in6_pcb.c revision 55679
1/* 2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. Neither the name of the project nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * $FreeBSD: head/sys/netinet6/in6_pcb.c 55679 2000-01-09 19:17:30Z shin $ 30 */ 31 32/* 33 * Copyright (c) 1982, 1986, 1991, 1993 34 * The Regents of the University of California. All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. All advertising materials mentioning features or use of this software 45 * must display the following acknowledgement: 46 * This product includes software developed by the University of 47 * California, Berkeley and its contributors. 48 * 4. Neither the name of the University nor the names of its contributors 49 * may be used to endorse or promote products derived from this software 50 * without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 62 * SUCH DAMAGE. 63 * 64 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 65 * $FreeBSD: head/sys/netinet6/in6_pcb.c 55679 2000-01-09 19:17:30Z shin $ 66 */ 67 68#include "opt_ipsec.h" 69 70#include <sys/param.h> 71#include <sys/systm.h> 72#include <sys/malloc.h> 73#include <sys/mbuf.h> 74#include <sys/domain.h> 75#include <sys/protosw.h> 76#include <sys/socket.h> 77#include <sys/socketvar.h> 78#include <sys/sockio.h> 79#include <sys/errno.h> 80#include <sys/time.h> 81#include <sys/proc.h> 82#include <sys/jail.h> 83 84#include <vm/vm_zone.h> 85 86#include <net/if.h> 87#include <net/if_types.h> 88#include <net/route.h> 89 90#include <netinet/in.h> 91#include <netinet/in_var.h> 92#include <netinet/in_systm.h> 93#include <netinet6/ip6.h> 94#include <netinet/ip_var.h> 95#include <netinet6/ip6_var.h> 96#include <netinet6/nd6.h> 97#include <netinet/in_pcb.h> 98#include <netinet6/in6_pcb.h> 99 100#include "faith.h" 101 102#ifdef IPSEC 103#include <netinet6/ipsec.h> 104#include <netinet6/ah.h> 105#include <netinet6/ipsec6.h> 106#include <netinet6/ah6.h> 107#include <netkey/key.h> 108#ifdef IPSEC_DEBUG 109#include <netkey/key_debug.h> 110#else 111#define KEYDEBUG(lev,arg) 112#endif /* IPSEC_DEBUG */ 113#endif /* IPSEC */ 114 115struct in6_addr zeroin6_addr; 116 117int 118in6_pcbbind(inp, nam, p) 119 register struct inpcb *inp; 120 struct sockaddr *nam; 121 struct proc *p; 122{ 123 struct socket *so = inp->inp_socket; 124 unsigned short *lastport; 125 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 126 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 127 u_short lport = 0; 128 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 129 int error; 130 131 if (!in6_ifaddr) /* XXX broken! */ 132 return (EADDRNOTAVAIL); 133 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 134 return(EINVAL); 135 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 136 wild = 1; 137 if (nam) { 138 sin6 = (struct sockaddr_in6 *)nam; 139 if (nam->sa_len != sizeof(*sin6)) 140 return(EINVAL); 141 /* 142 * family check. 143 */ 144 if (nam->sa_family != AF_INET6) 145 return(EAFNOSUPPORT); 146 147 /* 148 * If the scope of the destination is link-local, embed the 149 * interface index in the address. 150 */ 151 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) { 152 /* XXX boundary check is assumed to be already done. */ 153 /* XXX sin6_scope_id is weaker than advanced-api. */ 154 struct in6_pktinfo *pi; 155 if (inp->in6p_outputopts && 156 (pi = inp->in6p_outputopts->ip6po_pktinfo) && 157 pi->ipi6_ifindex) { 158 sin6->sin6_addr.s6_addr16[1] 159 = htons(pi->ipi6_ifindex); 160 } else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) 161 && inp->in6p_moptions 162 && inp->in6p_moptions->im6o_multicast_ifp) { 163 sin6->sin6_addr.s6_addr16[1] = 164 htons(inp->in6p_moptions->im6o_multicast_ifp->if_index); 165 } else if (sin6->sin6_scope_id) { 166 /* boundary check */ 167 if (sin6->sin6_scope_id < 0 168 || if_index < sin6->sin6_scope_id) { 169 return ENXIO; /* XXX EINVAL? */ 170 } 171 sin6->sin6_addr.s6_addr16[1] 172 = htons(sin6->sin6_scope_id & 0xffff);/*XXX*/ 173 /* this must be cleared for ifa_ifwithaddr() */ 174 sin6->sin6_scope_id = 0; 175 } 176 } 177 178 lport = sin6->sin6_port; 179 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 180 /* 181 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 182 * allow compepte duplication of binding if 183 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 184 * and a multicast address is bound on both 185 * new and duplicated sockets. 186 */ 187 if (so->so_options & SO_REUSEADDR) 188 reuseport = SO_REUSEADDR|SO_REUSEPORT; 189 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 190 struct ifaddr *ia = NULL; 191 192 sin6->sin6_port = 0; /* yech... */ 193 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 194 return(EADDRNOTAVAIL); 195 196 /* 197 * XXX: bind to an anycast address might accidentally 198 * cause sending a packet with anycast source address. 199 */ 200 if (ia && 201 ((struct in6_ifaddr *)ia)->ia6_flags & 202 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY| 203 IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) { 204 return(EADDRNOTAVAIL); 205 } 206 } 207 if (lport) { 208 struct inpcb *t; 209 210 /* GROSS */ 211 if (ntohs(lport) < IPV6PORT_RESERVED && p && 212 suser_xxx(0, p, PRISON_ROOT)) 213 return(EACCES); 214 if (so->so_cred->cr_uid != 0 && 215 !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 216 t = in6_pcblookup_local(pcbinfo, 217 &sin6->sin6_addr, lport, 218 INPLOOKUP_WILDCARD); 219 if (t && 220 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 221 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 222 (t->inp_socket->so_options & 223 SO_REUSEPORT) == 0) && 224 (so->so_cred->cr_uid != 225 t->inp_socket->so_cred->cr_uid)) 226 return (EADDRINUSE); 227 if (ip6_mapped_addr_on != 0 && 228 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 229 struct sockaddr_in sin; 230 231 in6_sin6_2_sin(&sin, sin6); 232 t = in_pcblookup_local(pcbinfo, 233 sin.sin_addr, lport, 234 INPLOOKUP_WILDCARD); 235 if (t && 236 (so->so_cred->cr_uid != 237 t->inp_socket->so_cred->cr_uid) && 238 (ntohl(t->inp_laddr.s_addr) != 239 INADDR_ANY || 240 INP_SOCKAF(so) == 241 INP_SOCKAF(t->inp_socket))) 242 return (EADDRINUSE); 243 } 244 } 245 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 246 lport, wild); 247 if (t && (reuseport & t->inp_socket->so_options) == 0) 248 return(EADDRINUSE); 249 if (ip6_mapped_addr_on != 0 && 250 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 251 struct sockaddr_in sin; 252 253 in6_sin6_2_sin(&sin, sin6); 254 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 255 lport, wild); 256 if (t && 257 (reuseport & t->inp_socket->so_options) 258 == 0 && 259 (ntohl(t->inp_laddr.s_addr) 260 != INADDR_ANY || 261 INP_SOCKAF(so) == 262 INP_SOCKAF(t->inp_socket))) 263 return (EADDRINUSE); 264 } 265 } 266 inp->in6p_laddr = sin6->sin6_addr; 267 } 268 if (lport == 0) { 269 ushort first, last; 270 int count; 271 272 inp->inp_flags |= INP_ANONPORT; 273 274 if (inp->inp_flags & INP_HIGHPORT) { 275 first = ipport_hifirstauto; /* sysctl */ 276 last = ipport_hilastauto; 277 lastport = &pcbinfo->lasthi; 278 } else if (inp->inp_flags & INP_LOWPORT) { 279 if (p && (error = suser_xxx(0, p, PRISON_ROOT))) 280 return error; 281 first = ipport_lowfirstauto; /* 1023 */ 282 last = ipport_lowlastauto; /* 600 */ 283 lastport = &pcbinfo->lastlow; 284 } else { 285 first = ipport_firstauto; /* sysctl */ 286 last = ipport_lastauto; 287 lastport = &pcbinfo->lastport; 288 } 289 /* 290 * Simple check to ensure all ports are not used up causing 291 * a deadlock here. 292 * 293 * We split the two cases (up and down) so that the direction 294 * is not being tested on each round of the loop. 295 */ 296 if (first > last) { 297 /* 298 * counting down 299 */ 300 count = first - last; 301 302 do { 303 if (count-- < 0) { /* completely used? */ 304 /* 305 * Undo any address bind that may have 306 * occurred above. 307 */ 308 inp->in6p_laddr = in6addr_any; 309 return (EAGAIN); 310 } 311 --*lastport; 312 if (*lastport > first || *lastport < last) 313 *lastport = first; 314 lport = htons(*lastport); 315 } while (in6_pcblookup_local(pcbinfo, 316 &inp->in6p_laddr, lport, wild)); 317 } else { 318 /* 319 * counting up 320 */ 321 count = last - first; 322 323 do { 324 if (count-- < 0) { /* completely used? */ 325 /* 326 * Undo any address bind that may have 327 * occurred above. 328 */ 329 inp->in6p_laddr = in6addr_any; 330 return (EAGAIN); 331 } 332 ++*lastport; 333 if (*lastport < first || *lastport > last) 334 *lastport = first; 335 lport = htons(*lastport); 336 } while (in6_pcblookup_local(pcbinfo, 337 &inp->in6p_laddr, lport, wild)); 338 } 339 } 340 inp->inp_lport = lport; 341 if (in_pcbinshash(inp) != 0) { 342 inp->in6p_laddr = in6addr_any; 343 inp->inp_lport = 0; 344 return (EAGAIN); 345 } 346 inp->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0; /*XXX*/ 347 return(0); 348} 349 350/* 351 * Transform old in6_pcbconnect() into an inner subroutine for new 352 * in6_pcbconnect(): Do some validity-checking on the remote 353 * address (in mbuf 'nam') and then determine local host address 354 * (i.e., which interface) to use to access that remote host. 355 * 356 * This preserves definition of in6_pcbconnect(), while supporting a 357 * slightly different version for T/TCP. (This is more than 358 * a bit of a kludge, but cleaning up the internal interfaces would 359 * have forced minor changes in every protocol). 360 */ 361 362int 363in6_pcbladdr(inp, nam, plocal_addr6) 364 register struct inpcb *inp; 365 struct sockaddr *nam; 366 struct in6_addr **plocal_addr6; 367{ 368 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 369 struct in6_pktinfo *pi; 370 struct ifnet *ifp = NULL; 371 int error = 0; 372 373 if (nam->sa_len != sizeof (*sin6)) 374 return (EINVAL); 375 if (sin6->sin6_family != AF_INET6) 376 return (EAFNOSUPPORT); 377 if (sin6->sin6_port == 0) 378 return (EADDRNOTAVAIL); 379 380 /* 381 * If the scope of the destination is link-local, embed the interface 382 * index in the address. 383 */ 384 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) { 385 /* XXX boundary check is assumed to be already done. */ 386 /* XXX sin6_scope_id is weaker than advanced-api. */ 387 if (inp->in6p_outputopts && 388 (pi = inp->in6p_outputopts->ip6po_pktinfo) && 389 pi->ipi6_ifindex) { 390 sin6->sin6_addr.s6_addr16[1] = htons(pi->ipi6_ifindex); 391 ifp = ifindex2ifnet[pi->ipi6_ifindex]; 392 } else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 393 inp->in6p_moptions && 394 inp->in6p_moptions->im6o_multicast_ifp) { 395 sin6->sin6_addr.s6_addr16[1] = 396 htons(inp->in6p_moptions->im6o_multicast_ifp->if_index); 397 ifp = ifindex2ifnet[inp->in6p_moptions->im6o_multicast_ifp->if_index]; 398 } else if (sin6->sin6_scope_id) { 399 /* boundary check */ 400 if (sin6->sin6_scope_id < 0 401 || if_index < sin6->sin6_scope_id) { 402 return ENXIO; /* XXX EINVAL? */ 403 } 404 sin6->sin6_addr.s6_addr16[1] 405 = htons(sin6->sin6_scope_id & 0xffff);/*XXX*/ 406 ifp = ifindex2ifnet[sin6->sin6_scope_id]; 407 } 408 } 409 410 if (in6_ifaddr) { 411 /* 412 * If the destination address is UNSPECIFIED addr, 413 * use the loopback addr, e.g ::1. 414 */ 415 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 416 sin6->sin6_addr = in6addr_loopback; 417 } 418 { 419 /* 420 * XXX: in6_selectsrc might replace the bound local address 421 * with the address specified by setsockopt(IPV6_PKTINFO). 422 * Is it the intended behavior? 423 */ 424 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts, 425 inp->in6p_moptions, 426 &inp->in6p_route, 427 &inp->in6p_laddr, &error); 428 if (*plocal_addr6 == 0) { 429 if (error == 0) 430 error = EADDRNOTAVAIL; 431 return(error); 432 } 433 /* 434 * Don't do pcblookup call here; return interface in 435 * plocal_addr6 436 * and exit to caller, that will do the lookup. 437 */ 438 } 439 440 if (inp->in6p_route.ro_rt) 441 ifp = inp->in6p_route.ro_rt->rt_ifp; 442 443 inp->in6p_ip6_hlim = (u_int8_t)in6_selecthlim(inp, ifp); 444 445 return(0); 446} 447 448/* 449 * Outer subroutine: 450 * Connect from a socket to a specified address. 451 * Both address and port must be specified in argument sin. 452 * If don't have a local address for this socket yet, 453 * then pick one. 454 */ 455int 456in6_pcbconnect(inp, nam, p) 457 register struct inpcb *inp; 458 struct sockaddr *nam; 459 struct proc *p; 460{ 461 struct in6_addr *addr6; 462 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 463 int error; 464 465 /* 466 * Call inner routine, to assign local interface address. 467 */ 468 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 469 return(error); 470 471 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr, 472 sin6->sin6_port, 473 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 474 ? addr6 : &inp->in6p_laddr, 475 inp->inp_lport, 0, NULL) != NULL) { 476 return (EADDRINUSE); 477 } 478 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 479 if (inp->inp_lport == 0) { 480 error = in6_pcbbind(inp, (struct sockaddr *)0, p); 481 if (error) 482 return (error); 483 } 484 inp->in6p_laddr = *addr6; 485 } 486 inp->in6p_faddr = sin6->sin6_addr; 487 inp->inp_fport = sin6->sin6_port; 488 /* 489 * xxx kazu flowlabel is necessary for connect? 490 * but if this line is missing, the garbage value remains. 491 */ 492 inp->in6p_flowinfo = sin6->sin6_flowinfo; 493#ifdef INET6 494 if ((inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) == 0 && 495 ip6_auto_flowlable != 0) 496 inp->in6p_flowinfo |= 497 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 498#endif 499 500 in_pcbrehash(inp); 501 return (0); 502} 503 504/* 505 * Return an IPv6 address, which is the most appropriate for given 506 * destination and user specified options. 507 * If necessary, this function lookups the routing table and return 508 * an entry to the caller for later use. 509 */ 510struct in6_addr * 511in6_selectsrc(dstsock, opts, mopts, ro, laddr, errorp) 512 struct sockaddr_in6 *dstsock; 513 struct ip6_pktopts *opts; 514 struct ip6_moptions *mopts; 515 struct route_in6 *ro; 516 struct in6_addr *laddr; 517 int *errorp; 518{ 519 struct in6_addr *dst; 520 struct in6_ifaddr *ia6 = 0; 521 struct in6_pktinfo *pi = NULL; 522 523 dst = &dstsock->sin6_addr; 524 *errorp = 0; 525 526 /* 527 * If the source address is explicitly specified by the caller, 528 * use it. 529 */ 530 if (opts && (pi = opts->ip6po_pktinfo) && 531 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) 532 return(&pi->ipi6_addr); 533 534 /* 535 * If the source address is not specified but the socket(if any) 536 * is already bound, use the bound address. 537 */ 538 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) 539 return(laddr); 540 541 /* 542 * If the caller doesn't specify the source address but 543 * the outgoing interface, use an address associated with 544 * the interface. 545 */ 546 if (pi && pi->ipi6_ifindex) { 547 /* XXX boundary check is assumed to be already done. */ 548 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex], 549 dst); 550 if (ia6 == 0) { 551 *errorp = EADDRNOTAVAIL; 552 return(0); 553 } 554 return(&satosin6(&ia6->ia_addr)->sin6_addr); 555 } 556 557 /* 558 * If the destination address is a link-local unicast address or 559 * a multicast address, and if the outgoing interface is specified 560 * by the sin6_scope_id filed, use an address associated with the 561 * interface. 562 * XXX: We're now trying to define more specific semantics of 563 * sin6_scope_id field, so this part will be rewritten in 564 * the near future. 565 */ 566 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) && 567 dstsock->sin6_scope_id) { 568 /* 569 * I'm not sure if boundary check for scope_id is done 570 * somewhere... 571 */ 572 if (dstsock->sin6_scope_id < 0 || 573 if_index < dstsock->sin6_scope_id) { 574 *errorp = ENXIO; /* XXX: better error? */ 575 return(0); 576 } 577 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id], 578 dst); 579 if (ia6 == 0) { 580 *errorp = EADDRNOTAVAIL; 581 return(0); 582 } 583 return(&satosin6(&ia6->ia_addr)->sin6_addr); 584 } 585 586 /* 587 * If the destination address is a multicast address and 588 * the outgoing interface for the address is specified 589 * by the caller, use an address associated with the interface. 590 * There is a sanity check here; if the destination has node-local 591 * scope, the outgoing interfacde should be a loopback address. 592 * Even if the outgoing interface is not specified, we also 593 * choose a loopback interface as the outgoing interface. 594 */ 595 if (IN6_IS_ADDR_MULTICAST(dst)) { 596 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL; 597 598 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) { 599 ifp = &loif[0]; 600 } 601 602 if (ifp) { 603 ia6 = in6_ifawithscope(ifp, dst); 604 if (ia6 == 0) { 605 *errorp = EADDRNOTAVAIL; 606 return(0); 607 } 608 return(&ia6->ia_addr.sin6_addr); 609 } 610 } 611 612 /* 613 * If the next hop address for the packet is specified 614 * by caller, use an address associated with the route 615 * to the next hop. 616 */ 617 { 618 struct sockaddr_in6 *sin6_next; 619 struct rtentry *rt; 620 621 if (opts && opts->ip6po_nexthop) { 622 sin6_next = satosin6(opts->ip6po_nexthop); 623 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL); 624 if (rt) { 625 ia6 = in6_ifawithscope(rt->rt_ifp, dst); 626 if (ia6 == 0) 627 ia6 = ifatoia6(rt->rt_ifa); 628 } 629 if (ia6 == 0) { 630 *errorp = EADDRNOTAVAIL; 631 return(0); 632 } 633 return(&satosin6(&ia6->ia_addr)->sin6_addr); 634 } 635 } 636 637 /* 638 * If route is known or can be allocated now, 639 * our src addr is taken from the i/f, else punt. 640 */ 641 if (ro) { 642 if (ro->ro_rt && 643 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) { 644 RTFREE(ro->ro_rt); 645 ro->ro_rt = (struct rtentry *)0; 646 } 647 if (ro->ro_rt == (struct rtentry *)0 || 648 ro->ro_rt->rt_ifp == (struct ifnet *)0) { 649 /* No route yet, so try to acquire one */ 650 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 651 ro->ro_dst.sin6_family = AF_INET6; 652 ro->ro_dst.sin6_len = sizeof(struct sockaddr_in6); 653 ro->ro_dst.sin6_addr = *dst; 654 if (IN6_IS_ADDR_MULTICAST(dst)) { 655 ro->ro_rt = rtalloc1(&((struct route *)ro) 656 ->ro_dst, 0, 0UL); 657 } else { 658 rtalloc((struct route *)ro); 659 } 660 } 661 662 /* 663 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 664 * the address. But we don't know why it does so. 665 * It is necessary to ensure the scope even for lo0 666 * so doesn't check out IFF_LOOPBACK. 667 */ 668 669 if (ro->ro_rt) { 670 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst); 671 if (ia6 == 0) /* xxx scope error ?*/ 672 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 673 } 674 if (ia6 == 0) { 675 *errorp = EHOSTUNREACH; /* no route */ 676 return(0); 677 } 678 return(&satosin6(&ia6->ia_addr)->sin6_addr); 679 } 680 681 *errorp = EADDRNOTAVAIL; 682 return(0); 683} 684 685/* 686 * Default hop limit selection. The precedence is as follows: 687 * 1. Hoplimit valued specified via ioctl. 688 * 2. (If the outgoing interface is detected) the current 689 * hop limit of the interface specified by router advertisement. 690 * 3. The system default hoplimit. 691*/ 692int 693in6_selecthlim(in6p, ifp) 694 struct in6pcb *in6p; 695 struct ifnet *ifp; 696{ 697 if (in6p && in6p->in6p_hops >= 0) 698 return(in6p->in6p_hops); 699 else if (ifp) 700 return(nd_ifinfo[ifp->if_index].chlim); 701 else 702 return(ip6_defhlim); 703} 704 705void 706in6_pcbdisconnect(inp) 707 struct inpcb *inp; 708{ 709 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 710 inp->inp_fport = 0; 711 in_pcbrehash(inp); 712 if (inp->inp_socket->so_state & SS_NOFDREF) 713 in6_pcbdetach(inp); 714} 715 716void 717in6_pcbdetach(inp) 718 struct inpcb *inp; 719{ 720 struct socket *so = inp->inp_socket; 721 struct inpcbinfo *ipi = inp->inp_pcbinfo; 722 723#ifdef IPSEC 724 if (sotoinpcb(so) != 0) 725 key_freeso(so); 726 ipsec6_delete_pcbpolicy(inp); 727#endif /* IPSEC */ 728 inp->inp_gencnt = ++ipi->ipi_gencnt; 729 in_pcbremlists(inp); 730 sotoinpcb(so) = 0; 731 sofree(so); 732 if (inp->in6p_options) 733 m_freem(inp->in6p_options); 734 if (inp->in6p_outputopts) { 735 if (inp->in6p_outputopts->ip6po_rthdr && 736 inp->in6p_outputopts->ip6po_route.ro_rt) 737 RTFREE(inp->in6p_outputopts->ip6po_route.ro_rt); 738 if (inp->in6p_outputopts->ip6po_m) 739 (void)m_free(inp->in6p_outputopts->ip6po_m); 740 free(inp->in6p_outputopts, M_IP6OPT); 741 } 742 if (inp->in6p_route.ro_rt) 743 rtfree(inp->in6p_route.ro_rt); 744 ip6_freemoptions(inp->in6p_moptions); 745 746 /* Check and free IPv4 related resources in case of mapped addr */ 747 if (inp->inp_options) 748 (void)m_free(inp->inp_options); 749 if (inp->inp_route.ro_rt) 750 rtfree(inp->inp_route.ro_rt); 751 ip_freemoptions(inp->inp_moptions); 752 753 inp->inp_vflag = 0; 754 zfreei(ipi->ipi_zone, inp); 755} 756 757/* 758 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was 759 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 760 * in struct pr_usrreqs, so that protocols can just reference then directly 761 * without the need for a wrapper function. The socket must have a valid 762 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 763 * except through a kernel programming error, so it is acceptable to panic 764 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 765 * because there actually /is/ a programming error somewhere... XXX) 766 */ 767int 768in6_setsockaddr(so, nam) 769 struct socket *so; 770 struct sockaddr **nam; 771{ 772 int s; 773 register struct inpcb *inp; 774 register struct sockaddr_in6 *sin6; 775 776 /* 777 * Do the malloc first in case it blocks. 778 */ 779 MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK); 780 bzero(sin6, sizeof *sin6); 781 sin6->sin6_family = AF_INET6; 782 sin6->sin6_len = sizeof(*sin6); 783 784 s = splnet(); 785 inp = sotoinpcb(so); 786 if (!inp) { 787 splx(s); 788 free(sin6, M_SONAME); 789 return EINVAL; 790 } 791 sin6->sin6_port = inp->inp_lport; 792 sin6->sin6_addr = inp->in6p_laddr; 793 splx(s); 794 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 795 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 796 else 797 sin6->sin6_scope_id = 0; /*XXX*/ 798 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 799 sin6->sin6_addr.s6_addr16[1] = 0; 800 801 *nam = (struct sockaddr *)sin6; 802 return 0; 803} 804 805int 806in6_setpeeraddr(so, nam) 807 struct socket *so; 808 struct sockaddr **nam; 809{ 810 int s; 811 struct inpcb *inp; 812 register struct sockaddr_in6 *sin6; 813 814 /* 815 * Do the malloc first in case it blocks. 816 */ 817 MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, M_WAITOK); 818 bzero((caddr_t)sin6, sizeof (*sin6)); 819 sin6->sin6_family = AF_INET6; 820 sin6->sin6_len = sizeof(struct sockaddr_in6); 821 822 s = splnet(); 823 inp = sotoinpcb(so); 824 if (!inp) { 825 splx(s); 826 free(sin6, M_SONAME); 827 return EINVAL; 828 } 829 sin6->sin6_port = inp->inp_fport; 830 sin6->sin6_addr = inp->in6p_faddr; 831 splx(s); 832 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 833 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]); 834 else 835 sin6->sin6_scope_id = 0; /*XXX*/ 836 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 837 sin6->sin6_addr.s6_addr16[1] = 0; 838 839 *nam = (struct sockaddr *)sin6; 840 return 0; 841} 842 843int 844in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 845{ 846 struct inpcb *inp = sotoinpcb(so); 847 int error; 848 849 if (inp == NULL) 850 return EINVAL; 851 if (inp->inp_vflag & INP_IPV4) { 852 error = in_setsockaddr(so, nam); 853 if (error == 0) 854 in6_sin_2_v4mapsin6_in_sock(nam); 855 } else 856 error = in6_setsockaddr(so, nam); 857 858 return error; 859} 860 861int 862in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 863{ 864 struct inpcb *inp = sotoinpcb(so); 865 int error; 866 867 if (inp == NULL) 868 return EINVAL; 869 if (inp->inp_vflag & INP_IPV4) { 870 error = in_setpeeraddr(so, nam); 871 if (error == 0) 872 in6_sin_2_v4mapsin6_in_sock(nam); 873 } else 874 error = in6_setpeeraddr(so, nam); 875 876 return error; 877} 878 879/* 880 * Pass some notification to all connections of a protocol 881 * associated with address dst. The local address and/or port numbers 882 * may be specified to limit the search. The "usual action" will be 883 * taken, depending on the ctlinput cmd. The caller must filter any 884 * cmds that are uninteresting (e.g., no error in the map). 885 * Call the protocol specific routine (if any) to report 886 * any errors for each matching socket. 887 * 888 * Must be called at splnet. 889 */ 890void 891in6_pcbnotify(head, dst, fport_arg, laddr6, lport_arg, cmd, notify) 892 struct inpcbhead *head; 893 struct sockaddr *dst; 894 u_int fport_arg, lport_arg; 895 struct in6_addr *laddr6; 896 int cmd; 897 void (*notify) __P((struct inpcb *, int)); 898{ 899 struct inpcb *inp, *oinp; 900 struct in6_addr faddr6; 901 u_short fport = fport_arg, lport = lport_arg; 902 int errno, s; 903 904 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6) 905 return; 906 faddr6 = ((struct sockaddr_in6 *)dst)->sin6_addr; 907 if (IN6_IS_ADDR_UNSPECIFIED(&faddr6)) 908 return; 909 910 /* 911 * Redirects go to all references to the destination, 912 * and use in_rtchange to invalidate the route cache. 913 * Dead host indications: notify all references to the destination. 914 * Otherwise, if we have knowledge of the local port and address, 915 * deliver only to that socket. 916 */ 917 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 918 fport = 0; 919 lport = 0; 920 bzero((caddr_t)laddr6, sizeof(*laddr6)); 921 if (cmd != PRC_HOSTDEAD) 922 notify = in6_rtchange; 923 } 924 errno = inet6ctlerrmap[cmd]; 925 s = splnet(); 926 for (inp = LIST_FIRST(head); inp != NULL;) { 927 if ((inp->inp_vflag & INP_IPV6) == 0) { 928 inp = LIST_NEXT(inp, inp_list); 929 continue; 930 } 931 if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &faddr6) || 932 inp->inp_socket == 0 || 933 (lport && inp->inp_lport != lport) || 934 (!IN6_IS_ADDR_UNSPECIFIED(laddr6) && 935 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr6)) || 936 (fport && inp->inp_fport != fport)) { 937 inp = LIST_NEXT(inp, inp_list); 938 continue; 939 } 940 oinp = inp; 941 inp = LIST_NEXT(inp, inp_list); 942 if (notify) 943 (*notify)(oinp, errno); 944 } 945 splx(s); 946} 947 948/* 949 * Lookup a PCB based on the local address and port. 950 */ 951struct inpcb * 952in6_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay) 953 struct inpcbinfo *pcbinfo; 954 struct in6_addr *laddr; 955 u_int lport_arg; 956 int wild_okay; 957{ 958 register struct inpcb *inp; 959 int matchwild = 3, wildcard; 960 u_short lport = lport_arg; 961 962 if (!wild_okay) { 963 struct inpcbhead *head; 964 /* 965 * Look for an unconnected (wildcard foreign addr) PCB that 966 * matches the local address and port we're looking for. 967 */ 968 head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, 969 pcbinfo->hashmask)]; 970 LIST_FOREACH(inp, head, inp_hash) { 971 if ((inp->inp_vflag & INP_IPV6) == 0) 972 continue; 973 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 974 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 975 inp->inp_lport == lport) { 976 /* 977 * Found. 978 */ 979 return (inp); 980 } 981 } 982 /* 983 * Not found. 984 */ 985 return (NULL); 986 } else { 987 struct inpcbporthead *porthash; 988 struct inpcbport *phd; 989 struct inpcb *match = NULL; 990 /* 991 * Best fit PCB lookup. 992 * 993 * First see if this local port is in use by looking on the 994 * port hash list. 995 */ 996 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 997 pcbinfo->porthashmask)]; 998 LIST_FOREACH(phd, porthash, phd_hash) { 999 if (phd->phd_port == lport) 1000 break; 1001 } 1002 if (phd != NULL) { 1003 /* 1004 * Port is in use by one or more PCBs. Look for best 1005 * fit. 1006 */ 1007 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 1008 wildcard = 0; 1009 if ((inp->inp_vflag & INP_IPV6) == 0) 1010 continue; 1011 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 1012 wildcard++; 1013 if (!IN6_IS_ADDR_UNSPECIFIED( 1014 &inp->in6p_laddr)) { 1015 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 1016 wildcard++; 1017 else if (!IN6_ARE_ADDR_EQUAL( 1018 &inp->in6p_laddr, laddr)) 1019 continue; 1020 } else { 1021 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 1022 wildcard++; 1023 } 1024 if (wildcard < matchwild) { 1025 match = inp; 1026 matchwild = wildcard; 1027 if (matchwild == 0) { 1028 break; 1029 } 1030 } 1031 } 1032 } 1033 return (match); 1034 } 1035} 1036 1037/* 1038 * Check for alternatives when higher level complains 1039 * about service problems. For now, invalidate cached 1040 * routing information. If the route was created dynamically 1041 * (by a redirect), time to try a default gateway again. 1042 */ 1043void 1044in6_losing(in6p) 1045 struct inpcb *in6p; 1046{ 1047 struct rtentry *rt; 1048 struct rt_addrinfo info; 1049 1050 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 1051 in6p->in6p_route.ro_rt = 0; 1052 bzero((caddr_t)&info, sizeof(info)); 1053 info.rti_info[RTAX_DST] = 1054 (struct sockaddr *)&in6p->in6p_route.ro_dst; 1055 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1056 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1057 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 1058 if (rt->rt_flags & RTF_DYNAMIC) 1059 (void)rtrequest(RTM_DELETE, rt_key(rt), 1060 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 1061 (struct rtentry **)0); 1062 else 1063 /* 1064 * A new route can be allocated 1065 * the next time output is attempted. 1066 */ 1067 rtfree(rt); 1068 } 1069} 1070 1071/* 1072 * After a routing change, flush old routing 1073 * and allocate a (hopefully) better one. 1074 */ 1075void 1076in6_rtchange(inp, errno) 1077 struct inpcb *inp; 1078 int errno; 1079{ 1080 if (inp->in6p_route.ro_rt) { 1081 rtfree(inp->in6p_route.ro_rt); 1082 inp->in6p_route.ro_rt = 0; 1083 /* 1084 * A new route can be allocated the next time 1085 * output is attempted. 1086 */ 1087 } 1088} 1089 1090/* 1091 * Lookup PCB in hash list. 1092 */ 1093struct inpcb * 1094in6_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, ifp) 1095 struct inpcbinfo *pcbinfo; 1096 struct in6_addr *faddr, *laddr; 1097 u_int fport_arg, lport_arg; 1098 int wildcard; 1099 struct ifnet *ifp; 1100{ 1101 struct inpcbhead *head; 1102 register struct inpcb *inp; 1103 u_short fport = fport_arg, lport = lport_arg; 1104 1105 /* 1106 * First look for an exact match. 1107 */ 1108 head = &pcbinfo->hashbase[INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, 1109 lport, fport, 1110 pcbinfo->hashmask)]; 1111 LIST_FOREACH(inp, head, inp_hash) { 1112 if ((inp->inp_vflag & INP_IPV6) == 0) 1113 continue; 1114 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 1115 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 1116 inp->inp_fport == fport && 1117 inp->inp_lport == lport) { 1118 /* 1119 * Found. 1120 */ 1121 return (inp); 1122 } 1123 } 1124 if (wildcard) { 1125 struct inpcb *local_wild = NULL; 1126 1127 head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, 1128 pcbinfo->hashmask)]; 1129 LIST_FOREACH(inp, head, inp_hash) { 1130 if ((inp->inp_vflag & INP_IPV6) == 0) 1131 continue; 1132 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 1133 inp->inp_lport == lport) { 1134#if defined(NFAITH) && NFAITH > 0 1135 if (ifp && ifp->if_type == IFT_FAITH && 1136 (inp->inp_flags & INP_FAITH) == 0) 1137 continue; 1138#endif 1139 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 1140 laddr)) 1141 return (inp); 1142 else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 1143 local_wild = inp; 1144 } 1145 } 1146 return (local_wild); 1147 } 1148 1149 /* 1150 * Not found. 1151 */ 1152 return (NULL); 1153} 1154 1155void 1156init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 1157{ 1158 struct ip6_hdr *ip; 1159 1160 ip = mtod(m, struct ip6_hdr *); 1161 bzero(sin6, sizeof(*sin6)); 1162 sin6->sin6_len = sizeof(*sin6); 1163 sin6->sin6_family = AF_INET6; 1164 sin6->sin6_addr = ip->ip6_src; 1165 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1166 sin6->sin6_addr.s6_addr16[1] = 0; 1167 sin6->sin6_scope_id = 1168 (m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) 1169 ? m->m_pkthdr.rcvif->if_index : 0; 1170 1171 return; 1172} 1173