in6_pcb.c revision 188144
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 188144 2009-02-05 14:06:09Z jamie $"); 65 66#include "opt_inet.h" 67#include "opt_inet6.h" 68#include "opt_ipsec.h" 69#include "opt_mac.h" 70 71#include <sys/param.h> 72#include <sys/systm.h> 73#include <sys/malloc.h> 74#include <sys/mbuf.h> 75#include <sys/domain.h> 76#include <sys/protosw.h> 77#include <sys/socket.h> 78#include <sys/socketvar.h> 79#include <sys/sockio.h> 80#include <sys/errno.h> 81#include <sys/time.h> 82#include <sys/priv.h> 83#include <sys/proc.h> 84#include <sys/jail.h> 85#include <sys/vimage.h> 86 87#include <vm/uma.h> 88 89#include <net/if.h> 90#include <net/if_types.h> 91#include <net/route.h> 92 93#include <netinet/in.h> 94#include <netinet/in_var.h> 95#include <netinet/in_systm.h> 96#include <netinet/tcp_var.h> 97#include <netinet/ip6.h> 98#include <netinet/ip_var.h> 99#include <netinet/vinet.h> 100 101#include <netinet6/ip6_var.h> 102#include <netinet6/nd6.h> 103#include <netinet/in_pcb.h> 104#include <netinet6/in6_pcb.h> 105#include <netinet6/scope6_var.h> 106#include <netinet6/vinet6.h> 107 108#include <security/mac/mac_framework.h> 109 110struct in6_addr zeroin6_addr; 111 112int 113in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam, 114 struct ucred *cred) 115{ 116 INIT_VNET_INET6(inp->inp_vnet); 117 INIT_VNET_INET(inp->inp_vnet); 118 struct socket *so = inp->inp_socket; 119 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 120 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 121 u_short lport = 0; 122 int error, wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 123 124 INP_INFO_WLOCK_ASSERT(pcbinfo); 125 INP_WLOCK_ASSERT(inp); 126 127 if (!V_in6_ifaddr) /* XXX broken! */ 128 return (EADDRNOTAVAIL); 129 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 130 return (EINVAL); 131 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 132 wild = INPLOOKUP_WILDCARD; 133 if (nam) { 134 sin6 = (struct sockaddr_in6 *)nam; 135 if (nam->sa_len != sizeof(*sin6)) 136 return (EINVAL); 137 /* 138 * family check. 139 */ 140 if (nam->sa_family != AF_INET6) 141 return (EAFNOSUPPORT); 142 143 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 144 return(error); 145 146 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 147 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 148 return (error); 149 150 lport = sin6->sin6_port; 151 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 152 /* 153 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 154 * allow compepte duplication of binding if 155 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 156 * and a multicast address is bound on both 157 * new and duplicated sockets. 158 */ 159 if (so->so_options & SO_REUSEADDR) 160 reuseport = SO_REUSEADDR|SO_REUSEPORT; 161 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 162 struct ifaddr *ia = NULL; 163 164 sin6->sin6_port = 0; /* yech... */ 165 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 166 return (EADDRNOTAVAIL); 167 168 /* 169 * XXX: bind to an anycast address might accidentally 170 * cause sending a packet with anycast source address. 171 * We should allow to bind to a deprecated address, since 172 * the application dares to use it. 173 */ 174 if (ia && 175 ((struct in6_ifaddr *)ia)->ia6_flags & 176 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 177 return (EADDRNOTAVAIL); 178 } 179 } 180 if (lport) { 181 struct inpcb *t; 182 183 /* GROSS */ 184 if (ntohs(lport) <= V_ipport_reservedhigh && 185 ntohs(lport) >= V_ipport_reservedlow && 186 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 187 0)) 188 return (EACCES); 189 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 190 priv_check_cred(inp->inp_cred, 191 PRIV_NETINET_REUSEPORT, 0) != 0) { 192 t = in6_pcblookup_local(pcbinfo, 193 &sin6->sin6_addr, lport, 194 INPLOOKUP_WILDCARD, cred); 195 if (t && 196 ((t->inp_vflag & INP_TIMEWAIT) == 0) && 197 (so->so_type != SOCK_STREAM || 198 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 199 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 200 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 201 (t->inp_socket->so_options & SO_REUSEPORT) 202 == 0) && (inp->inp_cred->cr_uid != 203 t->inp_cred->cr_uid)) 204 return (EADDRINUSE); 205 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 206 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 207 struct sockaddr_in sin; 208 209 in6_sin6_2_sin(&sin, sin6); 210 t = in_pcblookup_local(pcbinfo, 211 sin.sin_addr, lport, 212 INPLOOKUP_WILDCARD, cred); 213 if (t && 214 ((t->inp_vflag & 215 INP_TIMEWAIT) == 0) && 216 (so->so_type != SOCK_STREAM || 217 ntohl(t->inp_faddr.s_addr) == 218 INADDR_ANY) && 219 (inp->inp_cred->cr_uid != 220 t->inp_cred->cr_uid)) 221 return (EADDRINUSE); 222 } 223 } 224 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 225 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 226 return (error); 227 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 228 lport, wild, cred); 229 if (t && (reuseport & ((t->inp_vflag & INP_TIMEWAIT) ? 230 intotw(t)->tw_so_options : 231 t->inp_socket->so_options)) == 0) 232 return (EADDRINUSE); 233 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 234 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 235 struct sockaddr_in sin; 236 237 in6_sin6_2_sin(&sin, sin6); 238 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 239 lport, wild, cred); 240 if (t && t->inp_vflag & INP_TIMEWAIT) { 241 if ((reuseport & 242 intotw(t)->tw_so_options) == 0 && 243 (ntohl(t->inp_laddr.s_addr) != 244 INADDR_ANY || ((inp->inp_vflag & 245 INP_IPV6PROTO) == 246 (t->inp_vflag & INP_IPV6PROTO)))) 247 return (EADDRINUSE); 248 } 249 else if (t && 250 (reuseport & t->inp_socket->so_options) 251 == 0 && (ntohl(t->inp_laddr.s_addr) != 252 INADDR_ANY || INP_SOCKAF(so) == 253 INP_SOCKAF(t->inp_socket))) 254 return (EADDRINUSE); 255 } 256 } 257 inp->in6p_laddr = sin6->sin6_addr; 258 } 259 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 260 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 261 return (error); 262 if (lport == 0) { 263 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) 264 return (error); 265 } else { 266 inp->inp_lport = lport; 267 if (in_pcbinshash(inp) != 0) { 268 inp->in6p_laddr = in6addr_any; 269 inp->inp_lport = 0; 270 return (EAGAIN); 271 } 272 } 273 return (0); 274} 275 276/* 277 * Transform old in6_pcbconnect() into an inner subroutine for new 278 * in6_pcbconnect(): Do some validity-checking on the remote 279 * address (in mbuf 'nam') and then determine local host address 280 * (i.e., which interface) to use to access that remote host. 281 * 282 * This preserves definition of in6_pcbconnect(), while supporting a 283 * slightly different version for T/TCP. (This is more than 284 * a bit of a kludge, but cleaning up the internal interfaces would 285 * have forced minor changes in every protocol). 286 */ 287int 288in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam, 289 struct in6_addr **plocal_addr6) 290{ 291 INIT_VNET_INET6(inp->inp_vnet); 292 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 293 int error = 0; 294 struct ifnet *ifp = NULL; 295 int scope_ambiguous = 0; 296 297 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 298 INP_WLOCK_ASSERT(inp); 299 300 if (nam->sa_len != sizeof (*sin6)) 301 return (EINVAL); 302 if (sin6->sin6_family != AF_INET6) 303 return (EAFNOSUPPORT); 304 if (sin6->sin6_port == 0) 305 return (EADDRNOTAVAIL); 306 307 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 308 scope_ambiguous = 1; 309 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 310 return(error); 311 312 if (V_in6_ifaddr) { 313 /* 314 * If the destination address is UNSPECIFIED addr, 315 * use the loopback addr, e.g ::1. 316 */ 317 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 318 sin6->sin6_addr = in6addr_loopback; 319 } 320 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 321 return (error); 322 323 /* 324 * XXX: in6_selectsrc might replace the bound local address 325 * with the address specified by setsockopt(IPV6_PKTINFO). 326 * Is it the intended behavior? 327 */ 328 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts, 329 inp, NULL, 330 inp->inp_cred, 331 &ifp, &error); 332 if (ifp && scope_ambiguous && 333 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) { 334 return(error); 335 } 336 337 if (*plocal_addr6 == NULL) { 338 if (error == 0) 339 error = EADDRNOTAVAIL; 340 return (error); 341 } 342 /* 343 * Don't do pcblookup call here; return interface in 344 * plocal_addr6 345 * and exit to caller, that will do the lookup. 346 */ 347 348 return (0); 349} 350 351/* 352 * Outer subroutine: 353 * Connect from a socket to a specified address. 354 * Both address and port must be specified in argument sin. 355 * If don't have a local address for this socket yet, 356 * then pick one. 357 */ 358int 359in6_pcbconnect(register struct inpcb *inp, struct sockaddr *nam, 360 struct ucred *cred) 361{ 362 struct in6_addr *addr6; 363 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 364 int error; 365 366 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 367 INP_WLOCK_ASSERT(inp); 368 369 /* 370 * Call inner routine, to assign local interface address. 371 * in6_pcbladdr() may automatically fill in sin6_scope_id. 372 */ 373 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 374 return (error); 375 376 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr, 377 sin6->sin6_port, 378 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 379 ? addr6 : &inp->in6p_laddr, 380 inp->inp_lport, 0, NULL) != NULL) { 381 return (EADDRINUSE); 382 } 383 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 384 if (inp->inp_lport == 0) { 385 error = in6_pcbbind(inp, (struct sockaddr *)0, cred); 386 if (error) 387 return (error); 388 } 389 inp->in6p_laddr = *addr6; 390 } 391 inp->in6p_faddr = sin6->sin6_addr; 392 inp->inp_fport = sin6->sin6_port; 393 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 394 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 395 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 396 inp->inp_flow |= 397 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 398 399 in_pcbrehash(inp); 400 401 return (0); 402} 403 404void 405in6_pcbdisconnect(struct inpcb *inp) 406{ 407 408 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 409 INP_WLOCK_ASSERT(inp); 410 411 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 412 inp->inp_fport = 0; 413 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 414 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 415 in_pcbrehash(inp); 416} 417 418struct sockaddr * 419in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 420{ 421 struct sockaddr_in6 *sin6; 422 423 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 424 bzero(sin6, sizeof *sin6); 425 sin6->sin6_family = AF_INET6; 426 sin6->sin6_len = sizeof(*sin6); 427 sin6->sin6_port = port; 428 sin6->sin6_addr = *addr_p; 429 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 430 431 return (struct sockaddr *)sin6; 432} 433 434struct sockaddr * 435in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 436{ 437 struct sockaddr_in sin; 438 struct sockaddr_in6 *sin6_p; 439 440 bzero(&sin, sizeof sin); 441 sin.sin_family = AF_INET; 442 sin.sin_len = sizeof(sin); 443 sin.sin_port = port; 444 sin.sin_addr = *addr_p; 445 446 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 447 M_WAITOK); 448 in6_sin_2_v4mapsin6(&sin, sin6_p); 449 450 return (struct sockaddr *)sin6_p; 451} 452 453int 454in6_getsockaddr(struct socket *so, struct sockaddr **nam) 455{ 456 register struct inpcb *inp; 457 struct in6_addr addr; 458 in_port_t port; 459 460 inp = sotoinpcb(so); 461 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); 462 463 INP_RLOCK(inp); 464 port = inp->inp_lport; 465 addr = inp->in6p_laddr; 466 INP_RUNLOCK(inp); 467 468 *nam = in6_sockaddr(port, &addr); 469 return 0; 470} 471 472int 473in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 474{ 475 struct inpcb *inp; 476 struct in6_addr addr; 477 in_port_t port; 478 479 inp = sotoinpcb(so); 480 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 481 482 INP_RLOCK(inp); 483 port = inp->inp_fport; 484 addr = inp->in6p_faddr; 485 INP_RUNLOCK(inp); 486 487 *nam = in6_sockaddr(port, &addr); 488 return 0; 489} 490 491int 492in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 493{ 494 struct inpcb *inp; 495 int error; 496 497 inp = sotoinpcb(so); 498 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 499 500 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 501 error = in_getsockaddr(so, nam); 502 if (error == 0) 503 in6_sin_2_v4mapsin6_in_sock(nam); 504 } else { 505 /* scope issues will be handled in in6_getsockaddr(). */ 506 error = in6_getsockaddr(so, nam); 507 } 508 509 return error; 510} 511 512int 513in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 514{ 515 struct inpcb *inp; 516 int error; 517 518 inp = sotoinpcb(so); 519 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 520 521 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 522 error = in_getpeeraddr(so, nam); 523 if (error == 0) 524 in6_sin_2_v4mapsin6_in_sock(nam); 525 } else 526 /* scope issues will be handled in in6_getpeeraddr(). */ 527 error = in6_getpeeraddr(so, nam); 528 529 return error; 530} 531 532/* 533 * Pass some notification to all connections of a protocol 534 * associated with address dst. The local address and/or port numbers 535 * may be specified to limit the search. The "usual action" will be 536 * taken, depending on the ctlinput cmd. The caller must filter any 537 * cmds that are uninteresting (e.g., no error in the map). 538 * Call the protocol specific routine (if any) to report 539 * any errors for each matching socket. 540 */ 541void 542in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, 543 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, 544 int cmd, void *cmdarg, 545 struct inpcb *(*notify)(struct inpcb *, int)) 546{ 547 struct inpcb *inp, *inp_temp; 548 struct sockaddr_in6 sa6_src, *sa6_dst; 549 u_short fport = fport_arg, lport = lport_arg; 550 u_int32_t flowinfo; 551 int errno; 552 553 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 554 return; 555 556 sa6_dst = (struct sockaddr_in6 *)dst; 557 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 558 return; 559 560 /* 561 * note that src can be NULL when we get notify by local fragmentation. 562 */ 563 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 564 flowinfo = sa6_src.sin6_flowinfo; 565 566 /* 567 * Redirects go to all references to the destination, 568 * and use in6_rtchange to invalidate the route cache. 569 * Dead host indications: also use in6_rtchange to invalidate 570 * the cache, and deliver the error to all the sockets. 571 * Otherwise, if we have knowledge of the local port and address, 572 * deliver only to that socket. 573 */ 574 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 575 fport = 0; 576 lport = 0; 577 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 578 579 if (cmd != PRC_HOSTDEAD) 580 notify = in6_rtchange; 581 } 582 errno = inet6ctlerrmap[cmd]; 583 INP_INFO_WLOCK(pcbinfo); 584 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { 585 INP_WLOCK(inp); 586 if ((inp->inp_vflag & INP_IPV6) == 0) { 587 INP_WUNLOCK(inp); 588 continue; 589 } 590 591 /* 592 * If the error designates a new path MTU for a destination 593 * and the application (associated with this socket) wanted to 594 * know the value, notify. Note that we notify for all 595 * disconnected sockets if the corresponding application 596 * wanted. This is because some UDP applications keep sending 597 * sockets disconnected. 598 * XXX: should we avoid to notify the value to TCP sockets? 599 */ 600 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 && 601 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 602 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) { 603 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, 604 (u_int32_t *)cmdarg); 605 } 606 607 /* 608 * Detect if we should notify the error. If no source and 609 * destination ports are specifed, but non-zero flowinfo and 610 * local address match, notify the error. This is the case 611 * when the error is delivered with an encrypted buffer 612 * by ESP. Otherwise, just compare addresses and ports 613 * as usual. 614 */ 615 if (lport == 0 && fport == 0 && flowinfo && 616 inp->inp_socket != NULL && 617 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 618 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 619 goto do_notify; 620 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 621 &sa6_dst->sin6_addr) || 622 inp->inp_socket == 0 || 623 (lport && inp->inp_lport != lport) || 624 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 625 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 626 &sa6_src.sin6_addr)) || 627 (fport && inp->inp_fport != fport)) { 628 INP_WUNLOCK(inp); 629 continue; 630 } 631 632 do_notify: 633 if (notify) { 634 if ((*notify)(inp, errno)) 635 INP_WUNLOCK(inp); 636 } else 637 INP_WUNLOCK(inp); 638 } 639 INP_INFO_WUNLOCK(pcbinfo); 640} 641 642/* 643 * Lookup a PCB based on the local address and port. 644 */ 645struct inpcb * 646in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 647 u_short lport, int wild_okay, struct ucred *cred) 648{ 649 register struct inpcb *inp; 650 int matchwild = 3, wildcard; 651 652 INP_INFO_WLOCK_ASSERT(pcbinfo); 653 654 if (!wild_okay) { 655 struct inpcbhead *head; 656 /* 657 * Look for an unconnected (wildcard foreign addr) PCB that 658 * matches the local address and port we're looking for. 659 */ 660 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 661 0, pcbinfo->ipi_hashmask)]; 662 LIST_FOREACH(inp, head, inp_hash) { 663 /* XXX inp locking */ 664 if ((inp->inp_vflag & INP_IPV6) == 0) 665 continue; 666 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 667 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 668 inp->inp_lport == lport) { 669 /* Found. */ 670 if (cred == NULL || 671 inp->inp_cred->cr_prison == cred->cr_prison) 672 return (inp); 673 } 674 } 675 /* 676 * Not found. 677 */ 678 return (NULL); 679 } else { 680 struct inpcbporthead *porthash; 681 struct inpcbport *phd; 682 struct inpcb *match = NULL; 683 /* 684 * Best fit PCB lookup. 685 * 686 * First see if this local port is in use by looking on the 687 * port hash list. 688 */ 689 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 690 pcbinfo->ipi_porthashmask)]; 691 LIST_FOREACH(phd, porthash, phd_hash) { 692 if (phd->phd_port == lport) 693 break; 694 } 695 if (phd != NULL) { 696 /* 697 * Port is in use by one or more PCBs. Look for best 698 * fit. 699 */ 700 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 701 wildcard = 0; 702 if (cred != NULL && 703 inp->inp_cred->cr_prison != cred->cr_prison) 704 continue; 705 /* XXX inp locking */ 706 if ((inp->inp_vflag & INP_IPV6) == 0) 707 continue; 708 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 709 wildcard++; 710 if (!IN6_IS_ADDR_UNSPECIFIED( 711 &inp->in6p_laddr)) { 712 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 713 wildcard++; 714 else if (!IN6_ARE_ADDR_EQUAL( 715 &inp->in6p_laddr, laddr)) 716 continue; 717 } else { 718 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 719 wildcard++; 720 } 721 if (wildcard < matchwild) { 722 match = inp; 723 matchwild = wildcard; 724 if (matchwild == 0) 725 break; 726 } 727 } 728 } 729 return (match); 730 } 731} 732 733void 734in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 735{ 736 struct inpcb *in6p; 737 struct ip6_moptions *im6o; 738 struct in6_multi_mship *imm, *nimm; 739 740 INP_INFO_RLOCK(pcbinfo); 741 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) { 742 INP_WLOCK(in6p); 743 im6o = in6p->in6p_moptions; 744 if ((in6p->inp_vflag & INP_IPV6) && 745 im6o) { 746 /* 747 * Unselect the outgoing interface if it is being 748 * detached. 749 */ 750 if (im6o->im6o_multicast_ifp == ifp) 751 im6o->im6o_multicast_ifp = NULL; 752 753 /* 754 * Drop multicast group membership if we joined 755 * through the interface being detached. 756 * XXX controversial - is it really legal for kernel 757 * to force this? 758 */ 759 for (imm = im6o->im6o_memberships.lh_first; 760 imm != NULL; imm = nimm) { 761 nimm = imm->i6mm_chain.le_next; 762 if (imm->i6mm_maddr->in6m_ifp == ifp) { 763 LIST_REMOVE(imm, i6mm_chain); 764 in6_delmulti(imm->i6mm_maddr); 765 free(imm, M_IP6MADDR); 766 } 767 } 768 } 769 INP_WUNLOCK(in6p); 770 } 771 INP_INFO_RUNLOCK(pcbinfo); 772} 773 774/* 775 * Check for alternatives when higher level complains 776 * about service problems. For now, invalidate cached 777 * routing information. If the route was created dynamically 778 * (by a redirect), time to try a default gateway again. 779 */ 780void 781in6_losing(struct inpcb *in6p) 782{ 783 784 /* 785 * We don't store route pointers in the routing table anymore 786 */ 787 return; 788} 789 790/* 791 * After a routing change, flush old routing 792 * and allocate a (hopefully) better one. 793 */ 794struct inpcb * 795in6_rtchange(struct inpcb *inp, int errno) 796{ 797 /* 798 * We don't store route pointers in the routing table anymore 799 */ 800 return inp; 801} 802 803/* 804 * Lookup PCB in hash list. 805 */ 806struct inpcb * 807in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 808 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard, 809 struct ifnet *ifp) 810{ 811 struct inpcbhead *head; 812 struct inpcb *inp, *tmpinp; 813 u_short fport = fport_arg, lport = lport_arg; 814 int faith; 815 816 INP_INFO_LOCK_ASSERT(pcbinfo); 817 818 if (faithprefix_p != NULL) 819 faith = (*faithprefix_p)(laddr); 820 else 821 faith = 0; 822 823 /* 824 * First look for an exact match. 825 */ 826 tmpinp = NULL; 827 head = &pcbinfo->ipi_hashbase[ 828 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport, 829 pcbinfo->ipi_hashmask)]; 830 LIST_FOREACH(inp, head, inp_hash) { 831 /* XXX inp locking */ 832 if ((inp->inp_vflag & INP_IPV6) == 0) 833 continue; 834 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 835 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 836 inp->inp_fport == fport && 837 inp->inp_lport == lport) { 838 /* 839 * XXX We should be able to directly return 840 * the inp here, without any checks. 841 * Well unless both bound with SO_REUSEPORT? 842 */ 843 if (jailed(inp->inp_cred)) 844 return (inp); 845 if (tmpinp == NULL) 846 tmpinp = inp; 847 } 848 } 849 if (tmpinp != NULL) 850 return (tmpinp); 851 852 /* 853 * Then look for a wildcard match, if requested. 854 */ 855 if (wildcard == INPLOOKUP_WILDCARD) { 856 struct inpcb *local_wild = NULL, *local_exact = NULL; 857 struct inpcb *jail_wild = NULL; 858 int injail; 859 860 /* 861 * Order of socket selection - we always prefer jails. 862 * 1. jailed, non-wild. 863 * 2. jailed, wild. 864 * 3. non-jailed, non-wild. 865 * 4. non-jailed, wild. 866 */ 867 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 868 0, pcbinfo->ipi_hashmask)]; 869 LIST_FOREACH(inp, head, inp_hash) { 870 /* XXX inp locking */ 871 if ((inp->inp_vflag & INP_IPV6) == 0) 872 continue; 873 874 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 875 inp->inp_lport != lport) { 876 continue; 877 } 878 879 /* XXX inp locking */ 880 if (faith && (inp->inp_flags & INP_FAITH) == 0) 881 continue; 882 883 injail = jailed(inp->inp_cred); 884 if (injail) { 885 if (prison_check_ip6(inp->inp_cred, 886 laddr) != 0) 887 continue; 888 } else { 889 if (local_exact != NULL) 890 continue; 891 } 892 893 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 894 if (injail) 895 return (inp); 896 else 897 local_exact = inp; 898 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 899 if (injail) 900 jail_wild = inp; 901 else 902 local_wild = inp; 903 } 904 } /* LIST_FOREACH */ 905 906 if (jail_wild != NULL) 907 return (jail_wild); 908 if (local_exact != NULL) 909 return (local_exact); 910 if (local_wild != NULL) 911 return (local_wild); 912 } /* if (wildcard == INPLOOKUP_WILDCARD) */ 913 914 /* 915 * Not found. 916 */ 917 return (NULL); 918} 919 920void 921init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 922{ 923 struct ip6_hdr *ip; 924 925 ip = mtod(m, struct ip6_hdr *); 926 bzero(sin6, sizeof(*sin6)); 927 sin6->sin6_len = sizeof(*sin6); 928 sin6->sin6_family = AF_INET6; 929 sin6->sin6_addr = ip->ip6_src; 930 931 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 932 933 return; 934} 935