1 /* $FreeBSD: src/sys/netinet6/in6_pcb.c,v 1.10.2.9 2003/01/24 05:11:35 sam Exp $ */ 2 /* $DragonFly: src/sys/netinet6/in6_pcb.c,v 1.34 2008/01/05 14:02:40 swildner Exp $ */ 3 /* $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ */ 4 5 /* 6 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. Neither the name of the project 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 PROJECT 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 PROJECT 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 */ 34 35 /* 36 * Copyright (c) 1982, 1986, 1991, 1993 37 * The Regents of the University of California. All rights reserved. 38 * 39 * Redistribution and use in source and binary forms, with or without 40 * modification, are permitted provided that the following conditions 41 * are met: 42 * 1. Redistributions of source code must retain the above copyright 43 * notice, this list of conditions and the following disclaimer. 44 * 2. Redistributions in binary form must reproduce the above copyright 45 * notice, this list of conditions and the following disclaimer in the 46 * documentation and/or other materials provided with the distribution. 47 * 3. All advertising materials mentioning features or use of this software 48 * must display the following acknowledgement: 49 * This product includes software developed by the University of 50 * California, Berkeley and its contributors. 51 * 4. Neither the name of the University nor the names of its contributors 52 * may be used to endorse or promote products derived from this software 53 * without specific prior written permission. 54 * 55 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 58 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 65 * SUCH DAMAGE. 66 * 67 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 68 */ 69 70 #include "opt_inet.h" 71 #include "opt_inet6.h" 72 #include "opt_ipsec.h" 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/malloc.h> 77 #include <sys/mbuf.h> 78 #include <sys/domain.h> 79 #include <sys/protosw.h> 80 #include <sys/socket.h> 81 #include <sys/socketvar.h> 82 #include <sys/sockio.h> 83 #include <sys/errno.h> 84 #include <sys/time.h> 85 #include <sys/proc.h> 86 #include <sys/jail.h> 87 #include <sys/thread2.h> 88 89 #include <vm/vm_zone.h> 90 91 #include <net/if.h> 92 #include <net/if_types.h> 93 #include <net/route.h> 94 95 #include <netinet/in.h> 96 #include <netinet/in_var.h> 97 #include <netinet/in_systm.h> 98 #include <netinet/ip6.h> 99 #include <netinet/ip_var.h> 100 #include <netinet6/ip6_var.h> 101 #include <netinet6/nd6.h> 102 #include <netinet/in_pcb.h> 103 #include <netinet6/in6_pcb.h> 104 105 #ifdef IPSEC 106 #include <netinet6/ipsec.h> 107 #ifdef INET6 108 #include <netinet6/ipsec6.h> 109 #endif 110 #include <netinet6/ah.h> 111 #ifdef INET6 112 #include <netinet6/ah6.h> 113 #endif 114 #include <netproto/key/key.h> 115 #endif /* IPSEC */ 116 117 #ifdef FAST_IPSEC 118 #include <netproto/ipsec/ipsec.h> 119 #include <netproto/ipsec/ipsec6.h> 120 #include <netproto/ipsec/key.h> 121 #define IPSEC 122 #endif /* FAST_IPSEC */ 123 124 struct in6_addr zeroin6_addr; 125 126 int 127 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 128 { 129 struct socket *so = inp->inp_socket; 130 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 131 struct sockaddr_in6 jsin6; 132 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 133 struct proc *p = td->td_proc; 134 struct ucred *cred = NULL; 135 u_short lport = 0; 136 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 137 138 if (!in6_ifaddr) /* XXX broken! */ 139 return (EADDRNOTAVAIL); 140 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 141 return (EINVAL); 142 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 143 wild = 1; 144 if (p) 145 cred = p->p_ucred; 146 if (nam) { 147 sin6 = (struct sockaddr_in6 *)nam; 148 if (nam->sa_len != sizeof(*sin6)) 149 return (EINVAL); 150 /* 151 * family check. 152 */ 153 if (nam->sa_family != AF_INET6) 154 return (EAFNOSUPPORT); 155 156 if (!prison_replace_wildcards(td, nam)) 157 return (EINVAL); 158 159 /* KAME hack: embed scopeid */ 160 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, NULL) != 0) 161 return EINVAL; 162 /* this must be cleared for ifa_ifwithaddr() */ 163 sin6->sin6_scope_id = 0; 164 165 lport = sin6->sin6_port; 166 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 167 /* 168 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 169 * allow compepte duplication of binding if 170 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 171 * and a multicast address is bound on both 172 * new and duplicated sockets. 173 */ 174 if (so->so_options & SO_REUSEADDR) 175 reuseport = SO_REUSEADDR|SO_REUSEPORT; 176 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 177 struct ifaddr *ia = NULL; 178 179 sin6->sin6_port = 0; /* yech... */ 180 if (!prison_replace_wildcards(td, (struct sockaddr *)sin6)) { 181 sin6->sin6_addr = kin6addr_any; 182 return(EINVAL); 183 } 184 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 185 return (EADDRNOTAVAIL); 186 187 /* 188 * XXX: bind to an anycast address might accidentally 189 * cause sending a packet with anycast source address. 190 * We should allow to bind to a deprecated address, since 191 * the application dare to use it. 192 */ 193 if (ia && 194 ((struct in6_ifaddr *)ia)->ia6_flags & 195 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 196 return (EADDRNOTAVAIL); 197 } 198 } 199 if (lport) { 200 struct inpcb *t; 201 202 /* GROSS */ 203 if (ntohs(lport) < IPV6PORT_RESERVED && cred && 204 suser_cred(cred, PRISON_ROOT)) 205 return (EACCES); 206 if (so->so_cred->cr_uid != 0 && 207 !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 208 t = in6_pcblookup_local(pcbinfo, 209 &sin6->sin6_addr, lport, 210 INPLOOKUP_WILDCARD, cred); 211 if (t && 212 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 213 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 214 (t->inp_socket->so_options & 215 SO_REUSEPORT) == 0) && 216 (so->so_cred->cr_uid != 217 t->inp_socket->so_cred->cr_uid)) 218 return (EADDRINUSE); 219 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 220 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 221 struct sockaddr_in sin; 222 223 in6_sin6_2_sin(&sin, sin6); 224 t = in_pcblookup_local(pcbinfo, 225 sin.sin_addr, lport, 226 INPLOOKUP_WILDCARD, cred); 227 if (t && 228 (so->so_cred->cr_uid != 229 t->inp_socket->so_cred->cr_uid) && 230 (ntohl(t->inp_laddr.s_addr) != 231 INADDR_ANY || 232 INP_SOCKAF(so) == 233 INP_SOCKAF(t->inp_socket))) 234 return (EADDRINUSE); 235 } 236 } 237 if (cred && cred->cr_prison && 238 !prison_replace_wildcards(td, nam)) 239 return(EADDRNOTAVAIL); 240 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 241 lport, wild, cred); 242 if (t && (reuseport & t->inp_socket->so_options) == 0) 243 return (EADDRINUSE); 244 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 245 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 246 struct sockaddr_in sin; 247 248 in6_sin6_2_sin(&sin, sin6); 249 t = in_pcblookup_local(pcbinfo, 250 sin.sin_addr, lport, 251 wild, cred); 252 if (t && 253 (reuseport & t->inp_socket->so_options) 254 == 0 && 255 (ntohl(t->inp_laddr.s_addr) 256 != INADDR_ANY || 257 INP_SOCKAF(so) == 258 INP_SOCKAF(t->inp_socket))) 259 return (EADDRINUSE); 260 } 261 } 262 inp->in6p_laddr = sin6->sin6_addr; 263 } 264 if (lport == 0) { 265 int e; 266 267 jsin6.sin6_addr = inp->in6p_laddr; 268 jsin6.sin6_family = AF_INET6; 269 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin6)) { 270 inp->in6p_laddr = kin6addr_any; 271 inp->inp_lport = 0; 272 return (EINVAL); 273 } 274 275 if ((e = in6_pcbsetport(&inp->in6p_laddr, inp, td)) != 0) 276 return (e); 277 } 278 else { 279 inp->inp_lport = lport; 280 if (in_pcbinsporthash(inp) != 0) { 281 inp->in6p_laddr = kin6addr_any; 282 inp->inp_lport = 0; 283 return (EAGAIN); 284 } 285 } 286 return (0); 287 } 288 289 /* 290 * Transform old in6_pcbconnect() into an inner subroutine for new 291 * in6_pcbconnect(): Do some validity-checking on the remote 292 * address (in mbuf 'nam') and then determine local host address 293 * (i.e., which interface) to use to access that remote host. 294 * 295 * This preserves definition of in6_pcbconnect(), while supporting a 296 * slightly different version for T/TCP. (This is more than 297 * a bit of a kludge, but cleaning up the internal interfaces would 298 * have forced minor changes in every protocol). 299 */ 300 301 int 302 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam, 303 struct in6_addr **plocal_addr6, struct thread *td) 304 { 305 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 306 struct ifnet *ifp = NULL; 307 int error = 0; 308 309 if (nam->sa_len != sizeof (*sin6)) 310 return (EINVAL); 311 if (sin6->sin6_family != AF_INET6) 312 return (EAFNOSUPPORT); 313 if (sin6->sin6_port == 0) 314 return (EADDRNOTAVAIL); 315 316 /* KAME hack: embed scopeid */ 317 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, &ifp) != 0) 318 return EINVAL; 319 320 if (in6_ifaddr) { 321 /* 322 * If the destination address is UNSPECIFIED addr, 323 * use the loopback addr, e.g ::1. 324 */ 325 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 326 sin6->sin6_addr = kin6addr_loopback; 327 } 328 { 329 /* 330 * XXX: in6_selectsrc might replace the bound local address 331 * with the address specified by setsockopt(IPV6_PKTINFO). 332 * Is it the intended behavior? 333 */ 334 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts, 335 inp->in6p_moptions, 336 &inp->in6p_route, 337 &inp->in6p_laddr, &error, td); 338 if (*plocal_addr6 == 0) { 339 if (error == 0) 340 error = EADDRNOTAVAIL; 341 return (error); 342 } 343 /* 344 * Don't do pcblookup call here; return interface in 345 * plocal_addr6 346 * and exit to caller, that will do the lookup. 347 */ 348 } 349 350 if (inp->in6p_route.ro_rt) 351 ifp = inp->in6p_route.ro_rt->rt_ifp; 352 353 return (0); 354 } 355 356 /* 357 * Outer subroutine: 358 * Connect from a socket to a specified address. 359 * Both address and port must be specified in argument sin. 360 * If don't have a local address for this socket yet, 361 * then pick one. 362 */ 363 int 364 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 365 { 366 struct in6_addr *addr6; 367 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 368 int error; 369 370 /* 371 * Call inner routine, to assign local interface address. 372 * in6_pcbladdr() may automatically fill in sin6_scope_id. 373 */ 374 if ((error = in6_pcbladdr(inp, nam, &addr6, td)) != 0) 375 return (error); 376 377 if (in6_pcblookup_hash(inp->inp_cpcbinfo, &sin6->sin6_addr, 378 sin6->sin6_port, 379 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 380 ? addr6 : &inp->in6p_laddr, 381 inp->inp_lport, 0, NULL) != NULL) { 382 return (EADDRINUSE); 383 } 384 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 385 if (inp->inp_lport == 0) { 386 error = in6_pcbbind(inp, (struct sockaddr *)0, td); 387 if (error) 388 return (error); 389 } 390 inp->in6p_laddr = *addr6; 391 } 392 inp->in6p_faddr = sin6->sin6_addr; 393 inp->inp_fport = sin6->sin6_port; 394 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 395 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 396 if (inp->in6p_flags & IN6P_AUTOFLOWLABEL) 397 inp->in6p_flowinfo |= 398 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 399 400 in_pcbinsconnhash(inp); 401 return (0); 402 } 403 404 #if 0 405 /* 406 * Return an IPv6 address, which is the most appropriate for given 407 * destination and user specified options. 408 * If necessary, this function lookups the routing table and return 409 * an entry to the caller for later use. 410 */ 411 struct in6_addr * 412 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 413 struct ip6_moptions *mopts, struct route_in6 *ro, 414 struct in6_addr *laddr, int *errorp, struct thread *td) 415 { 416 struct sockaddr_in6 jsin6; 417 struct ucred *cred = NULL; 418 struct in6_addr *dst; 419 struct in6_ifaddr *ia6 = 0; 420 struct in6_pktinfo *pi = NULL; 421 int jailed = 0; 422 423 if (td && td->td_proc && td->td_proc->p_ucred) 424 cred = td->td_proc->p_ucred; 425 if (cred && cred->cr_prison) 426 jailed = 1; 427 jsin6.sin6_family = AF_INET6; 428 dst = &dstsock->sin6_addr; 429 *errorp = 0; 430 431 /* 432 * If the source address is explicitly specified by the caller, 433 * use it. 434 */ 435 if (opts && (pi = opts->ip6po_pktinfo) && 436 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) { 437 jsin6.sin6_addr = pi->ipi6_addr; 438 if (jailed && !jailed_ip(cred->cr_prison, 439 (struct sockaddr *)&jsin6)) { 440 return(0); 441 } else { 442 return (&pi->ipi6_addr); 443 } 444 } 445 446 /* 447 * If the source address is not specified but the socket(if any) 448 * is already bound, use the bound address. 449 */ 450 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) { 451 jsin6.sin6_addr = *laddr; 452 if (jailed && !jailed_ip(cred->cr_prison, 453 (struct sockaddr *)&jsin6)) { 454 return(0); 455 } else { 456 return (laddr); 457 } 458 } 459 460 /* 461 * If the caller doesn't specify the source address but 462 * the outgoing interface, use an address associated with 463 * the interface. 464 */ 465 if (pi && pi->ipi6_ifindex) { 466 /* XXX boundary check is assumed to be already done. */ 467 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex], 468 dst); 469 470 if (ia6 && jailed) { 471 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 472 if (!jailed_ip(cred->cr_prison, 473 (struct sockaddr *)&jsin6)) 474 ia6 = 0; 475 } 476 477 if (ia6 == 0) { 478 *errorp = EADDRNOTAVAIL; 479 return (0); 480 } 481 return (&satosin6(&ia6->ia_addr)->sin6_addr); 482 } 483 484 /* 485 * If the destination address is a link-local unicast address or 486 * a multicast address, and if the outgoing interface is specified 487 * by the sin6_scope_id filed, use an address associated with the 488 * interface. 489 * XXX: We're now trying to define more specific semantics of 490 * sin6_scope_id field, so this part will be rewritten in 491 * the near future. 492 */ 493 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) && 494 dstsock->sin6_scope_id) { 495 /* 496 * I'm not sure if boundary check for scope_id is done 497 * somewhere... 498 */ 499 if (dstsock->sin6_scope_id < 0 || 500 if_index < dstsock->sin6_scope_id) { 501 *errorp = ENXIO; /* XXX: better error? */ 502 return (0); 503 } 504 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id], 505 dst); 506 507 if (ia6 && jailed) { 508 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 509 if (!jailed_ip(cred->cr_prison, 510 (struct sockaddr *)&jsin6)) 511 ia6 = 0; 512 } 513 514 if (ia6 == 0) { 515 *errorp = EADDRNOTAVAIL; 516 return (0); 517 } 518 return (&satosin6(&ia6->ia_addr)->sin6_addr); 519 } 520 521 /* 522 * If the destination address is a multicast address and 523 * the outgoing interface for the address is specified 524 * by the caller, use an address associated with the interface. 525 * There is a sanity check here; if the destination has node-local 526 * scope, the outgoing interfacde should be a loopback address. 527 * Even if the outgoing interface is not specified, we also 528 * choose a loopback interface as the outgoing interface. 529 */ 530 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) { 531 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL; 532 533 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) { 534 ifp = &loif[0]; 535 } 536 537 if (ifp) { 538 ia6 = in6_ifawithscope(ifp, dst); 539 if (ia6 == 0) { 540 *errorp = EADDRNOTAVAIL; 541 return (0); 542 } 543 return (&ia6->ia_addr.sin6_addr); 544 } 545 } 546 547 /* 548 * If the next hop address for the packet is specified 549 * by caller, use an address associated with the route 550 * to the next hop. 551 */ 552 { 553 struct sockaddr_in6 *sin6_next; 554 struct rtentry *rt; 555 556 if (opts && opts->ip6po_nexthop) { 557 sin6_next = satosin6(opts->ip6po_nexthop); 558 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL); 559 if (rt) { 560 ia6 = in6_ifawithscope(rt->rt_ifp, dst); 561 if (ia6 == 0) 562 ia6 = ifatoia6(rt->rt_ifa); 563 } 564 if (ia6 && jailed) { 565 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 566 if (!jailed_ip(cred->cr_prison, 567 (struct sockaddr *)&jsin6)) 568 ia6 = 0; 569 } 570 571 if (ia6 == 0) { 572 *errorp = EADDRNOTAVAIL; 573 return (0); 574 } 575 return (&satosin6(&ia6->ia_addr)->sin6_addr); 576 } 577 } 578 579 /* 580 * If route is known or can be allocated now, 581 * our src addr is taken from the i/f, else punt. 582 */ 583 if (ro) { 584 if (ro->ro_rt && 585 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) { 586 RTFREE(ro->ro_rt); 587 ro->ro_rt = NULL; 588 } 589 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) { 590 struct sockaddr_in6 *dst6; 591 592 /* No route yet, so try to acquire one */ 593 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 594 dst6 = &ro->ro_dst; 595 dst6->sin6_family = AF_INET6; 596 dst6->sin6_len = sizeof(struct sockaddr_in6); 597 dst6->sin6_addr = *dst; 598 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) { 599 ro->ro_rt = 600 rtpurelookup((struct sockaddr *)&ro->ro_dst); 601 } else { 602 rtalloc((struct route *)ro); 603 } 604 } 605 606 /* 607 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 608 * the address. But we don't know why it does so. 609 * It is necessary to ensure the scope even for lo0 610 * so doesn't check out IFF_LOOPBACK. 611 */ 612 613 if (ro->ro_rt) { 614 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst); 615 if (ia6 && jailed) { 616 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 617 if (!jailed_ip(cred->cr_prison, 618 (struct sockaddr *)&jsin6)) 619 ia6 = 0; 620 } 621 622 if (ia6 == 0) /* xxx scope error ?*/ 623 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 624 625 if (ia6 && jailed) { 626 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 627 if (!jailed_ip(cred->cr_prison, 628 (struct sockaddr *)&jsin6)) 629 ia6 = 0; 630 } 631 } 632 if (ia6 == 0) { 633 *errorp = EHOSTUNREACH; /* no route */ 634 return (0); 635 } 636 return (&satosin6(&ia6->ia_addr)->sin6_addr); 637 } 638 639 *errorp = EADDRNOTAVAIL; 640 return (0); 641 } 642 643 /* 644 * Default hop limit selection. The precedence is as follows: 645 * 1. Hoplimit valued specified via ioctl. 646 * 2. (If the outgoing interface is detected) the current 647 * hop limit of the interface specified by router advertisement. 648 * 3. The system default hoplimit. 649 */ 650 int 651 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp) 652 { 653 if (in6p && in6p->in6p_hops >= 0) 654 return (in6p->in6p_hops); 655 else if (ifp) 656 return (ND_IFINFO(ifp)->chlim); 657 else 658 return (ip6_defhlim); 659 } 660 #endif 661 662 void 663 in6_pcbdisconnect(struct inpcb *inp) 664 { 665 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 666 inp->inp_fport = 0; 667 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 668 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 669 in_pcbremconnhash(inp); 670 if (inp->inp_socket->so_state & SS_NOFDREF) 671 in6_pcbdetach(inp); 672 } 673 674 void 675 in6_pcbdetach(struct inpcb *inp) 676 { 677 struct socket *so = inp->inp_socket; 678 struct inpcbinfo *ipi = inp->inp_pcbinfo; 679 680 #ifdef IPSEC 681 if (inp->in6p_sp != NULL) 682 ipsec6_delete_pcbpolicy(inp); 683 #endif /* IPSEC */ 684 inp->inp_gencnt = ++ipi->ipi_gencnt; 685 in_pcbremlists(inp); 686 so->so_pcb = NULL; 687 sofree(so); 688 689 if (inp->in6p_options) 690 m_freem(inp->in6p_options); 691 ip6_freepcbopts(inp->in6p_outputopts); 692 ip6_freemoptions(inp->in6p_moptions); 693 if (inp->in6p_route.ro_rt) 694 rtfree(inp->in6p_route.ro_rt); 695 /* Check and free IPv4 related resources in case of mapped addr */ 696 if (inp->inp_options) 697 m_free(inp->inp_options); 698 ip_freemoptions(inp->inp_moptions); 699 700 inp->inp_vflag = 0; 701 zfree(ipi->ipi_zone, inp); 702 } 703 704 /* 705 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was 706 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 707 * in struct pr_usrreqs, so that protocols can just reference then directly 708 * without the need for a wrapper function. The socket must have a valid 709 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 710 * except through a kernel programming error, so it is acceptable to panic 711 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 712 * because there actually /is/ a programming error somewhere... XXX) 713 */ 714 int 715 in6_setsockaddr(struct socket *so, struct sockaddr **nam) 716 { 717 struct inpcb *inp; 718 struct sockaddr_in6 *sin6; 719 720 /* 721 * Do the malloc first in case it blocks. 722 */ 723 MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, 724 M_WAITOK | M_ZERO); 725 sin6->sin6_family = AF_INET6; 726 sin6->sin6_len = sizeof(*sin6); 727 728 crit_enter(); 729 inp = so->so_pcb; 730 if (!inp) { 731 crit_exit(); 732 kfree(sin6, M_SONAME); 733 return EINVAL; 734 } 735 sin6->sin6_port = inp->inp_lport; 736 sin6->sin6_addr = inp->in6p_laddr; 737 crit_exit(); 738 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 739 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 740 else 741 sin6->sin6_scope_id = 0; /*XXX*/ 742 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 743 sin6->sin6_addr.s6_addr16[1] = 0; 744 745 *nam = (struct sockaddr *)sin6; 746 return 0; 747 } 748 749 int 750 in6_setpeeraddr(struct socket *so, struct sockaddr **nam) 751 { 752 struct inpcb *inp; 753 struct sockaddr_in6 *sin6; 754 755 /* 756 * Do the malloc first in case it blocks. 757 */ 758 MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, 759 M_WAITOK | M_ZERO); 760 sin6->sin6_family = AF_INET6; 761 sin6->sin6_len = sizeof(struct sockaddr_in6); 762 763 crit_enter(); 764 inp = so->so_pcb; 765 if (!inp) { 766 crit_exit(); 767 kfree(sin6, M_SONAME); 768 return EINVAL; 769 } 770 sin6->sin6_port = inp->inp_fport; 771 sin6->sin6_addr = inp->in6p_faddr; 772 crit_exit(); 773 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 774 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 775 else 776 sin6->sin6_scope_id = 0; /*XXX*/ 777 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 778 sin6->sin6_addr.s6_addr16[1] = 0; 779 780 *nam = (struct sockaddr *)sin6; 781 return 0; 782 } 783 784 int 785 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 786 { 787 struct inpcb *inp = so->so_pcb; 788 int error; 789 790 if (inp == NULL) 791 return EINVAL; 792 if (inp->inp_vflag & INP_IPV4) { 793 error = in_setsockaddr(so, nam); 794 if (error == 0) 795 in6_sin_2_v4mapsin6_in_sock(nam); 796 } else 797 /* scope issues will be handled in in6_setsockaddr(). */ 798 error = in6_setsockaddr(so, nam); 799 800 return error; 801 } 802 803 int 804 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 805 { 806 struct inpcb *inp = so->so_pcb; 807 int error; 808 809 if (inp == NULL) 810 return EINVAL; 811 if (inp->inp_vflag & INP_IPV4) { 812 error = in_setpeeraddr(so, nam); 813 if (error == 0) 814 in6_sin_2_v4mapsin6_in_sock(nam); 815 } else 816 /* scope issues will be handled in in6_setpeeraddr(). */ 817 error = in6_setpeeraddr(so, nam); 818 819 return error; 820 } 821 822 /* 823 * Pass some notification to all connections of a protocol 824 * associated with address dst. The local address and/or port numbers 825 * may be specified to limit the search. The "usual action" will be 826 * taken, depending on the ctlinput cmd. The caller must filter any 827 * cmds that are uninteresting (e.g., no error in the map). 828 * Call the protocol specific routine (if any) to report 829 * any errors for each matching socket. 830 * 831 * Must be called under crit_enter(). 832 */ 833 void 834 in6_pcbnotify(struct inpcbhead *head, struct sockaddr *dst, in_port_t fport, 835 const struct sockaddr *src, in_port_t lport, int cmd, int arg, 836 void (*notify) (struct inpcb *, int)) 837 { 838 struct inpcb *inp, *ninp; 839 struct sockaddr_in6 sa6_src, *sa6_dst; 840 u_int32_t flowinfo; 841 842 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 843 return; 844 845 sa6_dst = (struct sockaddr_in6 *)dst; 846 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 847 return; 848 849 /* 850 * note that src can be NULL when we get notify by local fragmentation. 851 */ 852 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 853 flowinfo = sa6_src.sin6_flowinfo; 854 855 /* 856 * Redirects go to all references to the destination, 857 * and use in6_rtchange to invalidate the route cache. 858 * Dead host indications: also use in6_rtchange to invalidate 859 * the cache, and deliver the error to all the sockets. 860 * Otherwise, if we have knowledge of the local port and address, 861 * deliver only to that socket. 862 */ 863 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 864 fport = 0; 865 lport = 0; 866 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 867 868 if (cmd != PRC_HOSTDEAD) 869 notify = in6_rtchange; 870 } 871 if (cmd != PRC_MSGSIZE) 872 arg = inet6ctlerrmap[cmd]; 873 crit_enter(); 874 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) { 875 ninp = LIST_NEXT(inp, inp_list); 876 877 if (inp->inp_flags & INP_PLACEMARKER) 878 continue; 879 880 if ((inp->inp_vflag & INP_IPV6) == 0) 881 continue; 882 883 /* 884 * Detect if we should notify the error. If no source and 885 * destination ports are specifed, but non-zero flowinfo and 886 * local address match, notify the error. This is the case 887 * when the error is delivered with an encrypted buffer 888 * by ESP. Otherwise, just compare addresses and ports 889 * as usual. 890 */ 891 if (lport == 0 && fport == 0 && flowinfo && 892 inp->inp_socket != NULL && 893 flowinfo == (inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 894 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 895 goto do_notify; 896 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 897 &sa6_dst->sin6_addr) || 898 inp->inp_socket == 0 || 899 (lport && inp->inp_lport != lport) || 900 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 901 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 902 &sa6_src.sin6_addr)) || 903 (fport && inp->inp_fport != fport)) 904 continue; 905 906 do_notify: 907 if (notify) 908 (*notify)(inp, arg); 909 } 910 crit_exit(); 911 } 912 913 /* 914 * Lookup a PCB based on the local address and port. 915 */ 916 struct inpcb * 917 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 918 u_int lport_arg, int wild_okay, struct ucred *cred) 919 { 920 struct inpcb *inp; 921 int matchwild = 3, wildcard; 922 u_short lport = lport_arg; 923 struct inpcbporthead *porthash; 924 struct inpcbport *phd; 925 struct inpcb *match = NULL; 926 927 /* 928 * Best fit PCB lookup. 929 * 930 * First see if this local port is in use by looking on the 931 * port hash list. 932 */ 933 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 934 pcbinfo->porthashmask)]; 935 LIST_FOREACH(phd, porthash, phd_hash) { 936 if (phd->phd_port == lport) 937 break; 938 } 939 if (phd != NULL) { 940 /* 941 * Port is in use by one or more PCBs. Look for best 942 * fit. 943 */ 944 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 945 wildcard = 0; 946 if ((inp->inp_vflag & INP_IPV6) == 0) 947 continue; 948 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 949 wildcard++; 950 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 951 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 952 wildcard++; 953 else if (!IN6_ARE_ADDR_EQUAL( 954 &inp->in6p_laddr, laddr)) 955 continue; 956 } else { 957 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 958 wildcard++; 959 } 960 if (wildcard && !wild_okay) 961 continue; 962 if (wildcard < matchwild && 963 (cred == NULL || 964 cred->cr_prison == 965 inp->inp_socket->so_cred->cr_prison)) { 966 match = inp; 967 matchwild = wildcard; 968 if (wildcard == 0) 969 break; 970 else 971 matchwild = wildcard; 972 } 973 } 974 } 975 return (match); 976 } 977 978 void 979 in6_pcbpurgeif0(struct in6pcb *head, struct ifnet *ifp) 980 { 981 struct in6pcb *in6p; 982 struct ip6_moptions *im6o; 983 struct in6_multi_mship *imm, *nimm; 984 985 for (in6p = head; in6p != NULL; in6p = LIST_NEXT(in6p, inp_list)) { 986 if (in6p->in6p_flags & INP_PLACEMARKER) 987 continue; 988 im6o = in6p->in6p_moptions; 989 if ((in6p->inp_vflag & INP_IPV6) && 990 im6o) { 991 /* 992 * Unselect the outgoing interface if it is being 993 * detached. 994 */ 995 if (im6o->im6o_multicast_ifp == ifp) 996 im6o->im6o_multicast_ifp = NULL; 997 998 /* 999 * Drop multicast group membership if we joined 1000 * through the interface being detached. 1001 * XXX controversial - is it really legal for kernel 1002 * to force this? 1003 */ 1004 for (imm = im6o->im6o_memberships.lh_first; 1005 imm != NULL; imm = nimm) { 1006 nimm = imm->i6mm_chain.le_next; 1007 if (imm->i6mm_maddr->in6m_ifp == ifp) { 1008 LIST_REMOVE(imm, i6mm_chain); 1009 in6_delmulti(imm->i6mm_maddr); 1010 kfree(imm, M_IPMADDR); 1011 } 1012 } 1013 } 1014 } 1015 } 1016 1017 /* 1018 * Check for alternatives when higher level complains 1019 * about service problems. For now, invalidate cached 1020 * routing information. If the route was created dynamically 1021 * (by a redirect), time to try a default gateway again. 1022 */ 1023 void 1024 in6_losing(struct inpcb *in6p) 1025 { 1026 struct rtentry *rt; 1027 struct rt_addrinfo info; 1028 1029 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 1030 bzero((caddr_t)&info, sizeof(info)); 1031 info.rti_flags = rt->rt_flags; 1032 info.rti_info[RTAX_DST] = rt_key(rt); 1033 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1034 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1035 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 1036 if (rt->rt_flags & RTF_DYNAMIC) 1037 rtrequest1_global(RTM_DELETE, &info, NULL, NULL); 1038 in6p->in6p_route.ro_rt = NULL; 1039 rtfree(rt); 1040 /* 1041 * A new route can be allocated 1042 * the next time output is attempted. 1043 */ 1044 } 1045 } 1046 1047 /* 1048 * After a routing change, flush old routing 1049 * and allocate a (hopefully) better one. 1050 */ 1051 void 1052 in6_rtchange(struct inpcb *inp, int error) 1053 { 1054 if (inp->in6p_route.ro_rt) { 1055 rtfree(inp->in6p_route.ro_rt); 1056 inp->in6p_route.ro_rt = 0; 1057 /* 1058 * A new route can be allocated the next time 1059 * output is attempted. 1060 */ 1061 } 1062 } 1063 1064 /* 1065 * Lookup PCB in hash list. 1066 */ 1067 struct inpcb * 1068 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1069 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, 1070 int wildcard, struct ifnet *ifp) 1071 { 1072 struct inpcbhead *head; 1073 struct inpcb *inp; 1074 struct inpcb *jinp = NULL; 1075 u_short fport = fport_arg, lport = lport_arg; 1076 int faith; 1077 1078 if (faithprefix_p != NULL) 1079 faith = (*faithprefix_p)(laddr); 1080 else 1081 faith = 0; 1082 1083 /* 1084 * First look for an exact match. 1085 */ 1086 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr->s6_addr32[3] /* XXX */, 1087 fport, 1088 laddr->s6_addr32[3], /* XXX JH */ 1089 lport, 1090 pcbinfo->hashmask)]; 1091 LIST_FOREACH(inp, head, inp_hash) { 1092 if ((inp->inp_vflag & INP_IPV6) == 0) 1093 continue; 1094 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 1095 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 1096 inp->inp_fport == fport && 1097 inp->inp_lport == lport) { 1098 /* 1099 * Found. 1100 */ 1101 if (inp->inp_socket == NULL || 1102 inp->inp_socket->so_cred->cr_prison == NULL) { 1103 return (inp); 1104 } else { 1105 if (jinp == NULL) 1106 jinp = inp; 1107 } 1108 } 1109 } 1110 if (jinp != NULL) 1111 return(jinp); 1112 if (wildcard) { 1113 struct inpcontainerhead *chead; 1114 struct inpcontainer *ic; 1115 struct inpcb *local_wild = NULL; 1116 struct inpcb *jinp_wild = NULL; 1117 struct sockaddr_in6 jsin6; 1118 struct ucred *cred; 1119 1120 /* 1121 * Order of socket selection: 1122 * 1. non-jailed, non-wild. 1123 * 2. non-jailed, wild. 1124 * 3. jailed, non-wild. 1125 * 4. jailed, wild. 1126 */ 1127 jsin6.sin6_family = AF_INET6; 1128 chead = &pcbinfo->wildcardhashbase[INP_PCBWILDCARDHASH(lport, 1129 pcbinfo->wildcardhashmask)]; 1130 LIST_FOREACH(ic, chead, ic_list) { 1131 inp = ic->ic_inp; 1132 1133 if (!(inp->inp_vflag & INP_IPV6)) 1134 continue; 1135 if (inp->inp_socket != NULL) 1136 cred = inp->inp_socket->so_cred; 1137 else 1138 cred = NULL; 1139 1140 if (cred != NULL && jailed(cred)) { 1141 if (jinp != NULL) { 1142 continue; 1143 } else { 1144 jsin6.sin6_addr = *laddr; 1145 if (!jailed_ip(cred->cr_prison, 1146 (struct sockaddr *)&jsin6)) 1147 continue; 1148 } 1149 } 1150 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 1151 inp->inp_lport == lport) { 1152 if (faith && (inp->inp_flags & INP_FAITH) == 0) 1153 continue; 1154 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 1155 laddr)) { 1156 if (cred != NULL && jailed(cred)) 1157 jinp = inp; 1158 else 1159 return (inp); 1160 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 1161 if (cred != NULL && jailed(cred)) 1162 jinp_wild = inp; 1163 else 1164 local_wild = inp; 1165 } 1166 } 1167 } 1168 if (local_wild != NULL) 1169 return (local_wild); 1170 if (jinp != NULL) 1171 return (jinp); 1172 return (jinp_wild); 1173 } 1174 1175 /* 1176 * Not found. 1177 */ 1178 return (NULL); 1179 } 1180 1181 void 1182 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 1183 { 1184 struct ip6_hdr *ip; 1185 1186 ip = mtod(m, struct ip6_hdr *); 1187 bzero(sin6, sizeof(*sin6)); 1188 sin6->sin6_len = sizeof(*sin6); 1189 sin6->sin6_family = AF_INET6; 1190 sin6->sin6_addr = ip->ip6_src; 1191 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1192 sin6->sin6_addr.s6_addr16[1] = 0; 1193 sin6->sin6_scope_id = 1194 (m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1195 ? m->m_pkthdr.rcvif->if_index : 0; 1196 1197 return; 1198 } 1199