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 1.21 2005/01/06 17:59:32 hsu 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 88 #include <vm/vm_zone.h> 89 90 #include <net/if.h> 91 #include <net/if_types.h> 92 #include <net/route.h> 93 94 #include <netinet/in.h> 95 #include <netinet/in_var.h> 96 #include <netinet/in_systm.h> 97 #include <netinet/ip6.h> 98 #include <netinet/ip_var.h> 99 #include <netinet6/ip6_var.h> 100 #include <netinet6/nd6.h> 101 #include <netinet/in_pcb.h> 102 #include <netinet6/in6_pcb.h> 103 104 #ifdef IPSEC 105 #include <netinet6/ipsec.h> 106 #ifdef INET6 107 #include <netinet6/ipsec6.h> 108 #endif 109 #include <netinet6/ah.h> 110 #ifdef INET6 111 #include <netinet6/ah6.h> 112 #endif 113 #include <netproto/key/key.h> 114 #endif /* IPSEC */ 115 116 #ifdef FAST_IPSEC 117 #include <netproto/ipsec/ipsec.h> 118 #include <netproto/ipsec/ipsec6.h> 119 #include <netproto/ipsec/key.h> 120 #define IPSEC 121 #endif /* FAST_IPSEC */ 122 123 struct in6_addr zeroin6_addr; 124 125 int 126 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 127 { 128 struct socket *so = inp->inp_socket; 129 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 130 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 131 u_short lport = 0; 132 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 133 134 if (!in6_ifaddr) /* XXX broken! */ 135 return (EADDRNOTAVAIL); 136 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 137 return(EINVAL); 138 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 139 wild = 1; 140 if (nam) { 141 sin6 = (struct sockaddr_in6 *)nam; 142 if (nam->sa_len != sizeof(*sin6)) 143 return(EINVAL); 144 /* 145 * family check. 146 */ 147 if (nam->sa_family != AF_INET6) 148 return(EAFNOSUPPORT); 149 150 /* KAME hack: embed scopeid */ 151 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, NULL) != 0) 152 return EINVAL; 153 /* this must be cleared for ifa_ifwithaddr() */ 154 sin6->sin6_scope_id = 0; 155 156 lport = sin6->sin6_port; 157 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 158 /* 159 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 160 * allow compepte duplication of binding if 161 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 162 * and a multicast address is bound on both 163 * new and duplicated sockets. 164 */ 165 if (so->so_options & SO_REUSEADDR) 166 reuseport = SO_REUSEADDR|SO_REUSEPORT; 167 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 168 struct ifaddr *ia = NULL; 169 170 sin6->sin6_port = 0; /* yech... */ 171 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 172 return(EADDRNOTAVAIL); 173 174 /* 175 * XXX: bind to an anycast address might accidentally 176 * cause sending a packet with anycast source address. 177 * We should allow to bind to a deprecated address, since 178 * the application dare to use it. 179 */ 180 if (ia && 181 ((struct in6_ifaddr *)ia)->ia6_flags & 182 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 183 return(EADDRNOTAVAIL); 184 } 185 } 186 if (lport) { 187 struct inpcb *t; 188 struct proc *p = td->td_proc; /* may be NULL */ 189 190 /* GROSS */ 191 if (ntohs(lport) < IPV6PORT_RESERVED && p && 192 suser_cred(p->p_ucred, PRISON_ROOT)) 193 return(EACCES); 194 if (so->so_cred->cr_uid != 0 && 195 !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 196 t = in6_pcblookup_local(pcbinfo, 197 &sin6->sin6_addr, lport, 198 INPLOOKUP_WILDCARD); 199 if (t && 200 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 201 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 202 (t->inp_socket->so_options & 203 SO_REUSEPORT) == 0) && 204 (so->so_cred->cr_uid != 205 t->inp_socket->so_cred->cr_uid)) 206 return (EADDRINUSE); 207 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 208 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 209 struct sockaddr_in sin; 210 211 in6_sin6_2_sin(&sin, sin6); 212 t = in_pcblookup_local(pcbinfo, 213 sin.sin_addr, lport, 214 INPLOOKUP_WILDCARD); 215 if (t && 216 (so->so_cred->cr_uid != 217 t->inp_socket->so_cred->cr_uid) && 218 (ntohl(t->inp_laddr.s_addr) != 219 INADDR_ANY || 220 INP_SOCKAF(so) == 221 INP_SOCKAF(t->inp_socket))) 222 return (EADDRINUSE); 223 } 224 } 225 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 226 lport, wild); 227 if (t && (reuseport & t->inp_socket->so_options) == 0) 228 return(EADDRINUSE); 229 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 230 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 231 struct sockaddr_in sin; 232 233 in6_sin6_2_sin(&sin, sin6); 234 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 235 lport, wild); 236 if (t && 237 (reuseport & t->inp_socket->so_options) 238 == 0 && 239 (ntohl(t->inp_laddr.s_addr) 240 != INADDR_ANY || 241 INP_SOCKAF(so) == 242 INP_SOCKAF(t->inp_socket))) 243 return (EADDRINUSE); 244 } 245 } 246 inp->in6p_laddr = sin6->sin6_addr; 247 } 248 if (lport == 0) { 249 int e; 250 if ((e = in6_pcbsetport(&inp->in6p_laddr, inp, td)) != 0) 251 return(e); 252 } 253 else { 254 inp->inp_lport = lport; 255 if (in_pcbinsporthash(inp) != 0) { 256 inp->in6p_laddr = in6addr_any; 257 inp->inp_lport = 0; 258 return (EAGAIN); 259 } 260 } 261 return(0); 262 } 263 264 /* 265 * Transform old in6_pcbconnect() into an inner subroutine for new 266 * in6_pcbconnect(): Do some validity-checking on the remote 267 * address (in mbuf 'nam') and then determine local host address 268 * (i.e., which interface) to use to access that remote host. 269 * 270 * This preserves definition of in6_pcbconnect(), while supporting a 271 * slightly different version for T/TCP. (This is more than 272 * a bit of a kludge, but cleaning up the internal interfaces would 273 * have forced minor changes in every protocol). 274 */ 275 276 int 277 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam, 278 struct in6_addr **plocal_addr6) 279 { 280 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 281 struct ifnet *ifp = NULL; 282 int error = 0; 283 284 if (nam->sa_len != sizeof (*sin6)) 285 return (EINVAL); 286 if (sin6->sin6_family != AF_INET6) 287 return (EAFNOSUPPORT); 288 if (sin6->sin6_port == 0) 289 return (EADDRNOTAVAIL); 290 291 /* KAME hack: embed scopeid */ 292 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, &ifp) != 0) 293 return EINVAL; 294 295 if (in6_ifaddr) { 296 /* 297 * If the destination address is UNSPECIFIED addr, 298 * use the loopback addr, e.g ::1. 299 */ 300 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 301 sin6->sin6_addr = in6addr_loopback; 302 } 303 { 304 /* 305 * XXX: in6_selectsrc might replace the bound local address 306 * with the address specified by setsockopt(IPV6_PKTINFO). 307 * Is it the intended behavior? 308 */ 309 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts, 310 inp->in6p_moptions, 311 &inp->in6p_route, 312 &inp->in6p_laddr, &error); 313 if (*plocal_addr6 == 0) { 314 if (error == 0) 315 error = EADDRNOTAVAIL; 316 return(error); 317 } 318 /* 319 * Don't do pcblookup call here; return interface in 320 * plocal_addr6 321 * and exit to caller, that will do the lookup. 322 */ 323 } 324 325 if (inp->in6p_route.ro_rt) 326 ifp = inp->in6p_route.ro_rt->rt_ifp; 327 328 return(0); 329 } 330 331 /* 332 * Outer subroutine: 333 * Connect from a socket to a specified address. 334 * Both address and port must be specified in argument sin. 335 * If don't have a local address for this socket yet, 336 * then pick one. 337 */ 338 int 339 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 340 { 341 struct in6_addr *addr6; 342 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 343 int error; 344 345 /* 346 * Call inner routine, to assign local interface address. 347 * in6_pcbladdr() may automatically fill in sin6_scope_id. 348 */ 349 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 350 return(error); 351 352 if (in6_pcblookup_hash(inp->inp_cpcbinfo, &sin6->sin6_addr, 353 sin6->sin6_port, 354 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 355 ? addr6 : &inp->in6p_laddr, 356 inp->inp_lport, 0, NULL) != NULL) { 357 return (EADDRINUSE); 358 } 359 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 360 if (inp->inp_lport == 0) { 361 error = in6_pcbbind(inp, (struct sockaddr *)0, td); 362 if (error) 363 return (error); 364 } 365 inp->in6p_laddr = *addr6; 366 } 367 inp->in6p_faddr = sin6->sin6_addr; 368 inp->inp_fport = sin6->sin6_port; 369 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 370 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 371 if (inp->in6p_flags & IN6P_AUTOFLOWLABEL) 372 inp->in6p_flowinfo |= 373 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 374 375 in_pcbinsconnhash(inp); 376 return (0); 377 } 378 379 #if 0 380 /* 381 * Return an IPv6 address, which is the most appropriate for given 382 * destination and user specified options. 383 * If necessary, this function lookups the routing table and return 384 * an entry to the caller for later use. 385 */ 386 struct in6_addr * 387 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 388 struct ip6_moptions *mopts, struct route_in6 *ro, 389 struct in6_addr *laddr, int *errorp) 390 { 391 struct in6_addr *dst; 392 struct in6_ifaddr *ia6 = 0; 393 struct in6_pktinfo *pi = NULL; 394 395 dst = &dstsock->sin6_addr; 396 *errorp = 0; 397 398 /* 399 * If the source address is explicitly specified by the caller, 400 * use it. 401 */ 402 if (opts && (pi = opts->ip6po_pktinfo) && 403 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) 404 return(&pi->ipi6_addr); 405 406 /* 407 * If the source address is not specified but the socket(if any) 408 * is already bound, use the bound address. 409 */ 410 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) 411 return(laddr); 412 413 /* 414 * If the caller doesn't specify the source address but 415 * the outgoing interface, use an address associated with 416 * the interface. 417 */ 418 if (pi && pi->ipi6_ifindex) { 419 /* XXX boundary check is assumed to be already done. */ 420 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex], 421 dst); 422 if (ia6 == 0) { 423 *errorp = EADDRNOTAVAIL; 424 return(0); 425 } 426 return(&satosin6(&ia6->ia_addr)->sin6_addr); 427 } 428 429 /* 430 * If the destination address is a link-local unicast address or 431 * a multicast address, and if the outgoing interface is specified 432 * by the sin6_scope_id filed, use an address associated with the 433 * interface. 434 * XXX: We're now trying to define more specific semantics of 435 * sin6_scope_id field, so this part will be rewritten in 436 * the near future. 437 */ 438 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) && 439 dstsock->sin6_scope_id) { 440 /* 441 * I'm not sure if boundary check for scope_id is done 442 * somewhere... 443 */ 444 if (dstsock->sin6_scope_id < 0 || 445 if_index < dstsock->sin6_scope_id) { 446 *errorp = ENXIO; /* XXX: better error? */ 447 return(0); 448 } 449 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id], 450 dst); 451 if (ia6 == 0) { 452 *errorp = EADDRNOTAVAIL; 453 return(0); 454 } 455 return(&satosin6(&ia6->ia_addr)->sin6_addr); 456 } 457 458 /* 459 * If the destination address is a multicast address and 460 * the outgoing interface for the address is specified 461 * by the caller, use an address associated with the interface. 462 * There is a sanity check here; if the destination has node-local 463 * scope, the outgoing interfacde should be a loopback address. 464 * Even if the outgoing interface is not specified, we also 465 * choose a loopback interface as the outgoing interface. 466 */ 467 if (IN6_IS_ADDR_MULTICAST(dst)) { 468 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL; 469 470 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) { 471 ifp = &loif[0]; 472 } 473 474 if (ifp) { 475 ia6 = in6_ifawithscope(ifp, dst); 476 if (ia6 == 0) { 477 *errorp = EADDRNOTAVAIL; 478 return(0); 479 } 480 return(&ia6->ia_addr.sin6_addr); 481 } 482 } 483 484 /* 485 * If the next hop address for the packet is specified 486 * by caller, use an address associated with the route 487 * to the next hop. 488 */ 489 { 490 struct sockaddr_in6 *sin6_next; 491 struct rtentry *rt; 492 493 if (opts && opts->ip6po_nexthop) { 494 sin6_next = satosin6(opts->ip6po_nexthop); 495 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL); 496 if (rt) { 497 ia6 = in6_ifawithscope(rt->rt_ifp, dst); 498 if (ia6 == 0) 499 ia6 = ifatoia6(rt->rt_ifa); 500 } 501 if (ia6 == 0) { 502 *errorp = EADDRNOTAVAIL; 503 return(0); 504 } 505 return(&satosin6(&ia6->ia_addr)->sin6_addr); 506 } 507 } 508 509 /* 510 * If route is known or can be allocated now, 511 * our src addr is taken from the i/f, else punt. 512 */ 513 if (ro) { 514 if (ro->ro_rt && 515 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) { 516 RTFREE(ro->ro_rt); 517 ro->ro_rt = NULL; 518 } 519 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) { 520 struct sockaddr_in6 *dst6; 521 522 /* No route yet, so try to acquire one */ 523 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 524 dst6 = &ro->ro_dst; 525 dst6->sin6_family = AF_INET6; 526 dst6->sin6_len = sizeof(struct sockaddr_in6); 527 dst6->sin6_addr = *dst; 528 if (IN6_IS_ADDR_MULTICAST(dst)) { 529 ro->ro_rt = 530 rtpurelookup((struct sockaddr *)&ro->ro_dst); 531 } else { 532 rtalloc((struct route *)ro); 533 } 534 } 535 536 /* 537 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 538 * the address. But we don't know why it does so. 539 * It is necessary to ensure the scope even for lo0 540 * so doesn't check out IFF_LOOPBACK. 541 */ 542 543 if (ro->ro_rt) { 544 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst); 545 if (ia6 == 0) /* xxx scope error ?*/ 546 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 547 } 548 if (ia6 == 0) { 549 *errorp = EHOSTUNREACH; /* no route */ 550 return(0); 551 } 552 return(&satosin6(&ia6->ia_addr)->sin6_addr); 553 } 554 555 *errorp = EADDRNOTAVAIL; 556 return(0); 557 } 558 559 /* 560 * Default hop limit selection. The precedence is as follows: 561 * 1. Hoplimit valued specified via ioctl. 562 * 2. (If the outgoing interface is detected) the current 563 * hop limit of the interface specified by router advertisement. 564 * 3. The system default hoplimit. 565 */ 566 int 567 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp) 568 { 569 if (in6p && in6p->in6p_hops >= 0) 570 return(in6p->in6p_hops); 571 else if (ifp) 572 return(nd_ifinfo[ifp->if_index].chlim); 573 else 574 return(ip6_defhlim); 575 } 576 #endif 577 578 void 579 in6_pcbdisconnect(struct inpcb *inp) 580 { 581 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 582 inp->inp_fport = 0; 583 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 584 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 585 in_pcbremconnhash(inp); 586 if (inp->inp_socket->so_state & SS_NOFDREF) 587 in6_pcbdetach(inp); 588 } 589 590 void 591 in6_pcbdetach(struct inpcb *inp) 592 { 593 struct socket *so = inp->inp_socket; 594 struct inpcbinfo *ipi = inp->inp_pcbinfo; 595 596 #ifdef IPSEC 597 if (inp->in6p_sp != NULL) 598 ipsec6_delete_pcbpolicy(inp); 599 #endif /* IPSEC */ 600 inp->inp_gencnt = ++ipi->ipi_gencnt; 601 in_pcbremlists(inp); 602 sotoinpcb(so) = 0; 603 sofree(so); 604 605 if (inp->in6p_options) 606 m_freem(inp->in6p_options); 607 ip6_freepcbopts(inp->in6p_outputopts); 608 ip6_freemoptions(inp->in6p_moptions); 609 if (inp->in6p_route.ro_rt) 610 rtfree(inp->in6p_route.ro_rt); 611 /* Check and free IPv4 related resources in case of mapped addr */ 612 if (inp->inp_options) 613 (void)m_free(inp->inp_options); 614 ip_freemoptions(inp->inp_moptions); 615 616 inp->inp_vflag = 0; 617 zfree(ipi->ipi_zone, inp); 618 } 619 620 /* 621 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was 622 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 623 * in struct pr_usrreqs, so that protocols can just reference then directly 624 * without the need for a wrapper function. The socket must have a valid 625 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 626 * except through a kernel programming error, so it is acceptable to panic 627 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 628 * because there actually /is/ a programming error somewhere... XXX) 629 */ 630 int 631 in6_setsockaddr(struct socket *so, struct sockaddr **nam) 632 { 633 int s; 634 struct inpcb *inp; 635 struct sockaddr_in6 *sin6; 636 637 /* 638 * Do the malloc first in case it blocks. 639 */ 640 MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK); 641 bzero(sin6, sizeof *sin6); 642 sin6->sin6_family = AF_INET6; 643 sin6->sin6_len = sizeof(*sin6); 644 645 s = splnet(); 646 inp = sotoinpcb(so); 647 if (!inp) { 648 splx(s); 649 free(sin6, M_SONAME); 650 return EINVAL; 651 } 652 sin6->sin6_port = inp->inp_lport; 653 sin6->sin6_addr = inp->in6p_laddr; 654 splx(s); 655 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 656 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 657 else 658 sin6->sin6_scope_id = 0; /*XXX*/ 659 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 660 sin6->sin6_addr.s6_addr16[1] = 0; 661 662 *nam = (struct sockaddr *)sin6; 663 return 0; 664 } 665 666 int 667 in6_setpeeraddr(struct socket *so, struct sockaddr **nam) 668 { 669 int s; 670 struct inpcb *inp; 671 struct sockaddr_in6 *sin6; 672 673 /* 674 * Do the malloc first in case it blocks. 675 */ 676 MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, M_WAITOK); 677 bzero((caddr_t)sin6, sizeof (*sin6)); 678 sin6->sin6_family = AF_INET6; 679 sin6->sin6_len = sizeof(struct sockaddr_in6); 680 681 s = splnet(); 682 inp = sotoinpcb(so); 683 if (!inp) { 684 splx(s); 685 free(sin6, M_SONAME); 686 return EINVAL; 687 } 688 sin6->sin6_port = inp->inp_fport; 689 sin6->sin6_addr = inp->in6p_faddr; 690 splx(s); 691 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 692 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 693 else 694 sin6->sin6_scope_id = 0; /*XXX*/ 695 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 696 sin6->sin6_addr.s6_addr16[1] = 0; 697 698 *nam = (struct sockaddr *)sin6; 699 return 0; 700 } 701 702 int 703 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 704 { 705 struct inpcb *inp = sotoinpcb(so); 706 int error; 707 708 if (inp == NULL) 709 return EINVAL; 710 if (inp->inp_vflag & INP_IPV4) { 711 error = in_setsockaddr(so, nam); 712 if (error == 0) 713 in6_sin_2_v4mapsin6_in_sock(nam); 714 } else 715 /* scope issues will be handled in in6_setsockaddr(). */ 716 error = in6_setsockaddr(so, nam); 717 718 return error; 719 } 720 721 int 722 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 723 { 724 struct inpcb *inp = sotoinpcb(so); 725 int error; 726 727 if (inp == NULL) 728 return EINVAL; 729 if (inp->inp_vflag & INP_IPV4) { 730 error = in_setpeeraddr(so, nam); 731 if (error == 0) 732 in6_sin_2_v4mapsin6_in_sock(nam); 733 } else 734 /* scope issues will be handled in in6_setpeeraddr(). */ 735 error = in6_setpeeraddr(so, nam); 736 737 return error; 738 } 739 740 /* 741 * Pass some notification to all connections of a protocol 742 * associated with address dst. The local address and/or port numbers 743 * may be specified to limit the search. The "usual action" will be 744 * taken, depending on the ctlinput cmd. The caller must filter any 745 * cmds that are uninteresting (e.g., no error in the map). 746 * Call the protocol specific routine (if any) to report 747 * any errors for each matching socket. 748 * 749 * Must be called at splnet. 750 */ 751 void 752 in6_pcbnotify(struct inpcbhead *head, struct sockaddr *dst, u_int fport_arg, 753 const struct sockaddr *src, u_int lport_arg, int cmd, 754 void (*notify) (struct inpcb *, int)) 755 { 756 struct inpcb *inp, *ninp; 757 struct sockaddr_in6 sa6_src, *sa6_dst; 758 u_short fport = fport_arg, lport = lport_arg; 759 u_int32_t flowinfo; 760 int errno, s; 761 762 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 763 return; 764 765 sa6_dst = (struct sockaddr_in6 *)dst; 766 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 767 return; 768 769 /* 770 * note that src can be NULL when we get notify by local fragmentation. 771 */ 772 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 773 flowinfo = sa6_src.sin6_flowinfo; 774 775 /* 776 * Redirects go to all references to the destination, 777 * and use in6_rtchange to invalidate the route cache. 778 * Dead host indications: also use in6_rtchange to invalidate 779 * the cache, and deliver the error to all the sockets. 780 * Otherwise, if we have knowledge of the local port and address, 781 * deliver only to that socket. 782 */ 783 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 784 fport = 0; 785 lport = 0; 786 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 787 788 if (cmd != PRC_HOSTDEAD) 789 notify = in6_rtchange; 790 } 791 errno = inet6ctlerrmap[cmd]; 792 s = splnet(); 793 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) { 794 ninp = LIST_NEXT(inp, inp_list); 795 796 if (inp->inp_flags & INP_PLACEMARKER) 797 continue; 798 799 if ((inp->inp_vflag & INP_IPV6) == 0) 800 continue; 801 802 /* 803 * Detect if we should notify the error. If no source and 804 * destination ports are specifed, but non-zero flowinfo and 805 * local address match, notify the error. This is the case 806 * when the error is delivered with an encrypted buffer 807 * by ESP. Otherwise, just compare addresses and ports 808 * as usual. 809 */ 810 if (lport == 0 && fport == 0 && flowinfo && 811 inp->inp_socket != NULL && 812 flowinfo == (inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 813 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 814 goto do_notify; 815 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 816 &sa6_dst->sin6_addr) || 817 inp->inp_socket == 0 || 818 (lport && inp->inp_lport != lport) || 819 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 820 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 821 &sa6_src.sin6_addr)) || 822 (fport && inp->inp_fport != fport)) 823 continue; 824 825 do_notify: 826 if (notify) 827 (*notify)(inp, errno); 828 } 829 splx(s); 830 } 831 832 /* 833 * Lookup a PCB based on the local address and port. 834 */ 835 struct inpcb * 836 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 837 u_int lport_arg, int wild_okay) 838 { 839 struct inpcb *inp; 840 int matchwild = 3, wildcard; 841 u_short lport = lport_arg; 842 struct inpcbporthead *porthash; 843 struct inpcbport *phd; 844 struct inpcb *match = NULL; 845 846 /* 847 * Best fit PCB lookup. 848 * 849 * First see if this local port is in use by looking on the 850 * port hash list. 851 */ 852 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 853 pcbinfo->porthashmask)]; 854 LIST_FOREACH(phd, porthash, phd_hash) { 855 if (phd->phd_port == lport) 856 break; 857 } 858 if (phd != NULL) { 859 /* 860 * Port is in use by one or more PCBs. Look for best 861 * fit. 862 */ 863 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 864 wildcard = 0; 865 if ((inp->inp_vflag & INP_IPV6) == 0) 866 continue; 867 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 868 wildcard++; 869 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 870 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 871 wildcard++; 872 else if (!IN6_ARE_ADDR_EQUAL( 873 &inp->in6p_laddr, laddr)) 874 continue; 875 } else { 876 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 877 wildcard++; 878 } 879 if (wildcard && !wild_okay) 880 continue; 881 if (wildcard < matchwild) { 882 match = inp; 883 if (wildcard == 0) 884 break; 885 else 886 matchwild = wildcard; 887 } 888 } 889 } 890 return (match); 891 } 892 893 void 894 in6_pcbpurgeif0(struct in6pcb *head, struct ifnet *ifp) 895 { 896 struct in6pcb *in6p; 897 struct ip6_moptions *im6o; 898 struct in6_multi_mship *imm, *nimm; 899 900 for (in6p = head; in6p != NULL; in6p = LIST_NEXT(in6p, inp_list)) { 901 if (in6p->in6p_flags & INP_PLACEMARKER) 902 continue; 903 im6o = in6p->in6p_moptions; 904 if ((in6p->inp_vflag & INP_IPV6) && 905 im6o) { 906 /* 907 * Unselect the outgoing interface if it is being 908 * detached. 909 */ 910 if (im6o->im6o_multicast_ifp == ifp) 911 im6o->im6o_multicast_ifp = NULL; 912 913 /* 914 * Drop multicast group membership if we joined 915 * through the interface being detached. 916 * XXX controversial - is it really legal for kernel 917 * to force this? 918 */ 919 for (imm = im6o->im6o_memberships.lh_first; 920 imm != NULL; imm = nimm) { 921 nimm = imm->i6mm_chain.le_next; 922 if (imm->i6mm_maddr->in6m_ifp == ifp) { 923 LIST_REMOVE(imm, i6mm_chain); 924 in6_delmulti(imm->i6mm_maddr); 925 free(imm, M_IPMADDR); 926 } 927 } 928 } 929 } 930 } 931 932 /* 933 * Check for alternatives when higher level complains 934 * about service problems. For now, invalidate cached 935 * routing information. If the route was created dynamically 936 * (by a redirect), time to try a default gateway again. 937 */ 938 void 939 in6_losing(struct inpcb *in6p) 940 { 941 struct rtentry *rt; 942 struct rt_addrinfo info; 943 944 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 945 bzero((caddr_t)&info, sizeof(info)); 946 info.rti_flags = rt->rt_flags; 947 info.rti_info[RTAX_DST] = rt_key(rt); 948 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 949 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 950 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 951 if (rt->rt_flags & RTF_DYNAMIC) 952 (void)rtrequest1(RTM_DELETE, &info, NULL); 953 in6p->in6p_route.ro_rt = NULL; 954 rtfree(rt); 955 /* 956 * A new route can be allocated 957 * the next time output is attempted. 958 */ 959 } 960 } 961 962 /* 963 * After a routing change, flush old routing 964 * and allocate a (hopefully) better one. 965 */ 966 void 967 in6_rtchange(struct inpcb *inp, int errno) 968 { 969 if (inp->in6p_route.ro_rt) { 970 rtfree(inp->in6p_route.ro_rt); 971 inp->in6p_route.ro_rt = 0; 972 /* 973 * A new route can be allocated the next time 974 * output is attempted. 975 */ 976 } 977 } 978 979 /* 980 * Lookup PCB in hash list. 981 */ 982 struct inpcb * 983 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 984 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, 985 int wildcard, struct ifnet *ifp) 986 { 987 struct inpcbhead *head; 988 struct inpcb *inp; 989 u_short fport = fport_arg, lport = lport_arg; 990 int faith; 991 992 if (faithprefix_p != NULL) 993 faith = (*faithprefix_p)(laddr); 994 else 995 faith = 0; 996 997 /* 998 * First look for an exact match. 999 */ 1000 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr->s6_addr32[3] /* XXX */, 1001 fport, 1002 laddr->s6_addr32[3], /* XXX JH */ 1003 lport, 1004 pcbinfo->hashmask)]; 1005 LIST_FOREACH(inp, head, inp_hash) { 1006 if ((inp->inp_vflag & INP_IPV6) == 0) 1007 continue; 1008 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 1009 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 1010 inp->inp_fport == fport && 1011 inp->inp_lport == lport) { 1012 /* 1013 * Found. 1014 */ 1015 return (inp); 1016 } 1017 } 1018 if (wildcard) { 1019 struct inpcontainerhead *chead; 1020 struct inpcontainer *ic; 1021 struct inpcb *local_wild = NULL; 1022 1023 chead = &pcbinfo->wildcardhashbase[INP_PCBWILDCARDHASH(lport, 1024 pcbinfo->wildcardhashmask)]; 1025 LIST_FOREACH(ic, chead, ic_list) { 1026 inp = ic->ic_inp; 1027 1028 if (!(inp->inp_vflag & INP_IPV6)) 1029 continue; 1030 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 1031 inp->inp_lport == lport) { 1032 if (faith && (inp->inp_flags & INP_FAITH) == 0) 1033 continue; 1034 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 1035 laddr)) 1036 return (inp); 1037 else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 1038 local_wild = inp; 1039 } 1040 } 1041 return (local_wild); 1042 } 1043 1044 /* 1045 * Not found. 1046 */ 1047 return (NULL); 1048 } 1049 1050 void 1051 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 1052 { 1053 struct ip6_hdr *ip; 1054 1055 ip = mtod(m, struct ip6_hdr *); 1056 bzero(sin6, sizeof(*sin6)); 1057 sin6->sin6_len = sizeof(*sin6); 1058 sin6->sin6_family = AF_INET6; 1059 sin6->sin6_addr = ip->ip6_src; 1060 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1061 sin6->sin6_addr.s6_addr16[1] = 0; 1062 sin6->sin6_scope_id = 1063 (m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1064 ? m->m_pkthdr.rcvif->if_index : 0; 1065 1066 return; 1067 } 1068