ip6_copyexthdr((mp), (caddr_t)(hp), \ 177 ((eh)->ip6e_len + 1) << 3); \ 178 if (error) \ 179 goto freehdrs; \ 180 } \ 181 } while (/*CONSTCOND*/ 0) 182 183 bzero(&exthdrs, sizeof(exthdrs)); 184 if (opt) { 185 /* Hop-by-Hop options header */ 186 MAKE_EXTHDR(opt->ip6po_hbh, &exthdrs.ip6e_hbh); 187 /* Destination options header(1st part) */ 188 MAKE_EXTHDR(opt->ip6po_dest1, &exthdrs.ip6e_dest1); 189 /* Routing header */ 190 MAKE_EXTHDR(opt->ip6po_rthdr, &exthdrs.ip6e_rthdr); 191 /* Destination options header(2nd part) */ 192 MAKE_EXTHDR(opt->ip6po_dest2, &exthdrs.ip6e_dest2); 193 } 194 195 #ifdef IPSEC 196 /* get a security policy for this packet */ 197 if (so == NULL) 198 sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, 0, &error); 199 else 200 sp = ipsec6_getpolicybysock(m, IPSEC_DIR_OUTBOUND, so, &error); 201 202 if (sp == NULL) { 203 ipsec6stat.out_inval++; 204 goto freehdrs; 205 } 206 207 error = 0; 208 209 /* check policy */ 210 switch (sp->policy) { 211 case IPSEC_POLICY_DISCARD: 212 /* 213 * This packet is just discarded. 214 */ 215 ipsec6stat.out_polvio++; 216 goto freehdrs; 217 218 case IPSEC_POLICY_BYPASS: 219 case IPSEC_POLICY_NONE: 220 /* no need to do IPsec. */ 221 needipsec = 0; 222 break; 223 224 case IPSEC_POLICY_IPSEC: 225 if (sp->req == NULL) { 226 /* XXX should be panic ? */ 227 printf("ip6_output: No IPsec request specified.\n"); 228 error = EINVAL; 229 goto freehdrs; 230 } 231 needipsec = 1; 232 break; 233 234 case IPSEC_POLICY_ENTRUST: 235 default: 236 printf("ip6_output: Invalid policy found. %d\n", sp->policy); 237 } 238 #endif /* IPSEC */ 239 240 /* 241 * Calculate the total length of the extension header chain. 242 * Keep the length of the unfragmentable part for fragmentation. 243 */ 244 optlen = 0; 245 if (exthdrs.ip6e_hbh) optlen += exthdrs.ip6e_hbh->m_len; 246 if (exthdrs.ip6e_dest1) optlen += exthdrs.ip6e_dest1->m_len; 247 if (exthdrs.ip6e_rthdr) optlen += exthdrs.ip6e_rthdr->m_len; 248 unfragpartlen = optlen + sizeof(struct ip6_hdr); 249 /* NOTE: we don't add AH/ESP length here. do that later. */ 250 if (exthdrs.ip6e_dest2) optlen += exthdrs.ip6e_dest2->m_len; 251 252 /* 253 * If we need IPsec, or there is at least one extension header, 254 * separate IP6 header from the payload. 255 */ 256 if ((needipsec || optlen) && !hdrsplit) { 257 if ((error = ip6_splithdr(m, &exthdrs)) != 0) { 258 m = NULL; 259 goto freehdrs; 260 } 261 m = exthdrs.ip6e_ip6; 262 hdrsplit++; 263 } 264 265 /* adjust pointer */ 266 ip6 = mtod(m, struct ip6_hdr *); 267 268 /* adjust mbuf packet header length */ 269 m->m_pkthdr.len += optlen; 270 plen = m->m_pkthdr.len - sizeof(*ip6); 271 272 /* If this is a jumbo payload, insert a jumbo payload option. */ 273 if (plen > IPV6_MAXPACKET) { 274 if (!hdrsplit) { 275 if ((error = ip6_splithdr(m, &exthdrs)) != 0) { 276 m = NULL; 277 goto freehdrs; 278 } 279 m = exthdrs.ip6e_ip6; 280 hdrsplit++; 281 } 282 /* adjust pointer */ 283 ip6 = mtod(m, struct ip6_hdr *); 284 if ((error = ip6_insert_jumboopt(&exthdrs, plen)) != 0) 285 goto freehdrs; 286 ip6->ip6_plen = 0; 287 } else 288 ip6->ip6_plen = htons(plen); 289 290 /* 291 * Concatenate headers and fill in next header fields. 292 * Here we have, on "m" 293 * IPv6 payload 294 * and we insert headers accordingly. Finally, we should be getting: 295 * IPv6 hbh dest1 rthdr ah* [esp* dest2 payload] 296 * 297 * during the header composing process, "m" points to IPv6 header. 298 * "mprev" points to an extension header prior to esp. 299 */ 300 { 301 u_char *nexthdrp = &ip6->ip6_nxt; 302 struct mbuf *mprev = m; 303 304 /* 305 * we treat dest2 specially. this makes IPsec processing 306 * much easier. 307 * 308 * result: IPv6 dest2 payload 309 * m and mprev will point to IPv6 header. 310 */ 311 if (exthdrs.ip6e_dest2) { 312 if (!hdrsplit) 313 panic("assumption failed: hdr not split"); 314 exthdrs.ip6e_dest2->m_next = m->m_next; 315 m->m_next = exthdrs.ip6e_dest2; 316 *mtod(exthdrs.ip6e_dest2, u_char *) = ip6->ip6_nxt; 317 ip6->ip6_nxt = IPPROTO_DSTOPTS; 318 } 319 320 #define MAKE_CHAIN(m, mp, p, i)\ 321 do {\ 322 if (m) {\ 323 if (!hdrsplit) \ 324 panic("assumption failed: hdr not split"); \ 325 *mtod((m), u_char *) = *(p);\ 326 *(p) = (i);\ 327 p = mtod((m), u_char *);\ 328 (m)->m_next = (mp)->m_next;\ 329 (mp)->m_next = (m);\ 330 (mp) = (m);\ 331 }\ 332 } while (/*CONSTCOND*/ 0) 333 /* 334 * result: IPv6 hbh dest1 rthdr dest2 payload 335 * m will point to IPv6 header. mprev will point to the 336 * extension header prior to dest2 (rthdr in the above case). 337 */ 338 MAKE_CHAIN(exthdrs.ip6e_hbh, mprev, nexthdrp, IPPROTO_HOPOPTS); 339 MAKE_CHAIN(exthdrs.ip6e_dest1, mprev, nexthdrp, 340 IPPROTO_DSTOPTS); 341 MAKE_CHAIN(exthdrs.ip6e_rthdr, mprev, nexthdrp, 342 IPPROTO_ROUTING); 343 344 #ifdef IPSEC 345 if (!needipsec) 346 goto skip_ipsec2; 347 348 /* 349 * pointers after IPsec headers are not valid any more. 350 * other pointers need a great care too. 351 * (IPsec routines should not mangle mbufs prior to AH/ESP) 352 */ 353 exthdrs.ip6e_dest2 = NULL; 354 355 { 356 struct ip6_rthdr *rh = NULL; 357 int segleft_org = 0; 358 struct ipsec_output_state state; 359 360 if (exthdrs.ip6e_rthdr) { 361 rh = mtod(exthdrs.ip6e_rthdr, struct ip6_rthdr *); 362 segleft_org = rh->ip6r_segleft; 363 rh->ip6r_segleft = 0; 364 } 365 366 bzero(&state, sizeof(state)); 367 state.m = m; 368 error = ipsec6_output_trans(&state, nexthdrp, mprev, sp, flags, 369 &needipsectun); 370 m = state.m; 371 if (error) { 372 /* mbuf is already reclaimed in ipsec6_output_trans. */ 373 m = NULL; 374 switch (error) { 375 case EHOSTUNREACH: 376 case ENETUNREACH: 377 case EMSGSIZE: 378 case ENOBUFS: 379 case ENOMEM: 380 break; 381 default: 382 printf("ip6_output (ipsec): error code %d\n", error); 383 /* FALLTHROUGH */ 384 case ENOENT: 385 /* don't show these error codes to the user */ 386 error = 0; 387 break; 388 } 389 goto bad; 390 } 391 if (exthdrs.ip6e_rthdr) { 392 /* ah6_output doesn't modify mbuf chain */ 393 rh->ip6r_segleft = segleft_org; 394 } 395 } 396 skip_ipsec2:; 397 #endif 398 } 399 400 /* 401 * If there is a routing header, replace destination address field 402 * with the first hop of the routing header. 403 */ 404 if (exthdrs.ip6e_rthdr) { 405 struct ip6_rthdr *rh; 406 struct ip6_rthdr0 *rh0; 407 struct in6_addr *addr; 408 409 rh = (struct ip6_rthdr *)(mtod(exthdrs.ip6e_rthdr, 410 struct ip6_rthdr *)); 411 finaldst = ip6->ip6_dst; 412 switch (rh->ip6r_type) { 413 case IPV6_RTHDR_TYPE_0: 414 rh0 = (struct ip6_rthdr0 *)rh; 415 addr = (struct in6_addr *)(rh0 + 1); 416 ip6->ip6_dst = addr[0]; 417 bcopy(&addr[1], &addr[0], 418 sizeof(struct in6_addr) * (rh0->ip6r0_segleft - 1)); 419 addr[rh0->ip6r0_segleft - 1] = finaldst; 420 break; 421 default: /* is it possible? */ 422 error = EINVAL; 423 goto bad; 424 } 425 } 426 427 /* Source address validation */ 428 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) && 429 (flags & IPV6_UNSPECSRC) == 0) { 430 error = EOPNOTSUPP; 431 ip6stat.ip6s_badscope++; 432 goto bad; 433 } 434 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) { 435 error = EOPNOTSUPP; 436 ip6stat.ip6s_badscope++; 437 goto bad; 438 } 439 440 ip6stat.ip6s_localout++; 441 442 /* 443 * Route packet. 444 */ 445 if (ro == 0) { 446 ro = &ip6route; 447 bzero((caddr_t)ro, sizeof(*ro)); 448 } 449 ro_pmtu = ro; 450 if (opt && opt->ip6po_rthdr) 451 ro = &opt->ip6po_route; 452 dst = (struct sockaddr_in6 *)&ro->ro_dst; 453 /* 454 * If there is a cached route, 455 * check that it is to the same destination 456 * and is still up. If not, free it and try again. 457 */ 458 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 459 dst->sin6_family != AF_INET6 || 460 !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &ip6->ip6_dst))) { 461 RTFREE(ro->ro_rt); 462 ro->ro_rt = (struct rtentry *)0; 463 } 464 if (ro->ro_rt == 0) { 465 bzero(dst, sizeof(*dst)); 466 dst->sin6_family = AF_INET6; 467 dst->sin6_len = sizeof(struct sockaddr_in6); 468 dst->sin6_addr = ip6->ip6_dst; 469 } 470 #ifdef IPSEC 471 if (needipsec && needipsectun) { 472 struct ipsec_output_state state; 473 474 /* 475 * All the extension headers will become inaccessible 476 * (since they can be encrypted). 477 * Don't panic, we need no more updates to extension headers 478 * on inner IPv6 packet (since they are now encapsulated). 479 * 480 * IPv6 [ESP|AH] IPv6 [extension headers] payload 481 */ 482 bzero(&exthdrs, sizeof(exthdrs)); 483 exthdrs.ip6e_ip6 = m; 484 485 bzero(&state, sizeof(state)); 486 state.m = m; 487 state.ro = (struct route *)ro; 488 state.dst = (struct sockaddr *)dst; 489 490 error = ipsec6_output_tunnel(&state, sp, flags); 491 492 m = state.m; 493 ro = (struct route_in6 *)state.ro; 494 dst = (struct sockaddr_in6 *)state.dst; 495 if (error) { 496 /* mbuf is already reclaimed in ipsec6_output_tunnel. */ 497 m0 = m = NULL; 498 m = NULL; 499 switch (error) { 500 case EHOSTUNREACH: 501 case ENETUNREACH: 502 case EMSGSIZE: 503 case ENOBUFS: 504 case ENOMEM: 505 break; 506 default: 507 printf("ip6_output (ipsec): error code %d\n", error); 508 /* FALLTHROUGH */ 509 case ENOENT: 510 /* don't show these error codes to the user */ 511 error = 0; 512 break; 513 } 514 goto bad; 515 } 516 517 exthdrs.ip6e_ip6 = m; 518 } 519 #endif /* IPSEC */ 520 521 if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 522 /* Unicast */ 523 524 #define ifatoia6(ifa) ((struct in6_ifaddr *)(ifa)) 525 #define sin6tosa(sin6) ((struct sockaddr *)(sin6)) 526 /* xxx 527 * interface selection comes here 528 * if an interface is specified from an upper layer, 529 * ifp must point it. 530 */ 531 if (ro->ro_rt == 0) { 532 /* 533 * non-bsdi always clone routes, if parent is 534 * PRF_CLONING. 535 */ 536 rtalloc((struct route *)ro); 537 } 538 if (ro->ro_rt == 0) { 539 ip6stat.ip6s_noroute++; 540 error = EHOSTUNREACH; 541 /* XXX in6_ifstat_inc(ifp, ifs6_out_discard); */ 542 goto bad; 543 } 544 ifp = ro->ro_rt->rt_ifp; 545 ro->ro_rt->rt_use++; 546 if (ro->ro_rt->rt_flags & RTF_GATEWAY) 547 dst = (struct sockaddr_in6 *)ro->ro_rt->rt_gateway; 548 m->m_flags &= ~(M_BCAST | M_MCAST); /* just in case */ 549 550 in6_ifstat_inc(ifp, ifs6_out_request); 551 552 /* 553 * Check if the outgoing interface conflicts with 554 * the interface specified by ifi6_ifindex (if specified). 555 * Note that loopback interface is always okay. 556 * (this may happen when we are sending a packet to one of 557 * our own addresses.) 558 */ 559 if (opt && opt->ip6po_pktinfo 560 && opt->ip6po_pktinfo->ipi6_ifindex) { 561 if (!(ifp->if_flags & IFF_LOOPBACK) 562 && ifp->if_index != opt->ip6po_pktinfo->ipi6_ifindex) { 563 ip6stat.ip6s_noroute++; 564 in6_ifstat_inc(ifp, ifs6_out_discard); 565 error = EHOSTUNREACH; 566 goto bad; 567 } 568 } 569 570 if (opt && opt->ip6po_hlim != -1) 571 ip6->ip6_hlim = opt->ip6po_hlim & 0xff; 572 } else { 573 /* Multicast */ 574 struct in6_multi *in6m; 575 576 m->m_flags = (m->m_flags & ~M_BCAST) | M_MCAST; 577 578 /* 579 * See if the caller provided any multicast options 580 */ 581 ifp = NULL; 582 if (im6o != NULL) { 583 ip6->ip6_hlim = im6o->im6o_multicast_hlim; 584 if (im6o->im6o_multicast_ifp != NULL) 585 ifp = im6o->im6o_multicast_ifp; 586 } else 587 ip6->ip6_hlim = ip6_defmcasthlim; 588 589 /* 590 * See if the caller provided the outgoing interface 591 * as an ancillary data. 592 * Boundary check for ifindex is assumed to be already done. 593 */ 594 if (opt && opt->ip6po_pktinfo && opt->ip6po_pktinfo->ipi6_ifindex) 595 ifp = ifindex2ifnet[opt->ip6po_pktinfo->ipi6_ifindex]; 596 597 /* 598 * If the destination is a node-local scope multicast, 599 * the packet should be loop-backed only. 600 */ 601 if (IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst)) { 602 /* 603 * If the outgoing interface is already specified, 604 * it should be a loopback interface. 605 */ 606 if (ifp && (ifp->if_flags & IFF_LOOPBACK) == 0) { 607 ip6stat.ip6s_badscope++; 608 error = ENETUNREACH; /* XXX: better error? */ 609 /* XXX correct ifp? */ 610 in6_ifstat_inc(ifp, ifs6_out_discard); 611 goto bad; 612 } else { 613 ifp = &loif[0]; 614 } 615 } 616 617 if (opt && opt->ip6po_hlim != -1) 618 ip6->ip6_hlim = opt->ip6po_hlim & 0xff; 619 620 /* 621 * If caller did not provide an interface lookup a 622 * default in the routing table. This is either a 623 * default for the speicfied group (i.e. a host 624 * route), or a multicast default (a route for the 625 * ``net'' ff00::/8). 626 */ 627 if (ifp == NULL) { 628 if (ro->ro_rt == 0) { 629 ro->ro_rt = rtalloc1((struct sockaddr *) 630 &ro->ro_dst, 0); 631 } 632 if (ro->ro_rt == 0) { 633 ip6stat.ip6s_noroute++; 634 error = EHOSTUNREACH; 635 /* XXX in6_ifstat_inc(ifp, ifs6_out_discard) */ 636 goto bad; 637 } 638 ifp = ro->ro_rt->rt_ifp; 639 ro->ro_rt->rt_use++; 640 } 641 642 if ((flags & IPV6_FORWARDING) == 0) 643 in6_ifstat_inc(ifp, ifs6_out_request); 644 in6_ifstat_inc(ifp, ifs6_out_mcast); 645 646 /* 647 * Confirm that the outgoing interface supports multicast. 648 */ 649 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 650 ip6stat.ip6s_noroute++; 651 in6_ifstat_inc(ifp, ifs6_out_discard); 652 error = ENETUNREACH; 653 goto bad; 654 } 655 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 656 if (in6m != NULL && 657 (im6o == NULL || im6o->im6o_multicast_loop)) { 658 /* 659 * If we belong to the destination multicast group 660 * on the outgoing interface, and the caller did not 661 * forbid loopback, loop back a copy. 662 */ 663 ip6_mloopback(ifp, m, dst); 664 } else { 665 /* 666 * If we are acting as a multicast router, perform 667 * multicast forwarding as if the packet had just 668 * arrived on the interface to which we are about 669 * to send. The multicast forwarding function 670 * recursively calls this function, using the 671 * IPV6_FORWARDING flag to prevent infinite recursion. 672 * 673 * Multicasts that are looped back by ip6_mloopback(), 674 * above, will be forwarded by the ip6_input() routine, 675 * if necessary. 676 */ 677 if (ip6_mrouter && (flags & IPV6_FORWARDING) == 0) { 678 if (ip6_mforward(ip6, ifp, m) != 0) { 679 m_freem(m); 680 goto done; 681 } 682 } 683 } 684 /* 685 * Multicasts with a hoplimit of zero may be looped back, 686 * above, but must not be transmitted on a network. 687 * Also, multicasts addressed to the loopback interface 688 * are not sent -- the above call to ip6_mloopback() will 689 * loop back a copy if this host actually belongs to the 690 * destination group on the loopback interface. 691 */ 692 if (ip6->ip6_hlim == 0 || (ifp->if_flags & IFF_LOOPBACK)) { 693 m_freem(m); 694 goto done; 695 } 696 } 697 698 /* 699 * Fill the outgoing inteface to tell the upper layer 700 * to increment per-interface statistics. 701 */ 702 if (ifpp) 703 *ifpp = ifp; 704 705 /* Determine path MTU. */ 706 if ((error = ip6_getpmtu(ro_pmtu, ro, ifp, &finaldst, &mtu)) != 0) 707 goto bad; 708 709 /* 710 * The caller of this function may specify to use the minimum MTU 711 * in some cases. 712 */ 713 if (mtu > IPV6_MMTU) { 714 if ((flags & IPV6_MINMTU)) 715 mtu = IPV6_MMTU; 716 } 717 718 /* Fake scoped addresses */ 719 if ((ifp->if_flags & IFF_LOOPBACK) != 0) { 720 /* 721 * If source or destination address is a scoped address, and 722 * the packet is going to be sent to a loopback interface, 723 * we should keep the original interface. 724 */ 725 726 /* 727 * XXX: this is a very experimental and temporary solution. 728 * We eventually have sockaddr_in6 and use the sin6_scope_id 729 * field of the structure here. 730 * We rely on the consistency between two scope zone ids 731 * of source add destination, which should already be assured 732 * Larger scopes than link will be supported in the near 733 * future. 734 */ 735 origifp = NULL; 736 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 737 origifp = ifindex2ifnet[ntohs(ip6->ip6_src.s6_addr16[1])]; 738 else if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 739 origifp = ifindex2ifnet[ntohs(ip6->ip6_dst.s6_addr16[1])]; 740 /* 741 * XXX: origifp can be NULL even in those two cases above. 742 * For example, if we remove the (only) link-local address 743 * from the loopback interface, and try to send a link-local 744 * address without link-id information. Then the source 745 * address is ::1, and the destination address is the 746 * link-local address with its s6_addr16[1] being zero. 747 * What is worse, if the packet goes to the loopback interface 748 * by a default rejected route, the null pointer would be 749 * passed to looutput, and the kernel would hang. 750 * The following last resort would prevent such disaster. 751 */ 752 if (origifp == NULL) 753 origifp = ifp; 754 } else 755 origifp = ifp; 756 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 757 ip6->ip6_src.s6_addr16[1] = 0; 758 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 759 ip6->ip6_dst.s6_addr16[1] = 0; 760 761 /* 762 * If the outgoing packet contains a hop-by-hop options header, 763 * it must be examined and processed even by the source node. 764 * (RFC 2460, section 4.) 765 */ 766 if (exthdrs.ip6e_hbh) { 767 struct ip6_hbh *hbh = mtod(exthdrs.ip6e_hbh, struct ip6_hbh *); 768 u_int32_t dummy1; /* XXX unused */ 769 u_int32_t dummy2; /* XXX unused */ 770 771 /* 772 * XXX: if we have to send an ICMPv6 error to the sender, 773 * we need the M_LOOP flag since icmp6_error() expects 774 * the IPv6 and the hop-by-hop options header are 775 * continuous unless the flag is set. 776 */ 777 m->m_flags |= M_LOOP; 778 m->m_pkthdr.rcvif = ifp; 779 if (ip6_process_hopopts(m, (u_int8_t *)(hbh + 1), 780 ((hbh->ip6h_len + 1) << 3) - sizeof(struct ip6_hbh), 781 &dummy1, &dummy2) < 0) { 782 /* m was already freed at this point */ 783 error = EINVAL;/* better error? */ 784 goto done; 785 } 786 m->m_flags &= ~M_LOOP; /* XXX */ 787 m->m_pkthdr.rcvif = NULL; 788 } 789 790 #ifdef PFIL_HOOKS 791 /* 792 * Run through list of hooks for output packets. 793 */ 794 if ((error = pfil_run_hooks(&inet6_pfil_hook, &m, ifp, PFIL_OUT)) != 0) 795 goto done; 796 if (m == NULL) 797 goto done; 798 ip6 = mtod(m, struct ip6_hdr *); 799 #endif /* PFIL_HOOKS */ 800 /* 801 * Send the packet to the outgoing interface. 802 * If necessary, do IPv6 fragmentation before sending. 803 */ 804 tlen = m->m_pkthdr.len; 805 if (tlen <= mtu) { 806 #ifdef IFA_STATS 807 struct in6_ifaddr *ia6; 808 ip6 = mtod(m, struct ip6_hdr *); 809 ia6 = in6_ifawithifp(ifp, &ip6->ip6_src); 810 if (ia6) { 811 /* Record statistics for this interface address. */ 812 ia6->ia_ifa.ifa_data.ifad_outbytes += 813 m->m_pkthdr.len; 814 } 815 #endif 816 #ifdef IPSEC 817 /* clean ipsec history once it goes out of the node */ 818 ipsec_delaux(m); 819 #endif 820 error = nd6_output(ifp, origifp, m, dst, ro->ro_rt); 821 goto done; 822 } else if (mtu < IPV6_MMTU) { 823 /* 824 * note that path MTU is never less than IPV6_MMTU 825 * (see icmp6_input). 826 */ 827 error = EMSGSIZE; 828 in6_ifstat_inc(ifp, ifs6_out_fragfail); 829 goto bad; 830 } else if (ip6->ip6_plen == 0) { /* jumbo payload cannot be fragmented */ 831 error = EMSGSIZE; 832 in6_ifstat_inc(ifp, ifs6_out_fragfail); 833 goto bad; 834 } else { 835 struct mbuf **mnext, *m_frgpart; 836 struct ip6_frag *ip6f; 837 u_int32_t id = htonl(ip6_randomid()); 838 u_char nextproto; 839 840 /* 841 * Too large for the destination or interface; 842 * fragment if possible. 843 * Must be able to put at least 8 bytes per fragment. 844 */ 845 hlen = unfragpartlen; 846 if (mtu > IPV6_MAXPACKET) 847 mtu = IPV6_MAXPACKET; 848 len = (mtu - hlen - sizeof(struct ip6_frag)) & ~7; 849 if (len < 8) { 850 error = EMSGSIZE; 851 in6_ifstat_inc(ifp, ifs6_out_fragfail); 852 goto bad; 853 } 854 855 mnext = &m->m_nextpkt; 856 857 /* 858 * Change the next header field of the last header in the 859 * unfragmentable part. 860 */ 861 if (exthdrs.ip6e_rthdr) { 862 nextproto = *mtod(exthdrs.ip6e_rthdr, u_char *); 863 *mtod(exthdrs.ip6e_rthdr, u_char *) = IPPROTO_FRAGMENT; 864 } else if (exthdrs.ip6e_dest1) { 865 nextproto = *mtod(exthdrs.ip6e_dest1, u_char *); 866 *mtod(exthdrs.ip6e_dest1, u_char *) = IPPROTO_FRAGMENT; 867 } else if (exthdrs.ip6e_hbh) { 868 nextproto = *mtod(exthdrs.ip6e_hbh, u_char *); 869 *mtod(exthdrs.ip6e_hbh, u_char *) = IPPROTO_FRAGMENT; 870 } else { 871 nextproto = ip6->ip6_nxt; 872 ip6->ip6_nxt = IPPROTO_FRAGMENT; 873 } 874 875 /* 876 * Loop through length of segment after first fragment, 877 * make new header and copy data of each part and link onto 878 * chain. 879 */ 880 m0 = m; 881 for (off = hlen; off < tlen; off += len) { 882 struct mbuf *mlast; 883 884 MGETHDR(m, M_DONTWAIT, MT_HEADER); 885 if (!m) { 886 error = ENOBUFS; 887 ip6stat.ip6s_odropped++; 888 goto sendorfree; 889 } 890 m->m_flags = m0->m_flags & M_COPYFLAGS; 891 *mnext = m; 892 mnext = &m->m_nextpkt; 893 m->m_data += max_linkhdr; 894 mhip6 = mtod(m, struct ip6_hdr *); 895 *mhip6 = *ip6; 896 m->m_len = sizeof(*mhip6); 897 error = ip6_insertfraghdr(m0, m, hlen, &ip6f); 898 if (error) { 899 ip6stat.ip6s_odropped++; 900 goto sendorfree; 901 } 902 ip6f->ip6f_offlg = htons((u_int16_t)((off - hlen) & ~7)); 903 if (off + len >= tlen) 904 len = tlen - off; 905 else 906 ip6f->ip6f_offlg |= IP6F_MORE_FRAG; 907 mhip6->ip6_plen = htons((u_int16_t)(len + hlen + 908 sizeof(*ip6f) - sizeof(struct ip6_hdr))); 909 if ((m_frgpart = m_copy(m0, off, len)) == 0) { 910 error = ENOBUFS; 911 ip6stat.ip6s_odropped++; 912 goto sendorfree; 913 } 914 for (mlast = m; mlast->m_next; mlast = mlast->m_next) 915 ; 916 mlast->m_next = m_frgpart; 917 m->m_pkthdr.len = len + hlen + sizeof(*ip6f); 918 m->m_pkthdr.rcvif = (struct ifnet *)0; 919 ip6f->ip6f_reserved = 0; 920 ip6f->ip6f_ident = id; 921 ip6f->ip6f_nxt = nextproto; 922 ip6stat.ip6s_ofragments++; 923 in6_ifstat_inc(ifp, ifs6_out_fragcreat); 924 } 925 926 in6_ifstat_inc(ifp, ifs6_out_fragok); 927 } 928 929 /* 930 * Remove leading garbages. 931 */ 932 sendorfree: 933 m = m0->m_nextpkt; 934 m0->m_nextpkt = 0; 935 m_freem(m0); 936 for (m0 = m; m; m = m0) { 937 m0 = m->m_nextpkt; 938 m->m_nextpkt = 0; 939 if (error == 0) { 940 #ifdef IFA_STATS 941 struct in6_ifaddr *ia6; 942 ip6 = mtod(m, struct ip6_hdr *); 943 ia6 = in6_ifawithifp(ifp, &ip6->ip6_src); 944 if (ia6) { 945 /* 946 * Record statistics for this interface 947 * address. 948 */ 949 ia6->ia_ifa.ifa_data.ifad_outbytes += 950 m->m_pkthdr.len; 951 } 952 #endif 953 #ifdef IPSEC 954 /* clean ipsec history once it goes out of the node */ 955 ipsec_delaux(m); 956 #endif 957 error = nd6_output(ifp, origifp, m, dst, ro->ro_rt); 958 } else 959 m_freem(m); 960 } 961 962 if (error == 0) 963 ip6stat.ip6s_fragmented++; 964 965 done: 966 if (ro == &ip6route && ro->ro_rt) { /* brace necessary for RTFREE */ 967 RTFREE(ro->ro_rt); 968 } else if (ro_pmtu == &ip6route && ro_pmtu->ro_rt) { 969 RTFREE(ro_pmtu->ro_rt); 970 } 971 972 #ifdef IPSEC 973 if (sp != NULL) 974 key_freesp(sp); 975 #endif /* IPSEC */ 976 977 return (error); 978 979 freehdrs: 980 m_freem(exthdrs.ip6e_hbh); /* m_freem will check if mbuf is 0 */ 981 m_freem(exthdrs.ip6e_dest1); 982 m_freem(exthdrs.ip6e_rthdr); 983 m_freem(exthdrs.ip6e_dest2); 984 /* FALLTHROUGH */ 985 bad: 986 m_freem(m); 987 goto done; 988 } 989 990 static int 991 ip6_copyexthdr(mp, hdr, hlen) 992 struct mbuf **mp; 993 caddr_t hdr; 994 int hlen; 995 { 996 struct mbuf *m; 997 998 if (hlen > MCLBYTES) 999 return (ENOBUFS); /* XXX */ 1000 1001 MGET(m, M_DONTWAIT, MT_DATA); 1002 if (!m) 1003 return (ENOBUFS); 1004 1005 if (hlen > MLEN) { 1006 MCLGET(m, M_DONTWAIT); 1007 if ((m->m_flags & M_EXT) == 0) { 1008 m_free(m); 1009 return (ENOBUFS); 1010 } 1011 } 1012 m->m_len = hlen; 1013 if (hdr) 1014 bcopy(hdr, mtod(m, caddr_t), hlen); 1015 1016 *mp = m; 1017 return (0); 1018 } 1019 1020 /* 1021 * Insert jumbo payload option. 1022 */ 1023 static int 1024 ip6_insert_jumboopt(exthdrs, plen) 1025 struct ip6_exthdrs *exthdrs; 1026 u_int32_t plen; 1027 { 1028 struct mbuf *mopt; 1029 u_int8_t *optbuf; 1030 u_int32_t v; 1031 1032 #define JUMBOOPTLEN 8 /* length of jumbo payload option and padding */ 1033 1034 /* 1035 * If there is no hop-by-hop options header, allocate new one. 1036 * If there is one but it doesn't have enough space to store the 1037 * jumbo payload option, allocate a cluster to store the whole options. 1038 * Otherwise, use it to store the options. 1039 */ 1040 if (exthdrs->ip6e_hbh == 0) { 1041 MGET(mopt, M_DONTWAIT, MT_DATA); 1042 if (mopt == 0) 1043 return (ENOBUFS); 1044 mopt->m_len = JUMBOOPTLEN; 1045 optbuf = mtod(mopt, u_int8_t *); 1046 optbuf[1] = 0; /* = ((JUMBOOPTLEN) >> 3) - 1 */ 1047 exthdrs->ip6e_hbh = mopt; 1048 } else { 1049 struct ip6_hbh *hbh; 1050 1051 mopt = exthdrs->ip6e_hbh; 1052 if (M_TRAILINGSPACE(mopt) < JUMBOOPTLEN) { 1053 /* 1054 * XXX assumption: 1055 * - exthdrs->ip6e_hbh is not referenced from places 1056 * other than exthdrs. 1057 * - exthdrs->ip6e_hbh is not an mbuf chain. 1058 */ 1059 int oldoptlen = mopt->m_len; 1060 struct mbuf *n; 1061 1062 /* 1063 * XXX: give up if the whole (new) hbh header does 1064 * not fit even in an mbuf cluster. 1065 */ 1066 if (oldoptlen + JUMBOOPTLEN > MCLBYTES) 1067 return (ENOBUFS); 1068 1069 /* 1070 * As a consequence, we must always prepare a cluster 1071 * at this point. 1072 */ 1073 MGET(n, M_DONTWAIT, MT_DATA); 1074 if (n) { 1075 MCLGET(n, M_DONTWAIT); 1076 if ((n->m_flags & M_EXT) == 0) { 1077 m_freem(n); 1078 n = NULL; 1079 } 1080 } 1081 if (!n) 1082 return (ENOBUFS); 1083 n->m_len = oldoptlen + JUMBOOPTLEN; 1084 bcopy(mtod(mopt, caddr_t), mtod(n, caddr_t), 1085 oldoptlen); 1086 optbuf = mtod(n, u_int8_t *) + oldoptlen; 1087 m_freem(mopt); 1088 mopt = exthdrs->ip6e_hbh = n; 1089 } else { 1090 optbuf = mtod(mopt, u_int8_t *) + mopt->m_len; 1091 mopt->m_len += JUMBOOPTLEN; 1092 } 1093 optbuf[0] = IP6OPT_PADN; 1094 optbuf[1] = 0; 1095 1096 /* 1097 * Adjust the header length according to the pad and 1098 * the jumbo payload option. 1099 */ 1100 hbh = mtod(mopt, struct ip6_hbh *); 1101 hbh->ip6h_len += (JUMBOOPTLEN >> 3); 1102 } 1103 1104 /* fill in the option. */ 1105 optbuf[2] = IP6OPT_JUMBO; 1106 optbuf[3] = 4; 1107 v = (u_int32_t)htonl(plen + JUMBOOPTLEN); 1108 bcopy(&v, &optbuf[4], sizeof(u_int32_t)); 1109 1110 /* finally, adjust the packet header length */ 1111 exthdrs->ip6e_ip6->m_pkthdr.len += JUMBOOPTLEN; 1112 1113 return (0); 1114 #undef JUMBOOPTLEN 1115 } 1116 1117 /* 1118 * Insert fragment header and copy unfragmentable header portions. 1119 */ 1120 static int 1121 ip6_insertfraghdr(m0, m, hlen, frghdrp) 1122 struct mbuf *m0, *m; 1123 int hlen; 1124 struct ip6_frag **frghdrp; 1125 { 1126 struct mbuf *n, *mlast; 1127 1128 if (hlen > sizeof(struct ip6_hdr)) { 1129 n = m_copym(m0, sizeof(struct ip6_hdr), 1130 hlen - sizeof(struct ip6_hdr), M_DONTWAIT); 1131 if (n == 0) 1132 return (ENOBUFS); 1133 m->m_next = n; 1134 } else 1135 n = m; 1136 1137 /* Search for the last mbuf of unfragmentable part. */ 1138 for (mlast = n; mlast->m_next; mlast = mlast->m_next) 1139 ; 1140 1141 if ((mlast->m_flags & M_EXT) == 0 && 1142 M_TRAILINGSPACE(mlast) >= sizeof(struct ip6_frag)) { 1143 /* use the trailing space of the last mbuf for the fragment hdr */ 1144 *frghdrp = (struct ip6_frag *)(mtod(mlast, caddr_t) + 1145 mlast->m_len); 1146 mlast->m_len += sizeof(struct ip6_frag); 1147 m->m_pkthdr.len += sizeof(struct ip6_frag); 1148 } else { 1149 /* allocate a new mbuf for the fragment header */ 1150 struct mbuf *mfrg; 1151 1152 MGET(mfrg, M_DONTWAIT, MT_DATA); 1153 if (mfrg == 0) 1154 return (ENOBUFS); 1155 mfrg->m_len = sizeof(struct ip6_frag); 1156 *frghdrp = mtod(mfrg, struct ip6_frag *); 1157 mlast->m_next = mfrg; 1158 } 1159 1160 return (0); 1161 } 1162 1163 static int 1164 ip6_getpmtu(ro_pmtu, ro, ifp, dst, mtup) 1165 struct route_in6 *ro_pmtu, *ro; 1166 struct ifnet *ifp; 1167 struct in6_addr *dst; 1168 u_long *mtup; 1169 { 1170 u_int32_t mtu = 0; 1171 int error = 0; 1172 1173 if (ro_pmtu != ro) { 1174 /* The first hop and the final destination may differ. */ 1175 struct sockaddr_in6 *sa6_dst = 1176 (struct sockaddr_in6 *)&ro_pmtu->ro_dst; 1177 if (ro_pmtu->ro_rt && 1178 ((ro_pmtu->ro_rt->rt_flags & RTF_UP) == 0 || 1179 !IN6_ARE_ADDR_EQUAL(&sa6_dst->sin6_addr, dst))) { 1180 RTFREE(ro_pmtu->ro_rt); 1181 ro_pmtu->ro_rt = (struct rtentry *)0; 1182 } 1183 if (ro_pmtu->ro_rt == 0) { 1184 bzero(sa6_dst, sizeof(*sa6_dst)); 1185 sa6_dst->sin6_family = AF_INET6; 1186 sa6_dst->sin6_len = sizeof(struct sockaddr_in6); 1187 sa6_dst->sin6_addr = *dst; 1188 1189 rtalloc((struct route *)ro_pmtu); 1190 } 1191 } 1192 if (ro_pmtu->ro_rt) { 1193 u_int32_t ifmtu; 1194 1195 if (ifp == NULL) 1196 ifp = ro_pmtu->ro_rt->rt_ifp; 1197 ifmtu = IN6_LINKMTU(ifp); 1198 mtu = ro_pmtu->ro_rt->rt_rmx.rmx_mtu; 1199 if (mtu == 0) 1200 mtu = ifmtu; 1201 else if (mtu > ifmtu) { 1202 /* 1203 * The MTU on the route is larger than the MTU on 1204 * the interface! This shouldn't happen, unless the 1205 * MTU of the interface has been changed after the 1206 * interface was brought up. Change the MTU in the 1207 * route to match the interface MTU (as long as the 1208 * field isn't locked). 1209 * 1210 * if MTU on the route is 0, we need to fix the MTU. 1211 * this case happens with path MTU discovery timeouts. 1212 */ 1213 mtu = ifmtu; 1214 if (!(ro_pmtu->ro_rt->rt_rmx.rmx_locks & RTV_MTU)) 1215 ro_pmtu->ro_rt->rt_rmx.rmx_mtu = mtu; 1216 } 1217 } else if (ifp) { 1218 mtu = IN6_LINKMTU(ifp); 1219 } else 1220 error = EHOSTUNREACH; /* XXX */ 1221 1222 *mtup = mtu; 1223 return (error); 1224 } 1225 1226 /* 1227 * IP6 socket option processing. 1228 */ 1229 int 1230 ip6_ctloutput(op, so, level, optname, mp) 1231 int op; 1232 struct socket *so; 1233 int level, optname; 1234 struct mbuf **mp; 1235 { 1236 struct in6pcb *in6p = sotoin6pcb(so); 1237 struct mbuf *m = *mp; 1238 int optval = 0; 1239 int error = 0; 1240 struct proc *p = curproc; /* XXX */ 1241 1242 if (level == IPPROTO_IPV6) { 1243 switch (op) { 1244 case PRCO_SETOPT: 1245 switch (optname) { 1246 case IPV6_PKTOPTIONS: 1247 /* m is freed in ip6_pcbopts */ 1248 return (ip6_pcbopts(&in6p->in6p_outputopts, 1249 m, so)); 1250 case IPV6_HOPOPTS: 1251 case IPV6_DSTOPTS: 1252 if (p == 0 || suser(p->p_ucred, &p->p_acflag)) { 1253 error = EPERM; 1254 break; 1255 } 1256 /* FALLTHROUGH */ 1257 case IPV6_UNICAST_HOPS: 1258 case IPV6_RECVOPTS: 1259 case IPV6_RECVRETOPTS: 1260 case IPV6_RECVDSTADDR: 1261 case IPV6_PKTINFO: 1262 case IPV6_HOPLIMIT: 1263 case IPV6_RTHDR: 1264 case IPV6_FAITH: 1265 case IPV6_V6ONLY: 1266 if (!m || m->m_len != sizeof(int)) { 1267 error = EINVAL; 1268 break; 1269 } 1270 optval = *mtod(m, int *); 1271 switch (optname) { 1272 1273 case IPV6_UNICAST_HOPS: 1274 if (optval < -1 || optval >= 256) 1275 error = EINVAL; 1276 else { 1277 /* -1 = kernel default */ 1278 in6p->in6p_hops = optval; 1279 } 1280 break; 1281 #define OPTSET(bit) \ 1282 do { \ 1283 if (optval) \ 1284 in6p->in6p_flags |= (bit); \ 1285 else \ 1286 in6p->in6p_flags &= ~(bit); \ 1287 } while (/*CONSTCOND*/ 0) 1288 1289 case IPV6_RECVOPTS: 1290 OPTSET(IN6P_RECVOPTS); 1291 break; 1292 1293 case IPV6_RECVRETOPTS: 1294 OPTSET(IN6P_RECVRETOPTS); 1295 break; 1296 1297 case IPV6_RECVDSTADDR: 1298 OPTSET(IN6P_RECVDSTADDR); 1299 break; 1300 1301 case IPV6_PKTINFO: 1302 OPTSET(IN6P_PKTINFO); 1303 break; 1304 1305 case IPV6_HOPLIMIT: 1306 OPTSET(IN6P_HOPLIMIT); 1307 break; 1308 1309 case IPV6_HOPOPTS: 1310 OPTSET(IN6P_HOPOPTS); 1311 break; 1312 1313 case IPV6_DSTOPTS: 1314 OPTSET(IN6P_DSTOPTS); 1315 break; 1316 1317 case IPV6_RTHDR: 1318 OPTSET(IN6P_RTHDR); 1319 break; 1320 1321 case IPV6_FAITH: 1322 OPTSET(IN6P_FAITH); 1323 break; 1324 1325 case IPV6_V6ONLY: 1326 /* 1327 * make setsockopt(IPV6_V6ONLY) 1328 * available only prior to bind(2). 1329 * see ipng mailing list, Jun 22 2001. 1330 */ 1331 if (in6p->in6p_lport || 1332 !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 1333 error = EINVAL; 1334 break; 1335 } 1336 #ifdef INET6_BINDV6ONLY 1337 if (!optval) 1338 error = EINVAL; 1339 #else 1340 OPTSET(IN6P_IPV6_V6ONLY); 1341 #endif 1342 break; 1343 } 1344 break; 1345 #undef OPTSET 1346 1347 case IPV6_MULTICAST_IF: 1348 case IPV6_MULTICAST_HOPS: 1349 case IPV6_MULTICAST_LOOP: 1350 case IPV6_JOIN_GROUP: 1351 case IPV6_LEAVE_GROUP: 1352 error = ip6_setmoptions(optname, 1353 &in6p->in6p_moptions, m); 1354 break; 1355 1356 case IPV6_PORTRANGE: 1357 optval = *mtod(m, int *); 1358 1359 switch (optval) { 1360 case IPV6_PORTRANGE_DEFAULT: 1361 in6p->in6p_flags &= ~(IN6P_LOWPORT); 1362 in6p->in6p_flags &= ~(IN6P_HIGHPORT); 1363 break; 1364 1365 case IPV6_PORTRANGE_HIGH: 1366 in6p->in6p_flags &= ~(IN6P_LOWPORT); 1367 in6p->in6p_flags |= IN6P_HIGHPORT; 1368 break; 1369 1370 case IPV6_PORTRANGE_LOW: 1371 in6p->in6p_flags &= ~(IN6P_HIGHPORT); 1372 in6p->in6p_flags |= IN6P_LOWPORT; 1373 break; 1374 1375 default: 1376 error = EINVAL; 1377 break; 1378 } 1379 break; 1380 1381 #ifdef IPSEC 1382 case IPV6_IPSEC_POLICY: 1383 { 1384 caddr_t req = NULL; 1385 size_t len = 0; 1386 1387 int priv = 0; 1388 if (p == 0 || suser(p->p_ucred, &p->p_acflag)) 1389 priv = 0; 1390 else 1391 priv = 1; 1392 if (m) { 1393 req = mtod(m, caddr_t); 1394 len = m->m_len; 1395 } 1396 error = ipsec6_set_policy(in6p, 1397 optname, req, len, priv); 1398 } 1399 break; 1400 #endif /* IPSEC */ 1401 1402 default: 1403 error = ENOPROTOOPT; 1404 break; 1405 } 1406 if (m) 1407 (void)m_free(m); 1408 break; 1409 1410 case PRCO_GETOPT: 1411 switch (optname) { 1412 1413 case IPV6_OPTIONS: 1414 case IPV6_RETOPTS: 1415 error = ENOPROTOOPT; 1416 break; 1417 1418 case IPV6_PKTOPTIONS: 1419 if (in6p->in6p_options) { 1420 *mp = m_copym(in6p->in6p_options, 0, 1421 M_COPYALL, M_WAIT); 1422 } else { 1423 *mp = m_get(M_WAIT, MT_SOOPTS); 1424 (*mp)->m_len = 0; 1425 } 1426 break; 1427 1428 case IPV6_HOPOPTS: 1429 case IPV6_DSTOPTS: 1430 if (p == 0 || suser(p->p_ucred, &p->p_acflag)) { 1431 error = EPERM; 1432 break; 1433 } 1434 /* FALLTHROUGH */ 1435 case IPV6_UNICAST_HOPS: 1436 case IPV6_RECVOPTS: 1437 case IPV6_RECVRETOPTS: 1438 case IPV6_RECVDSTADDR: 1439 case IPV6_PORTRANGE: 1440 case IPV6_PKTINFO: 1441 case IPV6_HOPLIMIT: 1442 case IPV6_RTHDR: 1443 case IPV6_FAITH: 1444 case IPV6_V6ONLY: 1445 *mp = m = m_get(M_WAIT, MT_SOOPTS); 1446 m->m_len = sizeof(int); 1447 switch (optname) { 1448 1449 case IPV6_UNICAST_HOPS: 1450 optval = in6p->in6p_hops; 1451 break; 1452 1453 #define OPTBIT(bit) (in6p->in6p_flags & bit ? 1 : 0) 1454 1455 case IPV6_RECVOPTS: 1456 optval = OPTBIT(IN6P_RECVOPTS); 1457 break; 1458 1459 case IPV6_RECVRETOPTS: 1460 optval = OPTBIT(IN6P_RECVRETOPTS); 1461 break; 1462 1463 case IPV6_RECVDSTADDR: 1464 optval = OPTBIT(IN6P_RECVDSTADDR); 1465 break; 1466 1467 case IPV6_PORTRANGE: 1468 { 1469 int flags; 1470 flags = in6p->in6p_flags; 1471 if (flags & IN6P_HIGHPORT) 1472 optval = IPV6_PORTRANGE_HIGH; 1473 else if (flags & IN6P_LOWPORT) 1474 optval = IPV6_PORTRANGE_LOW; 1475 else 1476 optval = 0; 1477 break; 1478 } 1479 1480 case IPV6_PKTINFO: 1481 optval = OPTBIT(IN6P_PKTINFO); 1482 break; 1483 1484 case IPV6_HOPLIMIT: 1485 optval = OPTBIT(IN6P_HOPLIMIT); 1486 break; 1487 1488 case IPV6_HOPOPTS: 1489 optval = OPTBIT(IN6P_HOPOPTS); 1490 break; 1491 1492 case IPV6_DSTOPTS: 1493 optval = OPTBIT(IN6P_DSTOPTS); 1494 break; 1495 1496 case IPV6_RTHDR: 1497 optval = OPTBIT(IN6P_RTHDR); 1498 break; 1499 1500 case IPV6_FAITH: 1501 optval = OPTBIT(IN6P_FAITH); 1502 break; 1503 1504 case IPV6_V6ONLY: 1505 optval = OPTBIT(IN6P_IPV6_V6ONLY); 1506 break; 1507 } 1508 *mtod(m, int *) = optval; 1509 break; 1510 1511 case IPV6_MULTICAST_IF: 1512 case IPV6_MULTICAST_HOPS: 1513 case IPV6_MULTICAST_LOOP: 1514 case IPV6_JOIN_GROUP: 1515 case IPV6_LEAVE_GROUP: 1516 error = ip6_getmoptions(optname, in6p->in6p_moptions, mp); 1517 break; 1518 1519 #ifdef IPSEC 1520 case IPV6_IPSEC_POLICY: 1521 { 1522 caddr_t req = NULL; 1523 size_t len = 0; 1524 1525 if (m) { 1526 req = mtod(m, caddr_t); 1527 len = m->m_len; 1528 } 1529 error = ipsec6_get_policy(in6p, req, len, mp); 1530 break; 1531 } 1532 #endif /* IPSEC */ 1533 1534 default: 1535 error = ENOPROTOOPT; 1536 break; 1537 } 1538 break; 1539 } 1540 } else { 1541 error = EINVAL; 1542 if (op == PRCO_SETOPT && *mp) 1543 (void)m_free(*mp); 1544 } 1545 return (error); 1546 } 1547 1548 int 1549 ip6_raw_ctloutput(op, so, level, optname, mp) 1550 int op; 1551 struct socket *so; 1552 int level, optname; 1553 struct mbuf **mp; 1554 { 1555 int error = 0, optval, optlen; 1556 const int icmp6off = offsetof(struct icmp6_hdr, icmp6_cksum); 1557 struct in6pcb *in6p = sotoin6pcb(so); 1558 struct mbuf *m = *mp; 1559 1560 optlen = m ? m->m_len : 0; 1561 1562 if (level != IPPROTO_IPV6) { 1563 if (op == PRCO_SETOPT && *mp) 1564 (void)m_free(*mp); 1565 return (EINVAL); 1566 } 1567 1568 switch (optname) { 1569 case IPV6_CHECKSUM: 1570 /* 1571 * For ICMPv6 sockets, no modification allowed for checksum 1572 * offset, permit "no change" values to help existing apps. 1573 * 1574 * XXX 2292bis says: "An attempt to set IPV6_CHECKSUM 1575 * for an ICMPv6 socket will fail." 1576 * The current behavior does not meet 2292bis. 1577 */ 1578 switch (op) { 1579 case PRCO_SETOPT: 1580 if (optlen != sizeof(int)) { 1581 error = EINVAL; 1582 break; 1583 } 1584 optval = *mtod(m, int *); 1585 if ((optval % 2) != 0) { 1586 /* the API assumes even offset values */ 1587 error = EINVAL; 1588 } else if (so->so_proto->pr_protocol == 1589 IPPROTO_ICMPV6) { 1590 if (optval != icmp6off) 1591 error = EINVAL; 1592 } else 1593 in6p->in6p_cksum = optval; 1594 break; 1595 1596 case PRCO_GETOPT: 1597 if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) 1598 optval = icmp6off; 1599 else 1600 optval = in6p->in6p_cksum; 1601 1602 *mp = m = m_get(M_WAIT, MT_SOOPTS); 1603 m->m_len = sizeof(int); 1604 *mtod(m, int *) = optval; 1605 break; 1606 1607 default: 1608 error = EINVAL; 1609 break; 1610 } 1611 break; 1612 1613 default: 1614 error = ENOPROTOOPT; 1615 break; 1616 } 1617 1618 if (op == PRCO_SETOPT && m) 1619 (void)m_free(m); 1620 1621 return (error); 1622 } 1623 1624 /* 1625 * Set up IP6 options in pcb for insertion in output packets. 1626 * Store in mbuf with pointer in pcbopt, adding pseudo-option 1627 * with destination address if source routed. 1628 */ 1629 static int 1630 ip6_pcbopts(pktopt, m, so) 1631 struct ip6_pktopts **pktopt; 1632 struct mbuf *m; 1633 struct socket *so; 1634 { 1635 struct ip6_pktopts *opt = *pktopt; 1636 int error = 0; 1637 struct proc *p = curproc; /* XXX */ 1638 int priv = 0; 1639 1640 /* turn off any old options. */ 1641 if (opt) { 1642 if (opt->ip6po_m) 1643 (void)m_free(opt->ip6po_m); 1644 } else 1645 opt = malloc(sizeof(*opt), M_IP6OPT, M_WAITOK); 1646 *pktopt = 0; 1647 1648 if (!m || m->m_len == 0) { 1649 /* 1650 * Only turning off any previous options. 1651 */ 1652 free(opt, M_IP6OPT); 1653 if (m) 1654 (void)m_free(m); 1655 return (0); 1656 } 1657 1658 /* set options specified by user. */ 1659 if (p && !suser(p->p_ucred, &p->p_acflag)) 1660 priv = 1; 1661 if ((error = ip6_setpktoptions(m, opt, priv)) != 0) { 1662 (void)m_free(m); 1663 free(opt, M_IP6OPT); 1664 return (error); 1665 } 1666 *pktopt = opt; 1667 return (0); 1668 } 1669 1670 /* 1671 * Set the IP6 multicast options in response to user setsockopt(). 1672 */ 1673 static int 1674 ip6_setmoptions(optname, im6op, m) 1675 int optname; 1676 struct ip6_moptions **im6op; 1677 struct mbuf *m; 1678 { 1679 int error = 0; 1680 u_int loop, ifindex; 1681 struct ipv6_mreq *mreq; 1682 struct ifnet *ifp; 1683 struct ip6_moptions *im6o = *im6op; 1684 struct route_in6 ro; 1685 struct sockaddr_in6 *dst; 1686 struct in6_multi_mship *imm; 1687 struct proc *p = curproc; /* XXX */ 1688 1689 if (im6o == NULL) { 1690 /* 1691 * No multicast option buffer attached to the pcb; 1692 * allocate one and initialize to default values. 1693 */ 1694 im6o = (struct ip6_moptions *) 1695 malloc(sizeof(*im6o), M_IPMOPTS, M_WAITOK); 1696 1697 if (im6o == NULL) 1698 return (ENOBUFS); 1699 *im6op = im6o; 1700 im6o->im6o_multicast_ifp = NULL; 1701 im6o->im6o_multicast_hlim = ip6_defmcasthlim; 1702 im6o->im6o_multicast_loop = IPV6_DEFAULT_MULTICAST_LOOP; 1703 LIST_INIT(&im6o->im6o_memberships); 1704 } 1705 1706 switch (optname) { 1707 1708 case IPV6_MULTICAST_IF: 1709 /* 1710 * Select the interface for outgoing multicast packets. 1711 */ 1712 if (m == NULL || m->m_len != sizeof(u_int)) { 1713 error = EINVAL; 1714 break; 1715 } 1716 bcopy(mtod(m, u_int *), &ifindex, sizeof(ifindex)); 1717 if (ifindex < 0 || if_indexlim <= ifindex || 1718 !ifindex2ifnet[ifindex]) { 1719 error = ENXIO; /* XXX EINVAL? */ 1720 break; 1721 } 1722 ifp = ifindex2ifnet[ifindex]; 1723 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1724 error = EADDRNOTAVAIL; 1725 break; 1726 } 1727 im6o->im6o_multicast_ifp = ifp; 1728 break; 1729 1730 case IPV6_MULTICAST_HOPS: 1731 { 1732 /* 1733 * Set the IP6 hoplimit for outgoing multicast packets. 1734 */ 1735 int optval; 1736 if (m == NULL || m->m_len != sizeof(int)) { 1737 error = EINVAL; 1738 break; 1739 } 1740 bcopy(mtod(m, u_int *), &optval, sizeof(optval)); 1741 if (optval < -1 || optval >= 256) 1742 error = EINVAL; 1743 else if (optval == -1) 1744 im6o->im6o_multicast_hlim = ip6_defmcasthlim; 1745 else 1746 im6o->im6o_multicast_hlim = optval; 1747 break; 1748 } 1749 1750 case IPV6_MULTICAST_LOOP: 1751 /* 1752 * Set the loopback flag for outgoing multicast packets. 1753 * Must be zero or one. 1754 */ 1755 if (m == NULL || m->m_len != sizeof(u_int)) { 1756 error = EINVAL; 1757 break; 1758 } 1759 bcopy(mtod(m, u_int *), &loop, sizeof(loop)); 1760 if (loop > 1) { 1761 error = EINVAL; 1762 break; 1763 } 1764 im6o->im6o_multicast_loop = loop; 1765 break; 1766 1767 case IPV6_JOIN_GROUP: 1768 /* 1769 * Add a multicast group membership. 1770 * Group must be a valid IP6 multicast address. 1771 */ 1772 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) { 1773 error = EINVAL; 1774 break; 1775 } 1776 mreq = mtod(m, struct ipv6_mreq *); 1777 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) { 1778 /* 1779 * We use the unspecified address to specify to accept 1780 * all multicast addresses. Only super user is allowed 1781 * to do this. 1782 */ 1783 if (suser(p->p_ucred, &p->p_acflag)) 1784 { 1785 error = EACCES; 1786 break; 1787 } 1788 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) { 1789 error = EINVAL; 1790 break; 1791 } 1792 1793 /* 1794 * If the interface is specified, validate it. 1795 */ 1796 if (mreq->ipv6mr_interface < 0 || 1797 if_indexlim <= mreq->ipv6mr_interface || 1798 !ifindex2ifnet[mreq->ipv6mr_interface]) { 1799 error = ENXIO; /* XXX EINVAL? */ 1800 break; 1801 } 1802 /* 1803 * If no interface was explicitly specified, choose an 1804 * appropriate one according to the given multicast address. 1805 */ 1806 if (mreq->ipv6mr_interface == 0) { 1807 /* 1808 * If the multicast address is in node-local scope, 1809 * the interface should be a loopback interface. 1810 * Otherwise, look up the routing table for the 1811 * address, and choose the outgoing interface. 1812 * XXX: is it a good approach? 1813 */ 1814 if (IN6_IS_ADDR_MC_NODELOCAL(&mreq->ipv6mr_multiaddr)) { 1815 ifp = &loif[0]; 1816 } else { 1817 ro.ro_rt = NULL; 1818 dst = (struct sockaddr_in6 *)&ro.ro_dst; 1819 bzero(dst, sizeof(*dst)); 1820 dst->sin6_len = sizeof(struct sockaddr_in6); 1821 dst->sin6_family = AF_INET6; 1822 dst->sin6_addr = mreq->ipv6mr_multiaddr; 1823 rtalloc((struct route *)&ro); 1824 if (ro.ro_rt == NULL) { 1825 error = EADDRNOTAVAIL; 1826 break; 1827 } 1828 ifp = ro.ro_rt->rt_ifp; 1829 rtfree(ro.ro_rt); 1830 } 1831 } else 1832 ifp = ifindex2ifnet[mreq->ipv6mr_interface]; 1833 1834 /* 1835 * See if we found an interface, and confirm that it 1836 * supports multicast 1837 */ 1838 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) { 1839 error = EADDRNOTAVAIL; 1840 break; 1841 } 1842 /* 1843 * Put interface index into the multicast address, 1844 * if the address has link-local scope. 1845 */ 1846 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) { 1847 mreq->ipv6mr_multiaddr.s6_addr16[1] = 1848 htons(ifp->if_index); 1849 } 1850 /* 1851 * See if the membership already exists. 1852 */ 1853 for (imm = im6o->im6o_memberships.lh_first; 1854 imm != NULL; imm = imm->i6mm_chain.le_next) 1855 if (imm->i6mm_maddr->in6m_ifp == ifp && 1856 IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr, 1857 &mreq->ipv6mr_multiaddr)) 1858 break; 1859 if (imm != NULL) { 1860 error = EADDRINUSE; 1861 break; 1862 } 1863 /* 1864 * Everything looks good; add a new record to the multicast 1865 * address list for the given interface. 1866 */ 1867 imm = in6_joingroup(ifp, &mreq->ipv6mr_multiaddr, &error); 1868 if (!imm) 1869 break; 1870 LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain); 1871 break; 1872 1873 case IPV6_LEAVE_GROUP: 1874 /* 1875 * Drop a multicast group membership. 1876 * Group must be a valid IP6 multicast address. 1877 */ 1878 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) { 1879 error = EINVAL; 1880 break; 1881 } 1882 mreq = mtod(m, struct ipv6_mreq *); 1883 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) { 1884 if (suser(p->p_ucred, &p->p_acflag)) 1885 { 1886 error = EACCES; 1887 break; 1888 } 1889 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) { 1890 error = EINVAL; 1891 break; 1892 } 1893 /* 1894 * If an interface address was specified, get a pointer 1895 * to its ifnet structure. 1896 */ 1897 if (mreq->ipv6mr_interface < 0 || 1898 if_indexlim <= mreq->ipv6mr_interface || 1899 !ifindex2ifnet[mreq->ipv6mr_interface]) { 1900 error = ENXIO; /* XXX EINVAL? */ 1901 break; 1902 } 1903 ifp = ifindex2ifnet[mreq->ipv6mr_interface]; 1904 /* 1905 * Put interface index into the multicast address, 1906 * if the address has link-local scope. 1907 */ 1908 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) { 1909 mreq->ipv6mr_multiaddr.s6_addr16[1] = 1910 htons(mreq->ipv6mr_interface); 1911 } 1912 /* 1913 * Find the membership in the membership list. 1914 */ 1915 for (imm = im6o->im6o_memberships.lh_first; 1916 imm != NULL; imm = imm->i6mm_chain.le_next) { 1917 if ((ifp == NULL || imm->i6mm_maddr->in6m_ifp == ifp) && 1918 IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr, 1919 &mreq->ipv6mr_multiaddr)) 1920 break; 1921 } 1922 if (imm == NULL) { 1923 /* Unable to resolve interface */ 1924 error = EADDRNOTAVAIL; 1925 break; 1926 } 1927 /* 1928 * Give up the multicast address record to which the 1929 * membership points. 1930 */ 1931 LIST_REMOVE(imm, i6mm_chain); 1932 in6_leavegroup(imm); 1933 break; 1934 1935 default: 1936 error = EOPNOTSUPP; 1937 break; 1938 } 1939 1940 /* 1941 * If all options have default values, no need to keep the mbuf. 1942 */ 1943 if (im6o->im6o_multicast_ifp == NULL && 1944 im6o->im6o_multicast_hlim == ip6_defmcasthlim && 1945 im6o->im6o_multicast_loop == IPV6_DEFAULT_MULTICAST_LOOP && 1946 im6o->im6o_memberships.lh_first == NULL) { 1947 free(*im6op, M_IPMOPTS); 1948 *im6op = NULL; 1949 } 1950 1951 return (error); 1952 } 1953 1954 /* 1955 * Return the IP6 multicast options in response to user getsockopt(). 1956 */ 1957 static int 1958 ip6_getmoptions(optname, im6o, mp) 1959 int optname; 1960 struct ip6_moptions *im6o; 1961 struct mbuf **mp; 1962 { 1963 u_int *hlim, *loop, *ifindex; 1964 1965 *mp = m_get(M_WAIT, MT_SOOPTS); 1966 1967 switch (optname) { 1968 1969 case IPV6_MULTICAST_IF: 1970 ifindex = mtod(*mp, u_int *); 1971 (*mp)->m_len = sizeof(u_int); 1972 if (im6o == NULL || im6o->im6o_multicast_ifp == NULL) 1973 *ifindex = 0; 1974 else 1975 *ifindex = im6o->im6o_multicast_ifp->if_index; 1976 return (0); 1977 1978 case IPV6_MULTICAST_HOPS: 1979 hlim = mtod(*mp, u_int *); 1980 (*mp)->m_len = sizeof(u_int); 1981 if (im6o == NULL) 1982 *hlim = ip6_defmcasthlim; 1983 else 1984 *hlim = im6o->im6o_multicast_hlim; 1985 return (0); 1986 1987 case IPV6_MULTICAST_LOOP: 1988 loop = mtod(*mp, u_int *); 1989 (*mp)->m_len = sizeof(u_int); 1990 if (im6o == NULL) 1991 *loop = ip6_defmcasthlim; 1992 else 1993 *loop = im6o->im6o_multicast_loop; 1994 return (0); 1995 1996 default: 1997 return (EOPNOTSUPP); 1998 } 1999 } 2000 2001 /* 2002 * Discard the IP6 multicast options. 2003 */ 2004 void 2005 ip6_freemoptions(im6o) 2006 struct ip6_moptions *im6o; 2007 { 2008 struct in6_multi_mship *imm; 2009 2010 if (im6o == NULL) 2011 return; 2012 2013 while ((imm = im6o->im6o_memberships.lh_first) != NULL) { 2014 LIST_REMOVE(imm, i6mm_chain); 2015 in6_leavegroup(imm); 2016 } 2017 free(im6o, M_IPMOPTS); 2018 } 2019 2020 /* 2021 * Set IPv6 outgoing packet options based on advanced API. 2022 */ 2023 int 2024 ip6_setpktoptions(control, opt, priv) 2025 struct mbuf *control; 2026 struct ip6_pktopts *opt; 2027 int priv; 2028 { 2029 struct cmsghdr *cm = 0; 2030 2031 if (control == 0 || opt == 0) 2032 return (EINVAL); 2033 2034 bzero(opt, sizeof(*opt)); 2035 opt->ip6po_hlim = -1; /* -1 means to use default hop limit */ 2036 2037 /* 2038 * XXX: Currently, we assume all the optional information is stored 2039 * in a single mbuf. 2040 */ 2041 if (control->m_next) 2042 return (EINVAL); 2043 2044 opt->ip6po_m = control; 2045 2046 for (; control->m_len; control->m_data += CMSG_ALIGN(cm->cmsg_len), 2047 control->m_len -= CMSG_ALIGN(cm->cmsg_len)) { 2048 cm = mtod(control, struct cmsghdr *); 2049 if (cm->cmsg_len == 0 || cm->cmsg_len > control->m_len) 2050 return (EINVAL); 2051 if (cm->cmsg_level != IPPROTO_IPV6) 2052 continue; 2053 2054 switch (cm->cmsg_type) { 2055 case IPV6_PKTINFO: 2056 if (cm->cmsg_len != CMSG_LEN(sizeof(struct in6_pktinfo))) 2057 return (EINVAL); 2058 opt->ip6po_pktinfo = (struct in6_pktinfo *)CMSG_DATA(cm); 2059 if (opt->ip6po_pktinfo->ipi6_ifindex && 2060 IN6_IS_ADDR_LINKLOCAL(&opt->ip6po_pktinfo->ipi6_addr)) 2061 opt->ip6po_pktinfo->ipi6_addr.s6_addr16[1] = 2062 htons(opt->ip6po_pktinfo->ipi6_ifindex); 2063 2064 if (opt->ip6po_pktinfo->ipi6_ifindex >= if_indexlim || 2065 opt->ip6po_pktinfo->ipi6_ifindex < 0) 2066 return (ENXIO); 2067 if (opt->ip6po_pktinfo->ipi6_ifindex > 0 && 2068 !ifindex2ifnet[opt->ip6po_pktinfo->ipi6_ifindex]) 2069 return (ENXIO); 2070 2071 /* 2072 * Check if the requested source address is indeed a 2073 * unicast address assigned to the node, and can be 2074 * used as the packet's source address. 2075 */ 2076 if (!IN6_IS_ADDR_UNSPECIFIED(&opt->ip6po_pktinfo->ipi6_addr)) { 2077 struct ifaddr *ia; 2078 struct in6_ifaddr *ia6; 2079 struct sockaddr_in6 sin6; 2080 2081 bzero(&sin6, sizeof(sin6)); 2082 sin6.sin6_len = sizeof(sin6); 2083 sin6.sin6_family = AF_INET6; 2084 sin6.sin6_addr = 2085 opt->ip6po_pktinfo->ipi6_addr; 2086 ia = ifa_ifwithaddr(sin6tosa(&sin6)); 2087 if (ia == NULL || 2088 (opt->ip6po_pktinfo->ipi6_ifindex && 2089 (ia->ifa_ifp->if_index != 2090 opt->ip6po_pktinfo->ipi6_ifindex))) { 2091 return (EADDRNOTAVAIL); 2092 } 2093 ia6 = (struct in6_ifaddr *)ia; 2094 if ((ia6->ia6_flags & (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY)) != 0) { 2095 return (EADDRNOTAVAIL); 2096 } 2097 2098 /* 2099 * Check if the requested source address is 2100 * indeed a unicast address assigned to the 2101 * node. 2102 */ 2103 if (IN6_IS_ADDR_MULTICAST(&opt->ip6po_pktinfo->ipi6_addr)) 2104 return (EADDRNOTAVAIL); 2105 } 2106 break; 2107 2108 case IPV6_HOPLIMIT: 2109 if (cm->cmsg_len != CMSG_LEN(sizeof(int))) 2110 return (EINVAL); 2111 else { 2112 int t; 2113 2114 bcopy(CMSG_DATA(cm), &t, sizeof(t)); 2115 if (t < -1 || t > 255) 2116 return (EINVAL); 2117 opt->ip6po_hlim = t; 2118 } 2119 break; 2120 2121 case IPV6_NEXTHOP: 2122 if (!priv) 2123 return (EPERM); 2124 2125 /* check if cmsg_len is large enough for sa_len */ 2126 if (cm->cmsg_len < sizeof(u_char) || 2127 cm->cmsg_len < CMSG_LEN(*CMSG_DATA(cm))) 2128 return (EINVAL); 2129 2130 opt->ip6po_nexthop = (struct sockaddr *)CMSG_DATA(cm); 2131 2132 break; 2133 2134 case IPV6_HOPOPTS: 2135 if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_hbh))) 2136 return (EINVAL); 2137 else { 2138 struct ip6_hbh *t; 2139 2140 t = (struct ip6_hbh *)CMSG_DATA(cm); 2141 if (cm->cmsg_len != 2142 CMSG_LEN((t->ip6h_len + 1) << 3)) 2143 return (EINVAL); 2144 opt->ip6po_hbh = t; 2145 } 2146 break; 2147 2148 case IPV6_DSTOPTS: 2149 if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_dest))) 2150 return (EINVAL); 2151 2152 /* 2153 * If there is no routing header yet, the destination 2154 * options header should be put on the 1st part. 2155 * Otherwise, the header should be on the 2nd part. 2156 * (See RFC 2460, section 4.1) 2157 */ 2158 if (opt->ip6po_rthdr == NULL) { 2159 struct ip6_dest *t; 2160 2161 t = (struct ip6_dest *)CMSG_DATA(cm); 2162 if (cm->cmsg_len != 2163 CMSG_LEN((t->ip6d_len + 1) << 3)); 2164 return (EINVAL); 2165 opt->ip6po_dest1 = t; 2166 } 2167 else { 2168 struct ip6_dest *t; 2169 2170 t = (struct ip6_dest *)CMSG_DATA(cm); 2171 if (cm->cmsg_len != 2172 CMSG_LEN((opt->ip6po_dest2->ip6d_len + 1) << 3)) 2173 return (EINVAL); 2174 opt->ip6po_dest2 = t; 2175 } 2176 break; 2177 2178 case IPV6_RTHDR: 2179 if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_rthdr))) 2180 return (EINVAL); 2181 else { 2182 struct ip6_rthdr *t; 2183 2184 t = (struct ip6_rthdr *)CMSG_DATA(cm); 2185 if (cm->cmsg_len != 2186 CMSG_LEN((t->ip6r_len + 1) << 3)) 2187 return (EINVAL); 2188 switch (t->ip6r_type) { 2189 case IPV6_RTHDR_TYPE_0: 2190 if (t->ip6r_segleft == 0) 2191 return (EINVAL); 2192 break; 2193 default: 2194 return (EINVAL); 2195 } 2196 opt->ip6po_rthdr = t; 2197 } 2198 break; 2199 2200 default: 2201 return (ENOPROTOOPT); 2202 } 2203 } 2204 2205 return (0); 2206 } 2207 2208 /* 2209 * Routine called from ip6_output() to loop back a copy of an IP6 multicast 2210 * packet to the input queue of a specified interface. Note that this 2211 * calls the output routine of the loopback "driver", but with an interface 2212 * pointer that might NOT be &loif -- easier than replicating that code here. 2213 */ 2214 void 2215 ip6_mloopback(ifp, m, dst) 2216 struct ifnet *ifp; 2217 struct mbuf *m; 2218 struct sockaddr_in6 *dst; 2219 { 2220 struct mbuf *copym; 2221 struct ip6_hdr *ip6; 2222 2223 copym = m_copy(m, 0, M_COPYALL); 2224 if (copym == NULL) 2225 return; 2226 2227 /* 2228 * Make sure to deep-copy IPv6 header portion in case the data 2229 * is in an mbuf cluster, so that we can safely override the IPv6 2230 * header portion later. 2231 */ 2232 if ((copym->m_flags & M_EXT) != 0 || 2233 copym->m_len < sizeof(struct ip6_hdr)) { 2234 copym = m_pullup(copym, sizeof(struct ip6_hdr)); 2235 if (copym == NULL) 2236 return; 2237 } 2238 2239 #ifdef DIAGNOSTIC 2240 if (copym->m_len < sizeof(*ip6)) { 2241 m_freem(copym); 2242 return; 2243 } 2244 #endif 2245 2246 ip6 = mtod(copym, struct ip6_hdr *); 2247 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 2248 ip6->ip6_src.s6_addr16[1] = 0; 2249 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 2250 ip6->ip6_dst.s6_addr16[1] = 0; 2251 2252 (void)looutput(ifp, copym, (struct sockaddr *)dst, NULL); 2253 } 2254 2255 /* 2256 * Chop IPv6 header off from the payload. 2257 */ 2258 static int 2259 ip6_splithdr(m, exthdrs) 2260 struct mbuf *m; 2261 struct ip6_exthdrs *exthdrs; 2262 { 2263 struct mbuf *mh; 2264 struct ip6_hdr *ip6; 2265 2266 ip6 = mtod(m, struct ip6_hdr *); 2267 if (m->m_len > sizeof(*ip6)) { 2268 MGETHDR(mh, M_DONTWAIT, MT_HEADER); 2269 if (mh == 0) { 2270 m_freem(m); 2271 return ENOBUFS; 2272 } 2273 M_COPY_PKTHDR(mh, m); 2274 MH_ALIGN(mh, sizeof(*ip6)); 2275 m_tag_delete_chain(m, NULL); 2276 m->m_flags &= ~M_PKTHDR; 2277 m->m_len -= sizeof(*ip6); 2278 m->m_data += sizeof(*ip6); 2279 mh->m_next = m; 2280 m = mh; 2281 m->m_len = sizeof(*ip6); 2282 bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(*ip6)); 2283 } 2284 exthdrs->ip6e_ip6 = m; 2285 return 0; 2286 } 2287 2288 /* 2289 * Compute IPv6 extension header length. 2290 */ 2291 int 2292 ip6_optlen(in6p) 2293 struct in6pcb *in6p; 2294 { 2295 int len; 2296 2297 if (!in6p->in6p_outputopts) 2298 return 0; 2299 2300 len = 0; 2301 #define elen(x) \ 2302 (((struct ip6_ext *)(x)) ? (((struct ip6_ext *)(x))->ip6e_len + 1) << 3 : 0) 2303 2304 len += elen(in6p->in6p_outputopts->ip6po_hbh); 2305 len += elen(in6p->in6p_outputopts->ip6po_dest1); 2306 len += elen(in6p->in6p_outputopts->ip6po_rthdr); 2307 len += elen(in6p->in6p_outputopts->ip6po_dest2); 2308 return len; 2309 #undef elen 2310 } 2311
Indexes created Sun Sep 21 20:09:37 GMT 2025