in6_pcb.c revision 221247
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 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ 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 * 4. Neither the name of the University nor the names of its contributors 45 * may be used to endorse or promote products derived from this software 46 * without specific prior written permission. 47 * 48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 58 * SUCH DAMAGE. 59 * 60 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 61 */ 62 63#include <sys/cdefs.h> 64__FBSDID("$FreeBSD: head/sys/netinet6/in6_pcb.c 221247 2011-04-30 11:04:34Z bz $"); 65 66#include "opt_inet.h" 67#include "opt_inet6.h" 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/priv.h> 82#include <sys/proc.h> 83#include <sys/jail.h> 84 85#include <vm/uma.h> 86 87#include <net/if.h> 88#include <net/if_types.h> 89#include <net/route.h> 90 91#include <netinet/in.h> 92#include <netinet/in_var.h> 93#include <netinet/in_systm.h> 94#include <netinet/tcp_var.h> 95#include <netinet/ip6.h> 96#include <netinet/ip_var.h> 97 98#include <netinet6/ip6_var.h> 99#include <netinet6/nd6.h> 100#include <netinet/in_pcb.h> 101#include <netinet6/in6_pcb.h> 102#include <netinet6/scope6_var.h> 103 104struct in6_addr zeroin6_addr; 105 106int 107in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam, 108 struct ucred *cred) 109{ 110 struct socket *so = inp->inp_socket; 111 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 112 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 113 u_short lport = 0; 114 int error, wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 115 116 INP_INFO_WLOCK_ASSERT(pcbinfo); 117 INP_WLOCK_ASSERT(inp); 118 119 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */ 120 return (EADDRNOTAVAIL); 121 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 122 return (EINVAL); 123 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 124 wild = INPLOOKUP_WILDCARD; 125 if (nam == NULL) { 126 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 127 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 128 return (error); 129 } else { 130 sin6 = (struct sockaddr_in6 *)nam; 131 if (nam->sa_len != sizeof(*sin6)) 132 return (EINVAL); 133 /* 134 * family check. 135 */ 136 if (nam->sa_family != AF_INET6) 137 return (EAFNOSUPPORT); 138 139 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 140 return(error); 141 142 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 143 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 144 return (error); 145 146 lport = sin6->sin6_port; 147 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 148 /* 149 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 150 * allow compepte duplication of binding if 151 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 152 * and a multicast address is bound on both 153 * new and duplicated sockets. 154 */ 155 if (so->so_options & SO_REUSEADDR) 156 reuseport = SO_REUSEADDR|SO_REUSEPORT; 157 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 158 struct ifaddr *ifa; 159 160 sin6->sin6_port = 0; /* yech... */ 161 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == 162 NULL && 163 (inp->inp_flags & INP_BINDANY) == 0) { 164 return (EADDRNOTAVAIL); 165 } 166 167 /* 168 * XXX: bind to an anycast address might accidentally 169 * cause sending a packet with anycast source address. 170 * We should allow to bind to a deprecated address, since 171 * the application dares to use it. 172 */ 173 if (ifa != NULL && 174 ((struct in6_ifaddr *)ifa)->ia6_flags & 175 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 176 ifa_free(ifa); 177 return (EADDRNOTAVAIL); 178 } 179 if (ifa != NULL) 180 ifa_free(ifa); 181 } 182 if (lport) { 183 struct inpcb *t; 184 185 /* GROSS */ 186 if (ntohs(lport) <= V_ipport_reservedhigh && 187 ntohs(lport) >= V_ipport_reservedlow && 188 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 189 0)) 190 return (EACCES); 191 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 192 priv_check_cred(inp->inp_cred, 193 PRIV_NETINET_REUSEPORT, 0) != 0) { 194 t = in6_pcblookup_local(pcbinfo, 195 &sin6->sin6_addr, lport, 196 INPLOOKUP_WILDCARD, cred); 197 if (t && 198 ((t->inp_flags & INP_TIMEWAIT) == 0) && 199 (so->so_type != SOCK_STREAM || 200 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 201 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 202 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 203 (t->inp_socket->so_options & SO_REUSEPORT) 204 == 0) && (inp->inp_cred->cr_uid != 205 t->inp_cred->cr_uid)) 206 return (EADDRINUSE); 207#ifdef INET 208 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 209 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 210 struct sockaddr_in sin; 211 212 in6_sin6_2_sin(&sin, sin6); 213 t = in_pcblookup_local(pcbinfo, 214 sin.sin_addr, lport, 215 INPLOOKUP_WILDCARD, cred); 216 if (t && 217 ((t->inp_flags & 218 INP_TIMEWAIT) == 0) && 219 (so->so_type != SOCK_STREAM || 220 ntohl(t->inp_faddr.s_addr) == 221 INADDR_ANY) && 222 (inp->inp_cred->cr_uid != 223 t->inp_cred->cr_uid)) 224 return (EADDRINUSE); 225 } 226#endif 227 } 228 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 229 lport, wild, cred); 230 if (t && (reuseport & ((t->inp_flags & INP_TIMEWAIT) ? 231 intotw(t)->tw_so_options : 232 t->inp_socket->so_options)) == 0) 233 return (EADDRINUSE); 234#ifdef INET 235 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 236 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 237 struct sockaddr_in sin; 238 239 in6_sin6_2_sin(&sin, sin6); 240 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 241 lport, wild, cred); 242 if (t && t->inp_flags & INP_TIMEWAIT) { 243 if ((reuseport & 244 intotw(t)->tw_so_options) == 0 && 245 (ntohl(t->inp_laddr.s_addr) != 246 INADDR_ANY || ((inp->inp_vflag & 247 INP_IPV6PROTO) == 248 (t->inp_vflag & INP_IPV6PROTO)))) 249 return (EADDRINUSE); 250 } 251 else if (t && 252 (reuseport & t->inp_socket->so_options) 253 == 0 && (ntohl(t->inp_laddr.s_addr) != 254 INADDR_ANY || INP_SOCKAF(so) == 255 INP_SOCKAF(t->inp_socket))) 256 return (EADDRINUSE); 257 } 258#endif 259 } 260 inp->in6p_laddr = sin6->sin6_addr; 261 } 262 if (lport == 0) { 263 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) { 264 /* Undo an address bind that may have occurred. */ 265 inp->in6p_laddr = in6addr_any; 266 return (error); 267 } 268 } else { 269 inp->inp_lport = lport; 270 if (in_pcbinshash(inp) != 0) { 271 inp->in6p_laddr = in6addr_any; 272 inp->inp_lport = 0; 273 return (EAGAIN); 274 } 275 } 276 return (0); 277} 278 279/* 280 * Transform old in6_pcbconnect() into an inner subroutine for new 281 * in6_pcbconnect(): Do some validity-checking on the remote 282 * address (in mbuf 'nam') and then determine local host address 283 * (i.e., which interface) to use to access that remote host. 284 * 285 * This preserves definition of in6_pcbconnect(), while supporting a 286 * slightly different version for T/TCP. (This is more than 287 * a bit of a kludge, but cleaning up the internal interfaces would 288 * have forced minor changes in every protocol). 289 */ 290int 291in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam, 292 struct in6_addr *plocal_addr6) 293{ 294 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 295 int error = 0; 296 struct ifnet *ifp = NULL; 297 int scope_ambiguous = 0; 298 struct in6_addr in6a; 299 300 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 301 INP_WLOCK_ASSERT(inp); 302 303 if (nam->sa_len != sizeof (*sin6)) 304 return (EINVAL); 305 if (sin6->sin6_family != AF_INET6) 306 return (EAFNOSUPPORT); 307 if (sin6->sin6_port == 0) 308 return (EADDRNOTAVAIL); 309 310 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 311 scope_ambiguous = 1; 312 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 313 return(error); 314 315 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) { 316 /* 317 * If the destination address is UNSPECIFIED addr, 318 * use the loopback addr, e.g ::1. 319 */ 320 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 321 sin6->sin6_addr = in6addr_loopback; 322 } 323 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 324 return (error); 325 326 error = in6_selectsrc(sin6, inp->in6p_outputopts, 327 inp, NULL, inp->inp_cred, &ifp, &in6a); 328 if (error) 329 return (error); 330 331 if (ifp && scope_ambiguous && 332 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) { 333 return(error); 334 } 335 336 /* 337 * Do not update this earlier, in case we return with an error. 338 * 339 * XXX: this in6_selectsrc result might replace the bound local 340 * address with the address specified by setsockopt(IPV6_PKTINFO). 341 * Is it the intended behavior? 342 */ 343 *plocal_addr6 = in6a; 344 345 /* 346 * Don't do pcblookup call here; return interface in 347 * plocal_addr6 348 * and exit to caller, that will do the lookup. 349 */ 350 351 return (0); 352} 353 354/* 355 * Outer subroutine: 356 * Connect from a socket to a specified address. 357 * Both address and port must be specified in argument sin. 358 * If don't have a local address for this socket yet, 359 * then pick one. 360 */ 361int 362in6_pcbconnect(register struct inpcb *inp, struct sockaddr *nam, 363 struct ucred *cred) 364{ 365 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 366 struct in6_addr addr6; 367 int error; 368 369 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 370 INP_WLOCK_ASSERT(inp); 371 372 /* 373 * Call inner routine, to assign local interface address. 374 * in6_pcbladdr() may automatically fill in sin6_scope_id. 375 */ 376 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 377 return (error); 378 379 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr, 380 sin6->sin6_port, 381 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 382 ? &addr6 : &inp->in6p_laddr, 383 inp->inp_lport, 0, NULL) != NULL) { 384 return (EADDRINUSE); 385 } 386 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 387 if (inp->inp_lport == 0) { 388 error = in6_pcbbind(inp, (struct sockaddr *)0, cred); 389 if (error) 390 return (error); 391 } 392 inp->in6p_laddr = addr6; 393 } 394 inp->in6p_faddr = sin6->sin6_addr; 395 inp->inp_fport = sin6->sin6_port; 396 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 397 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 398 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 399 inp->inp_flow |= 400 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 401 402 in_pcbrehash(inp); 403 404 return (0); 405} 406 407void 408in6_pcbdisconnect(struct inpcb *inp) 409{ 410 411 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 412 INP_WLOCK_ASSERT(inp); 413 414 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 415 inp->inp_fport = 0; 416 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 417 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 418 in_pcbrehash(inp); 419} 420 421struct sockaddr * 422in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 423{ 424 struct sockaddr_in6 *sin6; 425 426 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 427 bzero(sin6, sizeof *sin6); 428 sin6->sin6_family = AF_INET6; 429 sin6->sin6_len = sizeof(*sin6); 430 sin6->sin6_port = port; 431 sin6->sin6_addr = *addr_p; 432 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 433 434 return (struct sockaddr *)sin6; 435} 436 437struct sockaddr * 438in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 439{ 440 struct sockaddr_in sin; 441 struct sockaddr_in6 *sin6_p; 442 443 bzero(&sin, sizeof sin); 444 sin.sin_family = AF_INET; 445 sin.sin_len = sizeof(sin); 446 sin.sin_port = port; 447 sin.sin_addr = *addr_p; 448 449 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 450 M_WAITOK); 451 in6_sin_2_v4mapsin6(&sin, sin6_p); 452 453 return (struct sockaddr *)sin6_p; 454} 455 456int 457in6_getsockaddr(struct socket *so, struct sockaddr **nam) 458{ 459 register struct inpcb *inp; 460 struct in6_addr addr; 461 in_port_t port; 462 463 inp = sotoinpcb(so); 464 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); 465 466 INP_RLOCK(inp); 467 port = inp->inp_lport; 468 addr = inp->in6p_laddr; 469 INP_RUNLOCK(inp); 470 471 *nam = in6_sockaddr(port, &addr); 472 return 0; 473} 474 475int 476in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 477{ 478 struct inpcb *inp; 479 struct in6_addr addr; 480 in_port_t port; 481 482 inp = sotoinpcb(so); 483 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 484 485 INP_RLOCK(inp); 486 port = inp->inp_fport; 487 addr = inp->in6p_faddr; 488 INP_RUNLOCK(inp); 489 490 *nam = in6_sockaddr(port, &addr); 491 return 0; 492} 493 494int 495in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 496{ 497 struct inpcb *inp; 498 int error; 499 500 inp = sotoinpcb(so); 501 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 502 503#ifdef INET 504 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 505 error = in_getsockaddr(so, nam); 506 if (error == 0) 507 in6_sin_2_v4mapsin6_in_sock(nam); 508 } else 509#endif 510 { 511 /* scope issues will be handled in in6_getsockaddr(). */ 512 error = in6_getsockaddr(so, nam); 513 } 514 515 return error; 516} 517 518int 519in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 520{ 521 struct inpcb *inp; 522 int error; 523 524 inp = sotoinpcb(so); 525 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 526 527#ifdef INET 528 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 529 error = in_getpeeraddr(so, nam); 530 if (error == 0) 531 in6_sin_2_v4mapsin6_in_sock(nam); 532 } else 533#endif 534 /* scope issues will be handled in in6_getpeeraddr(). */ 535 error = in6_getpeeraddr(so, nam); 536 537 return error; 538} 539 540/* 541 * Pass some notification to all connections of a protocol 542 * associated with address dst. The local address and/or port numbers 543 * may be specified to limit the search. The "usual action" will be 544 * taken, depending on the ctlinput cmd. The caller must filter any 545 * cmds that are uninteresting (e.g., no error in the map). 546 * Call the protocol specific routine (if any) to report 547 * any errors for each matching socket. 548 */ 549void 550in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, 551 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, 552 int cmd, void *cmdarg, 553 struct inpcb *(*notify)(struct inpcb *, int)) 554{ 555 struct inpcb *inp, *inp_temp; 556 struct sockaddr_in6 sa6_src, *sa6_dst; 557 u_short fport = fport_arg, lport = lport_arg; 558 u_int32_t flowinfo; 559 int errno; 560 561 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 562 return; 563 564 sa6_dst = (struct sockaddr_in6 *)dst; 565 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 566 return; 567 568 /* 569 * note that src can be NULL when we get notify by local fragmentation. 570 */ 571 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 572 flowinfo = sa6_src.sin6_flowinfo; 573 574 /* 575 * Redirects go to all references to the destination, 576 * and use in6_rtchange to invalidate the route cache. 577 * Dead host indications: also use in6_rtchange to invalidate 578 * the cache, and deliver the error to all the sockets. 579 * Otherwise, if we have knowledge of the local port and address, 580 * deliver only to that socket. 581 */ 582 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 583 fport = 0; 584 lport = 0; 585 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 586 587 if (cmd != PRC_HOSTDEAD) 588 notify = in6_rtchange; 589 } 590 errno = inet6ctlerrmap[cmd]; 591 INP_INFO_WLOCK(pcbinfo); 592 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { 593 INP_WLOCK(inp); 594 if ((inp->inp_vflag & INP_IPV6) == 0) { 595 INP_WUNLOCK(inp); 596 continue; 597 } 598 599 /* 600 * If the error designates a new path MTU for a destination 601 * and the application (associated with this socket) wanted to 602 * know the value, notify. Note that we notify for all 603 * disconnected sockets if the corresponding application 604 * wanted. This is because some UDP applications keep sending 605 * sockets disconnected. 606 * XXX: should we avoid to notify the value to TCP sockets? 607 */ 608 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 && 609 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 610 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) { 611 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, 612 (u_int32_t *)cmdarg); 613 } 614 615 /* 616 * Detect if we should notify the error. If no source and 617 * destination ports are specifed, but non-zero flowinfo and 618 * local address match, notify the error. This is the case 619 * when the error is delivered with an encrypted buffer 620 * by ESP. Otherwise, just compare addresses and ports 621 * as usual. 622 */ 623 if (lport == 0 && fport == 0 && flowinfo && 624 inp->inp_socket != NULL && 625 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 626 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 627 goto do_notify; 628 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 629 &sa6_dst->sin6_addr) || 630 inp->inp_socket == 0 || 631 (lport && inp->inp_lport != lport) || 632 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 633 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 634 &sa6_src.sin6_addr)) || 635 (fport && inp->inp_fport != fport)) { 636 INP_WUNLOCK(inp); 637 continue; 638 } 639 640 do_notify: 641 if (notify) { 642 if ((*notify)(inp, errno)) 643 INP_WUNLOCK(inp); 644 } else 645 INP_WUNLOCK(inp); 646 } 647 INP_INFO_WUNLOCK(pcbinfo); 648} 649 650/* 651 * Lookup a PCB based on the local address and port. 652 */ 653struct inpcb * 654in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 655 u_short lport, int wild_okay, struct ucred *cred) 656{ 657 register struct inpcb *inp; 658 int matchwild = 3, wildcard; 659 660 INP_INFO_WLOCK_ASSERT(pcbinfo); 661 662 if (!wild_okay) { 663 struct inpcbhead *head; 664 /* 665 * Look for an unconnected (wildcard foreign addr) PCB that 666 * matches the local address and port we're looking for. 667 */ 668 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 669 0, pcbinfo->ipi_hashmask)]; 670 LIST_FOREACH(inp, head, inp_hash) { 671 /* XXX inp locking */ 672 if ((inp->inp_vflag & INP_IPV6) == 0) 673 continue; 674 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 675 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 676 inp->inp_lport == lport) { 677 /* Found. */ 678 if (cred == NULL || 679 prison_equal_ip6(cred->cr_prison, 680 inp->inp_cred->cr_prison)) 681 return (inp); 682 } 683 } 684 /* 685 * Not found. 686 */ 687 return (NULL); 688 } else { 689 struct inpcbporthead *porthash; 690 struct inpcbport *phd; 691 struct inpcb *match = NULL; 692 /* 693 * Best fit PCB lookup. 694 * 695 * First see if this local port is in use by looking on the 696 * port hash list. 697 */ 698 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 699 pcbinfo->ipi_porthashmask)]; 700 LIST_FOREACH(phd, porthash, phd_hash) { 701 if (phd->phd_port == lport) 702 break; 703 } 704 if (phd != NULL) { 705 /* 706 * Port is in use by one or more PCBs. Look for best 707 * fit. 708 */ 709 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 710 wildcard = 0; 711 if (cred != NULL && 712 !prison_equal_ip6(cred->cr_prison, 713 inp->inp_cred->cr_prison)) 714 continue; 715 /* XXX inp locking */ 716 if ((inp->inp_vflag & INP_IPV6) == 0) 717 continue; 718 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 719 wildcard++; 720 if (!IN6_IS_ADDR_UNSPECIFIED( 721 &inp->in6p_laddr)) { 722 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 723 wildcard++; 724 else if (!IN6_ARE_ADDR_EQUAL( 725 &inp->in6p_laddr, laddr)) 726 continue; 727 } else { 728 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 729 wildcard++; 730 } 731 if (wildcard < matchwild) { 732 match = inp; 733 matchwild = wildcard; 734 if (matchwild == 0) 735 break; 736 } 737 } 738 } 739 return (match); 740 } 741} 742 743void 744in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 745{ 746 struct inpcb *in6p; 747 struct ip6_moptions *im6o; 748 int i, gap; 749 750 INP_INFO_RLOCK(pcbinfo); 751 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) { 752 INP_WLOCK(in6p); 753 im6o = in6p->in6p_moptions; 754 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) { 755 /* 756 * Unselect the outgoing ifp for multicast if it 757 * is being detached. 758 */ 759 if (im6o->im6o_multicast_ifp == ifp) 760 im6o->im6o_multicast_ifp = NULL; 761 /* 762 * Drop multicast group membership if we joined 763 * through the interface being detached. 764 */ 765 gap = 0; 766 for (i = 0; i < im6o->im6o_num_memberships; i++) { 767 if (im6o->im6o_membership[i]->in6m_ifp == 768 ifp) { 769 in6_mc_leave(im6o->im6o_membership[i], 770 NULL); 771 gap++; 772 } else if (gap != 0) { 773 im6o->im6o_membership[i - gap] = 774 im6o->im6o_membership[i]; 775 } 776 } 777 im6o->im6o_num_memberships -= gap; 778 } 779 INP_WUNLOCK(in6p); 780 } 781 INP_INFO_RUNLOCK(pcbinfo); 782} 783 784/* 785 * Check for alternatives when higher level complains 786 * about service problems. For now, invalidate cached 787 * routing information. If the route was created dynamically 788 * (by a redirect), time to try a default gateway again. 789 */ 790void 791in6_losing(struct inpcb *in6p) 792{ 793 794 /* 795 * We don't store route pointers in the routing table anymore 796 */ 797 return; 798} 799 800/* 801 * After a routing change, flush old routing 802 * and allocate a (hopefully) better one. 803 */ 804struct inpcb * 805in6_rtchange(struct inpcb *inp, int errno) 806{ 807 /* 808 * We don't store route pointers in the routing table anymore 809 */ 810 return inp; 811} 812 813/* 814 * Lookup PCB in hash list. 815 */ 816struct inpcb * 817in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 818 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard, 819 struct ifnet *ifp) 820{ 821 struct inpcbhead *head; 822 struct inpcb *inp, *tmpinp; 823 u_short fport = fport_arg, lport = lport_arg; 824 int faith; 825 826 INP_INFO_LOCK_ASSERT(pcbinfo); 827 828 if (faithprefix_p != NULL) 829 faith = (*faithprefix_p)(laddr); 830 else 831 faith = 0; 832 833 /* 834 * First look for an exact match. 835 */ 836 tmpinp = NULL; 837 head = &pcbinfo->ipi_hashbase[ 838 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport, 839 pcbinfo->ipi_hashmask)]; 840 LIST_FOREACH(inp, head, inp_hash) { 841 /* XXX inp locking */ 842 if ((inp->inp_vflag & INP_IPV6) == 0) 843 continue; 844 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 845 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 846 inp->inp_fport == fport && 847 inp->inp_lport == lport) { 848 /* 849 * XXX We should be able to directly return 850 * the inp here, without any checks. 851 * Well unless both bound with SO_REUSEPORT? 852 */ 853 if (prison_flag(inp->inp_cred, PR_IP6)) 854 return (inp); 855 if (tmpinp == NULL) 856 tmpinp = inp; 857 } 858 } 859 if (tmpinp != NULL) 860 return (tmpinp); 861 862 /* 863 * Then look for a wildcard match, if requested. 864 */ 865 if (wildcard == INPLOOKUP_WILDCARD) { 866 struct inpcb *local_wild = NULL, *local_exact = NULL; 867 struct inpcb *jail_wild = NULL; 868 int injail; 869 870 /* 871 * Order of socket selection - we always prefer jails. 872 * 1. jailed, non-wild. 873 * 2. jailed, wild. 874 * 3. non-jailed, non-wild. 875 * 4. non-jailed, wild. 876 */ 877 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 878 0, pcbinfo->ipi_hashmask)]; 879 LIST_FOREACH(inp, head, inp_hash) { 880 /* XXX inp locking */ 881 if ((inp->inp_vflag & INP_IPV6) == 0) 882 continue; 883 884 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 885 inp->inp_lport != lport) { 886 continue; 887 } 888 889 /* XXX inp locking */ 890 if (faith && (inp->inp_flags & INP_FAITH) == 0) 891 continue; 892 893 injail = prison_flag(inp->inp_cred, PR_IP6); 894 if (injail) { 895 if (prison_check_ip6(inp->inp_cred, 896 laddr) != 0) 897 continue; 898 } else { 899 if (local_exact != NULL) 900 continue; 901 } 902 903 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 904 if (injail) 905 return (inp); 906 else 907 local_exact = inp; 908 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 909 if (injail) 910 jail_wild = inp; 911 else 912 local_wild = inp; 913 } 914 } /* LIST_FOREACH */ 915 916 if (jail_wild != NULL) 917 return (jail_wild); 918 if (local_exact != NULL) 919 return (local_exact); 920 if (local_wild != NULL) 921 return (local_wild); 922 } /* if (wildcard == INPLOOKUP_WILDCARD) */ 923 924 /* 925 * Not found. 926 */ 927 return (NULL); 928} 929 930void 931init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 932{ 933 struct ip6_hdr *ip; 934 935 ip = mtod(m, struct ip6_hdr *); 936 bzero(sin6, sizeof(*sin6)); 937 sin6->sin6_len = sizeof(*sin6); 938 sin6->sin6_family = AF_INET6; 939 sin6->sin6_addr = ip->ip6_src; 940 941 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 942 943 return; 944} 945