1 /*- 2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 3 * Copyright (c) 2010-2011 Juniper Networks, Inc. 4 * All rights reserved. 5 * 6 * Portions of this software were developed by Robert N. M. Watson under 7 * contract to Juniper Networks, Inc. 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 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ 34 */ 35 36 /*- 37 * Copyright (c) 1982, 1986, 1991, 1993 38 * The Regents of the University of California. All rights reserved. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 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 */ 66 67 #include <sys/cdefs.h> 68 __FBSDID("$FreeBSD$"); 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/priv.h> 86 #include <sys/proc.h> 87 #include <sys/jail.h> 88 89 #include <vm/uma.h> 90 91 #include <net/if.h> 92 #include <net/if_types.h> 93 #include <net/route.h> 94 95 #include <netinet/in.h> 96 #include <netinet/in_var.h> 97 #include <netinet/in_systm.h> 98 #include <netinet/tcp_var.h> 99 #include <netinet/ip6.h> 100 #include <netinet/ip_var.h> 101 102 #include <netinet6/ip6_var.h> 103 #include <netinet6/nd6.h> 104 #include <netinet/in_pcb.h> 105 #include <netinet6/in6_pcb.h> 106 #include <netinet6/scope6_var.h> 107 108 struct in6_addr zeroin6_addr; 109 110 int 111 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam, 112 struct ucred *cred) 113 { 114 struct socket *so = inp->inp_socket; 115 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 116 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 117 u_short lport = 0; 118 int error, lookupflags = 0; 119 int reuseport = (so->so_options & SO_REUSEPORT); 120 121 INP_WLOCK_ASSERT(inp); 122 INP_HASH_WLOCK_ASSERT(pcbinfo); 123 124 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */ 125 return (EADDRNOTAVAIL); 126 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 127 return (EINVAL); 128 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 129 lookupflags = INPLOOKUP_WILDCARD; 130 if (nam == NULL) { 131 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 132 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 133 return (error); 134 } else { 135 sin6 = (struct sockaddr_in6 *)nam; 136 if (nam->sa_len != sizeof(*sin6)) 137 return (EINVAL); 138 /* 139 * family check. 140 */ 141 if (nam->sa_family != AF_INET6) 142 return (EAFNOSUPPORT); 143 144 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 145 return(error); 146 147 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 148 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 149 return (error); 150 151 lport = sin6->sin6_port; 152 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 153 /* 154 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 155 * allow compepte duplication of binding if 156 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 157 * and a multicast address is bound on both 158 * new and duplicated sockets. 159 */ 160 if (so->so_options & SO_REUSEADDR) 161 reuseport = SO_REUSEADDR|SO_REUSEPORT; 162 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 163 struct ifaddr *ifa; 164 165 sin6->sin6_port = 0; /* yech... */ 166 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == 167 NULL && 168 (inp->inp_flags & INP_BINDANY) == 0) { 169 return (EADDRNOTAVAIL); 170 } 171 172 /* 173 * XXX: bind to an anycast address might accidentally 174 * cause sending a packet with anycast source address. 175 * We should allow to bind to a deprecated address, since 176 * the application dares to use it. 177 */ 178 if (ifa != NULL && 179 ((struct in6_ifaddr *)ifa)->ia6_flags & 180 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 181 ifa_free(ifa); 182 return (EADDRNOTAVAIL); 183 } 184 if (ifa != NULL) 185 ifa_free(ifa); 186 } 187 if (lport) { 188 struct inpcb *t; 189 190 /* GROSS */ 191 if (ntohs(lport) <= V_ipport_reservedhigh && 192 ntohs(lport) >= V_ipport_reservedlow && 193 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 194 0)) 195 return (EACCES); 196 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 197 priv_check_cred(inp->inp_cred, 198 PRIV_NETINET_REUSEPORT, 0) != 0) { 199 t = in6_pcblookup_local(pcbinfo, 200 &sin6->sin6_addr, lport, 201 INPLOOKUP_WILDCARD, cred); 202 if (t && 203 ((t->inp_flags & INP_TIMEWAIT) == 0) && 204 (so->so_type != SOCK_STREAM || 205 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 206 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 207 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 208 (t->inp_socket->so_options & SO_REUSEPORT) 209 == 0) && (inp->inp_cred->cr_uid != 210 t->inp_cred->cr_uid)) 211 return (EADDRINUSE); 212 #ifdef INET 213 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 214 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 215 struct sockaddr_in sin; 216 217 in6_sin6_2_sin(&sin, sin6); 218 t = in_pcblookup_local(pcbinfo, 219 sin.sin_addr, lport, 220 INPLOOKUP_WILDCARD, cred); 221 if (t && 222 ((t->inp_flags & 223 INP_TIMEWAIT) == 0) && 224 (so->so_type != SOCK_STREAM || 225 ntohl(t->inp_faddr.s_addr) == 226 INADDR_ANY) && 227 (inp->inp_cred->cr_uid != 228 t->inp_cred->cr_uid)) 229 return (EADDRINUSE); 230 } 231 #endif 232 } 233 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 234 lport, lookupflags, cred); 235 if (t && (reuseport & ((t->inp_flags & INP_TIMEWAIT) ? 236 intotw(t)->tw_so_options : 237 t->inp_socket->so_options)) == 0) 238 return (EADDRINUSE); 239 #ifdef INET 240 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 241 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 242 struct sockaddr_in sin; 243 244 in6_sin6_2_sin(&sin, sin6); 245 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 246 lport, lookupflags, cred); 247 if (t && t->inp_flags & INP_TIMEWAIT) { 248 if ((reuseport & 249 intotw(t)->tw_so_options) == 0 && 250 (ntohl(t->inp_laddr.s_addr) != 251 INADDR_ANY || ((inp->inp_vflag & 252 INP_IPV6PROTO) == 253 (t->inp_vflag & INP_IPV6PROTO)))) 254 return (EADDRINUSE); 255 } 256 else if (t && 257 (reuseport & t->inp_socket->so_options) 258 == 0 && (ntohl(t->inp_laddr.s_addr) != 259 INADDR_ANY || INP_SOCKAF(so) == 260 INP_SOCKAF(t->inp_socket))) 261 return (EADDRINUSE); 262 } 263 #endif 264 } 265 inp->in6p_laddr = sin6->sin6_addr; 266 } 267 if (lport == 0) { 268 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) { 269 /* Undo an address bind that may have occurred. */ 270 inp->in6p_laddr = in6addr_any; 271 return (error); 272 } 273 } else { 274 inp->inp_lport = lport; 275 if (in_pcbinshash(inp) != 0) { 276 inp->in6p_laddr = in6addr_any; 277 inp->inp_lport = 0; 278 return (EAGAIN); 279 } 280 } 281 return (0); 282 } 283 284 /* 285 * Transform old in6_pcbconnect() into an inner subroutine for new 286 * in6_pcbconnect(): Do some validity-checking on the remote 287 * address (in mbuf 'nam') and then determine local host address 288 * (i.e., which interface) to use to access that remote host. 289 * 290 * This preserves definition of in6_pcbconnect(), while supporting a 291 * slightly different version for T/TCP. (This is more than 292 * a bit of a kludge, but cleaning up the internal interfaces would 293 * have forced minor changes in every protocol). 294 */ 295 int 296 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam, 297 struct in6_addr *plocal_addr6) 298 { 299 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 300 int error = 0; 301 struct ifnet *ifp = NULL; 302 int scope_ambiguous = 0; 303 struct in6_addr in6a; 304 305 INP_WLOCK_ASSERT(inp); 306 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */ 307 308 if (nam->sa_len != sizeof (*sin6)) 309 return (EINVAL); 310 if (sin6->sin6_family != AF_INET6) 311 return (EAFNOSUPPORT); 312 if (sin6->sin6_port == 0) 313 return (EADDRNOTAVAIL); 314 315 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 316 scope_ambiguous = 1; 317 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 318 return(error); 319 320 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) { 321 /* 322 * If the destination address is UNSPECIFIED addr, 323 * use the loopback addr, e.g ::1. 324 */ 325 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 326 sin6->sin6_addr = in6addr_loopback; 327 } 328 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 329 return (error); 330 331 error = in6_selectsrc(sin6, inp->in6p_outputopts, 332 inp, NULL, inp->inp_cred, &ifp, &in6a); 333 if (error) 334 return (error); 335 336 if (ifp && scope_ambiguous && 337 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) { 338 return(error); 339 } 340 341 /* 342 * Do not update this earlier, in case we return with an error. 343 * 344 * XXX: this in6_selectsrc result might replace the bound local 345 * address with the address specified by setsockopt(IPV6_PKTINFO). 346 * Is it the intended behavior? 347 */ 348 *plocal_addr6 = in6a; 349 350 /* 351 * Don't do pcblookup call here; return interface in 352 * plocal_addr6 353 * and exit to caller, that will do the lookup. 354 */ 355 356 return (0); 357 } 358 359 /* 360 * Outer subroutine: 361 * Connect from a socket to a specified address. 362 * Both address and port must be specified in argument sin. 363 * If don't have a local address for this socket yet, 364 * then pick one. 365 */ 366 int 367 in6_pcbconnect_mbuf(register struct inpcb *inp, struct sockaddr *nam, 368 struct ucred *cred, struct mbuf *m) 369 { 370 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 371 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 372 struct in6_addr addr6; 373 int error; 374 375 INP_WLOCK_ASSERT(inp); 376 INP_HASH_WLOCK_ASSERT(pcbinfo); 377 378 /* 379 * Call inner routine, to assign local interface address. 380 * in6_pcbladdr() may automatically fill in sin6_scope_id. 381 */ 382 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 383 return (error); 384 385 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr, 386 sin6->sin6_port, 387 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 388 ? &addr6 : &inp->in6p_laddr, 389 inp->inp_lport, 0, NULL) != NULL) { 390 return (EADDRINUSE); 391 } 392 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 393 if (inp->inp_lport == 0) { 394 error = in6_pcbbind(inp, (struct sockaddr *)0, cred); 395 if (error) 396 return (error); 397 } 398 inp->in6p_laddr = addr6; 399 } 400 inp->in6p_faddr = sin6->sin6_addr; 401 inp->inp_fport = sin6->sin6_port; 402 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 403 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 404 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 405 inp->inp_flow |= 406 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 407 408 in_pcbrehash_mbuf(inp, m); 409 410 return (0); 411 } 412 413 int 414 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) 415 { 416 417 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL)); 418 } 419 420 void 421 in6_pcbdisconnect(struct inpcb *inp) 422 { 423 424 INP_WLOCK_ASSERT(inp); 425 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); 426 427 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 428 inp->inp_fport = 0; 429 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 430 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 431 in_pcbrehash(inp); 432 } 433 434 struct sockaddr * 435 in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 436 { 437 struct sockaddr_in6 *sin6; 438 439 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 440 bzero(sin6, sizeof *sin6); 441 sin6->sin6_family = AF_INET6; 442 sin6->sin6_len = sizeof(*sin6); 443 sin6->sin6_port = port; 444 sin6->sin6_addr = *addr_p; 445 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 446 447 return (struct sockaddr *)sin6; 448 } 449 450 struct sockaddr * 451 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 452 { 453 struct sockaddr_in sin; 454 struct sockaddr_in6 *sin6_p; 455 456 bzero(&sin, sizeof sin); 457 sin.sin_family = AF_INET; 458 sin.sin_len = sizeof(sin); 459 sin.sin_port = port; 460 sin.sin_addr = *addr_p; 461 462 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 463 M_WAITOK); 464 in6_sin_2_v4mapsin6(&sin, sin6_p); 465 466 return (struct sockaddr *)sin6_p; 467 } 468 469 int 470 in6_getsockaddr(struct socket *so, struct sockaddr **nam) 471 { 472 register struct inpcb *inp; 473 struct in6_addr addr; 474 in_port_t port; 475 476 inp = sotoinpcb(so); 477 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); 478 479 INP_RLOCK(inp); 480 port = inp->inp_lport; 481 addr = inp->in6p_laddr; 482 INP_RUNLOCK(inp); 483 484 *nam = in6_sockaddr(port, &addr); 485 return 0; 486 } 487 488 int 489 in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 490 { 491 struct inpcb *inp; 492 struct in6_addr addr; 493 in_port_t port; 494 495 inp = sotoinpcb(so); 496 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 497 498 INP_RLOCK(inp); 499 port = inp->inp_fport; 500 addr = inp->in6p_faddr; 501 INP_RUNLOCK(inp); 502 503 *nam = in6_sockaddr(port, &addr); 504 return 0; 505 } 506 507 int 508 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 509 { 510 struct inpcb *inp; 511 int error; 512 513 inp = sotoinpcb(so); 514 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 515 516 #ifdef INET 517 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 518 error = in_getsockaddr(so, nam); 519 if (error == 0) 520 in6_sin_2_v4mapsin6_in_sock(nam); 521 } else 522 #endif 523 { 524 /* scope issues will be handled in in6_getsockaddr(). */ 525 error = in6_getsockaddr(so, nam); 526 } 527 528 return error; 529 } 530 531 int 532 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 533 { 534 struct inpcb *inp; 535 int error; 536 537 inp = sotoinpcb(so); 538 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 539 540 #ifdef INET 541 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 542 error = in_getpeeraddr(so, nam); 543 if (error == 0) 544 in6_sin_2_v4mapsin6_in_sock(nam); 545 } else 546 #endif 547 /* scope issues will be handled in in6_getpeeraddr(). */ 548 error = in6_getpeeraddr(so, nam); 549 550 return error; 551 } 552 553 /* 554 * Pass some notification to all connections of a protocol 555 * associated with address dst. The local address and/or port numbers 556 * may be specified to limit the search. The "usual action" will be 557 * taken, depending on the ctlinput cmd. The caller must filter any 558 * cmds that are uninteresting (e.g., no error in the map). 559 * Call the protocol specific routine (if any) to report 560 * any errors for each matching socket. 561 */ 562 void 563 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, 564 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, 565 int cmd, void *cmdarg, 566 struct inpcb *(*notify)(struct inpcb *, int)) 567 { 568 struct inpcb *inp, *inp_temp; 569 struct sockaddr_in6 sa6_src, *sa6_dst; 570 u_short fport = fport_arg, lport = lport_arg; 571 u_int32_t flowinfo; 572 int errno; 573 574 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 575 return; 576 577 sa6_dst = (struct sockaddr_in6 *)dst; 578 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 579 return; 580 581 /* 582 * note that src can be NULL when we get notify by local fragmentation. 583 */ 584 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 585 flowinfo = sa6_src.sin6_flowinfo; 586 587 /* 588 * Redirects go to all references to the destination, 589 * and use in6_rtchange to invalidate the route cache. 590 * Dead host indications: also use in6_rtchange to invalidate 591 * the cache, and deliver the error to all the sockets. 592 * Otherwise, if we have knowledge of the local port and address, 593 * deliver only to that socket. 594 */ 595 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 596 fport = 0; 597 lport = 0; 598 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 599 600 if (cmd != PRC_HOSTDEAD) 601 notify = in6_rtchange; 602 } 603 errno = inet6ctlerrmap[cmd]; 604 INP_INFO_WLOCK(pcbinfo); 605 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { 606 INP_WLOCK(inp); 607 if ((inp->inp_vflag & INP_IPV6) == 0) { 608 INP_WUNLOCK(inp); 609 continue; 610 } 611 612 /* 613 * If the error designates a new path MTU for a destination 614 * and the application (associated with this socket) wanted to 615 * know the value, notify. Note that we notify for all 616 * disconnected sockets if the corresponding application 617 * wanted. This is because some UDP applications keep sending 618 * sockets disconnected. 619 * XXX: should we avoid to notify the value to TCP sockets? 620 */ 621 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 && 622 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 623 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) { 624 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, 625 (u_int32_t *)cmdarg); 626 } 627 628 /* 629 * Detect if we should notify the error. If no source and 630 * destination ports are specifed, but non-zero flowinfo and 631 * local address match, notify the error. This is the case 632 * when the error is delivered with an encrypted buffer 633 * by ESP. Otherwise, just compare addresses and ports 634 * as usual. 635 */ 636 if (lport == 0 && fport == 0 && flowinfo && 637 inp->inp_socket != NULL && 638 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 639 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 640 goto do_notify; 641 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 642 &sa6_dst->sin6_addr) || 643 inp->inp_socket == 0 || 644 (lport && inp->inp_lport != lport) || 645 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 646 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 647 &sa6_src.sin6_addr)) || 648 (fport && inp->inp_fport != fport)) { 649 INP_WUNLOCK(inp); 650 continue; 651 } 652 653 do_notify: 654 if (notify) { 655 if ((*notify)(inp, errno)) 656 INP_WUNLOCK(inp); 657 } else 658 INP_WUNLOCK(inp); 659 } 660 INP_INFO_WUNLOCK(pcbinfo); 661 } 662 663 /* 664 * Lookup a PCB based on the local address and port. Caller must hold the 665 * hash lock. No inpcb locks or references are acquired. 666 */ 667 struct inpcb * 668 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 669 u_short lport, int lookupflags, struct ucred *cred) 670 { 671 register struct inpcb *inp; 672 int matchwild = 3, wildcard; 673 674 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, 675 ("%s: invalid lookup flags %d", __func__, lookupflags)); 676 677 INP_HASH_WLOCK_ASSERT(pcbinfo); 678 679 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) { 680 struct inpcbhead *head; 681 /* 682 * Look for an unconnected (wildcard foreign addr) PCB that 683 * matches the local address and port we're looking for. 684 */ 685 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 686 0, pcbinfo->ipi_hashmask)]; 687 LIST_FOREACH(inp, head, inp_hash) { 688 /* XXX inp locking */ 689 if ((inp->inp_vflag & INP_IPV6) == 0) 690 continue; 691 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 692 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 693 inp->inp_lport == lport) { 694 /* Found. */ 695 if (cred == NULL || 696 prison_equal_ip6(cred->cr_prison, 697 inp->inp_cred->cr_prison)) 698 return (inp); 699 } 700 } 701 /* 702 * Not found. 703 */ 704 return (NULL); 705 } else { 706 struct inpcbporthead *porthash; 707 struct inpcbport *phd; 708 struct inpcb *match = NULL; 709 /* 710 * Best fit PCB lookup. 711 * 712 * First see if this local port is in use by looking on the 713 * port hash list. 714 */ 715 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 716 pcbinfo->ipi_porthashmask)]; 717 LIST_FOREACH(phd, porthash, phd_hash) { 718 if (phd->phd_port == lport) 719 break; 720 } 721 if (phd != NULL) { 722 /* 723 * Port is in use by one or more PCBs. Look for best 724 * fit. 725 */ 726 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 727 wildcard = 0; 728 if (cred != NULL && 729 !prison_equal_ip6(cred->cr_prison, 730 inp->inp_cred->cr_prison)) 731 continue; 732 /* XXX inp locking */ 733 if ((inp->inp_vflag & INP_IPV6) == 0) 734 continue; 735 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 736 wildcard++; 737 if (!IN6_IS_ADDR_UNSPECIFIED( 738 &inp->in6p_laddr)) { 739 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 740 wildcard++; 741 else if (!IN6_ARE_ADDR_EQUAL( 742 &inp->in6p_laddr, laddr)) 743 continue; 744 } else { 745 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 746 wildcard++; 747 } 748 if (wildcard < matchwild) { 749 match = inp; 750 matchwild = wildcard; 751 if (matchwild == 0) 752 break; 753 } 754 } 755 } 756 return (match); 757 } 758 } 759 760 void 761 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 762 { 763 struct inpcb *in6p; 764 struct ip6_moptions *im6o; 765 int i, gap; 766 767 INP_INFO_RLOCK(pcbinfo); 768 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) { 769 INP_WLOCK(in6p); 770 im6o = in6p->in6p_moptions; 771 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) { 772 /* 773 * Unselect the outgoing ifp for multicast if it 774 * is being detached. 775 */ 776 if (im6o->im6o_multicast_ifp == ifp) 777 im6o->im6o_multicast_ifp = NULL; 778 /* 779 * Drop multicast group membership if we joined 780 * through the interface being detached. 781 */ 782 gap = 0; 783 for (i = 0; i < im6o->im6o_num_memberships; i++) { 784 if (im6o->im6o_membership[i]->in6m_ifp == 785 ifp) { 786 in6_mc_leave(im6o->im6o_membership[i], 787 NULL); 788 gap++; 789 } else if (gap != 0) { 790 im6o->im6o_membership[i - gap] = 791 im6o->im6o_membership[i]; 792 } 793 } 794 im6o->im6o_num_memberships -= gap; 795 } 796 INP_WUNLOCK(in6p); 797 } 798 INP_INFO_RUNLOCK(pcbinfo); 799 } 800 801 /* 802 * Check for alternatives when higher level complains 803 * about service problems. For now, invalidate cached 804 * routing information. If the route was created dynamically 805 * (by a redirect), time to try a default gateway again. 806 */ 807 void 808 in6_losing(struct inpcb *in6p) 809 { 810 811 /* 812 * We don't store route pointers in the routing table anymore 813 */ 814 return; 815 } 816 817 /* 818 * After a routing change, flush old routing 819 * and allocate a (hopefully) better one. 820 */ 821 struct inpcb * 822 in6_rtchange(struct inpcb *inp, int errno) 823 { 824 /* 825 * We don't store route pointers in the routing table anymore 826 */ 827 return inp; 828 } 829 830 /* 831 * Lookup PCB in hash list. 832 */ 833 struct inpcb * 834 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 835 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, 836 int lookupflags, struct ifnet *ifp) 837 { 838 struct inpcbhead *head; 839 struct inpcb *inp, *tmpinp; 840 u_short fport = fport_arg, lport = lport_arg; 841 int faith; 842 843 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, 844 ("%s: invalid lookup flags %d", __func__, lookupflags)); 845 846 INP_HASH_LOCK_ASSERT(pcbinfo); 847 848 if (faithprefix_p != NULL) 849 faith = (*faithprefix_p)(laddr); 850 else 851 faith = 0; 852 853 /* 854 * First look for an exact match. 855 */ 856 tmpinp = NULL; 857 head = &pcbinfo->ipi_hashbase[ 858 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport, 859 pcbinfo->ipi_hashmask)]; 860 LIST_FOREACH(inp, head, inp_hash) { 861 /* XXX inp locking */ 862 if ((inp->inp_vflag & INP_IPV6) == 0) 863 continue; 864 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 865 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 866 inp->inp_fport == fport && 867 inp->inp_lport == lport) { 868 /* 869 * XXX We should be able to directly return 870 * the inp here, without any checks. 871 * Well unless both bound with SO_REUSEPORT? 872 */ 873 if (prison_flag(inp->inp_cred, PR_IP6)) 874 return (inp); 875 if (tmpinp == NULL) 876 tmpinp = inp; 877 } 878 } 879 if (tmpinp != NULL) 880 return (tmpinp); 881 882 /* 883 * Then look for a wildcard match, if requested. 884 */ 885 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { 886 struct inpcb *local_wild = NULL, *local_exact = NULL; 887 struct inpcb *jail_wild = NULL; 888 int injail; 889 890 /* 891 * Order of socket selection - we always prefer jails. 892 * 1. jailed, non-wild. 893 * 2. jailed, wild. 894 * 3. non-jailed, non-wild. 895 * 4. non-jailed, wild. 896 */ 897 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 898 0, pcbinfo->ipi_hashmask)]; 899 LIST_FOREACH(inp, head, inp_hash) { 900 /* XXX inp locking */ 901 if ((inp->inp_vflag & INP_IPV6) == 0) 902 continue; 903 904 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 905 inp->inp_lport != lport) { 906 continue; 907 } 908 909 /* XXX inp locking */ 910 if (faith && (inp->inp_flags & INP_FAITH) == 0) 911 continue; 912 913 injail = prison_flag(inp->inp_cred, PR_IP6); 914 if (injail) { 915 if (prison_check_ip6(inp->inp_cred, 916 laddr) != 0) 917 continue; 918 } else { 919 if (local_exact != NULL) 920 continue; 921 } 922 923 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 924 if (injail) 925 return (inp); 926 else 927 local_exact = inp; 928 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 929 if (injail) 930 jail_wild = inp; 931 else 932 local_wild = inp; 933 } 934 } /* LIST_FOREACH */ 935 936 if (jail_wild != NULL) 937 return (jail_wild); 938 if (local_exact != NULL) 939 return (local_exact); 940 if (local_wild != NULL) 941 return (local_wild); 942 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */ 943 944 /* 945 * Not found. 946 */ 947 return (NULL); 948 } 949 950 /* 951 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the 952 * hash list lock, and will return the inpcb locked (i.e., requires 953 * INPLOOKUP_LOCKPCB). 954 */ 955 static struct inpcb * 956 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 957 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 958 struct ifnet *ifp) 959 { 960 struct inpcb *inp; 961 962 INP_HASH_RLOCK(pcbinfo); 963 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport, 964 (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp); 965 if (inp != NULL) { 966 in_pcbref(inp); 967 INP_HASH_RUNLOCK(pcbinfo); 968 if (lookupflags & INPLOOKUP_WLOCKPCB) { 969 INP_WLOCK(inp); 970 if (in_pcbrele_wlocked(inp)) 971 return (NULL); 972 } else if (lookupflags & INPLOOKUP_RLOCKPCB) { 973 INP_RLOCK(inp); 974 if (in_pcbrele_rlocked(inp)) 975 return (NULL); 976 } else 977 panic("%s: locking bug", __func__); 978 } else 979 INP_HASH_RUNLOCK(pcbinfo); 980 return (inp); 981 } 982 983 /* 984 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf 985 * from which a pre-calculated hash value may be extracted. 986 */ 987 struct inpcb * 988 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, 989 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp) 990 { 991 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, 992 ("%s: invalid lookup flags %d", __func__, lookupflags)); 993 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, 994 ("%s: LOCKPCB not set", __func__)); 995 996 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, 997 lookupflags, ifp)); 998 } 999 1000 struct inpcb * 1001 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1002 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 1003 struct ifnet *ifp, struct mbuf *m) 1004 { 1005 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, 1006 ("%s: invalid lookup flags %d", __func__, lookupflags)); 1007 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, 1008 ("%s: LOCKPCB not set", __func__)); 1009 1010 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, 1011 lookupflags, ifp)); 1012 } 1013 1014 void 1015 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 1016 { 1017 struct ip6_hdr *ip; 1018 1019 ip = mtod(m, struct ip6_hdr *); 1020 bzero(sin6, sizeof(*sin6)); 1021 sin6->sin6_len = sizeof(*sin6); 1022 sin6->sin6_family = AF_INET6; 1023 sin6->sin6_addr = ip->ip6_src; 1024 1025 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 1026 1027 return; 1028 } 1029