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("$FreeBSD$"); 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 #include <sys/vimage.h> 85 86 #include <vm/uma.h> 87 88 #include <net/if.h> 89 #include <net/if_types.h> 90 #include <net/route.h> 91 92 #include <netinet/in.h> 93 #include <netinet/in_var.h> 94 #include <netinet/in_systm.h> 95 #include <netinet/tcp_var.h> 96 #include <netinet/ip6.h> 97 #include <netinet/ip_var.h> 98 #include <netinet/vinet.h> 99 100 #include <netinet6/ip6_var.h> 101 #include <netinet6/nd6.h> 102 #include <netinet/in_pcb.h> 103 #include <netinet6/in6_pcb.h> 104 #include <netinet6/scope6_var.h> 105 #include <netinet6/vinet6.h> 106 107 #include <security/mac/mac_framework.h> 108 109 struct in6_addr zeroin6_addr; 110 111 int 112 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam, 113 struct ucred *cred) 114 { 115 INIT_VNET_INET6(inp->inp_vnet); 116 INIT_VNET_INET(inp->inp_vnet); 117 struct socket *so = inp->inp_socket; 118 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 119 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 120 u_short lport = 0; 121 int error, wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 122 123 INP_INFO_WLOCK_ASSERT(pcbinfo); 124 INP_WLOCK_ASSERT(inp); 125 126 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */ 127 return (EADDRNOTAVAIL); 128 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 129 return (EINVAL); 130 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 131 wild = INPLOOKUP_WILDCARD; 132 if (nam == NULL) { 133 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 134 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 135 return (error); 136 } else { 137 sin6 = (struct sockaddr_in6 *)nam; 138 if (nam->sa_len != sizeof(*sin6)) 139 return (EINVAL); 140 /* 141 * family check. 142 */ 143 if (nam->sa_family != AF_INET6) 144 return (EAFNOSUPPORT); 145 146 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 147 return(error); 148 149 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 150 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 151 return (error); 152 153 lport = sin6->sin6_port; 154 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 155 /* 156 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 157 * allow compepte duplication of binding if 158 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 159 * and a multicast address is bound on both 160 * new and duplicated sockets. 161 */ 162 if (so->so_options & SO_REUSEADDR) 163 reuseport = SO_REUSEADDR|SO_REUSEPORT; 164 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 165 struct ifaddr *ifa; 166 167 sin6->sin6_port = 0; /* yech... */ 168 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == 169 NULL && 170 (inp->inp_flags & INP_BINDANY) == 0) { 171 return (EADDRNOTAVAIL); 172 } 173 174 /* 175 * XXX: bind to an anycast address might accidentally 176 * cause sending a packet with anycast source address. 177 * We should allow to bind to a deprecated address, since 178 * the application dares to use it. 179 */ 180 if (ifa != NULL && 181 ((struct in6_ifaddr *)ifa)->ia6_flags & 182 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 183 ifa_free(ifa); 184 return (EADDRNOTAVAIL); 185 } 186 if (ifa != NULL) 187 ifa_free(ifa); 188 } 189 if (lport) { 190 struct inpcb *t; 191 192 /* GROSS */ 193 if (ntohs(lport) <= V_ipport_reservedhigh && 194 ntohs(lport) >= V_ipport_reservedlow && 195 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 196 0)) 197 return (EACCES); 198 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 199 priv_check_cred(inp->inp_cred, 200 PRIV_NETINET_REUSEPORT, 0) != 0) { 201 t = in6_pcblookup_local(pcbinfo, 202 &sin6->sin6_addr, lport, 203 INPLOOKUP_WILDCARD, cred); 204 if (t && 205 ((t->inp_flags & INP_TIMEWAIT) == 0) && 206 (so->so_type != SOCK_STREAM || 207 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 208 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 209 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 210 (t->inp_socket->so_options & SO_REUSEPORT) 211 == 0) && (inp->inp_cred->cr_uid != 212 t->inp_cred->cr_uid)) 213 return (EADDRINUSE); 214 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 215 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 216 struct sockaddr_in sin; 217 218 in6_sin6_2_sin(&sin, sin6); 219 t = in_pcblookup_local(pcbinfo, 220 sin.sin_addr, lport, 221 INPLOOKUP_WILDCARD, cred); 222 if (t && 223 ((t->inp_flags & 224 INP_TIMEWAIT) == 0) && 225 (so->so_type != SOCK_STREAM || 226 ntohl(t->inp_faddr.s_addr) == 227 INADDR_ANY) && 228 (inp->inp_cred->cr_uid != 229 t->inp_cred->cr_uid)) 230 return (EADDRINUSE); 231 } 232 } 233 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 234 lport, wild, 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 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 240 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 241 struct sockaddr_in sin; 242 243 in6_sin6_2_sin(&sin, sin6); 244 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 245 lport, wild, cred); 246 if (t && t->inp_flags & INP_TIMEWAIT) { 247 if ((reuseport & 248 intotw(t)->tw_so_options) == 0 && 249 (ntohl(t->inp_laddr.s_addr) != 250 INADDR_ANY || ((inp->inp_vflag & 251 INP_IPV6PROTO) == 252 (t->inp_vflag & INP_IPV6PROTO)))) 253 return (EADDRINUSE); 254 } 255 else if (t && 256 (reuseport & t->inp_socket->so_options) 257 == 0 && (ntohl(t->inp_laddr.s_addr) != 258 INADDR_ANY || INP_SOCKAF(so) == 259 INP_SOCKAF(t->inp_socket))) 260 return (EADDRINUSE); 261 } 262 } 263 inp->in6p_laddr = sin6->sin6_addr; 264 } 265 if (lport == 0) { 266 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) 267 return (error); 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 */ 290 int 291 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam, 292 struct in6_addr *plocal_addr6) 293 { 294 INIT_VNET_INET6(inp->inp_vnet); 295 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 296 int error = 0; 297 struct ifnet *ifp = NULL; 298 int scope_ambiguous = 0; 299 struct in6_addr in6a; 300 301 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 302 INP_WLOCK_ASSERT(inp); 303 304 if (nam->sa_len != sizeof (*sin6)) 305 return (EINVAL); 306 if (sin6->sin6_family != AF_INET6) 307 return (EAFNOSUPPORT); 308 if (sin6->sin6_port == 0) 309 return (EADDRNOTAVAIL); 310 311 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 312 scope_ambiguous = 1; 313 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 314 return(error); 315 316 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) { 317 /* 318 * If the destination address is UNSPECIFIED addr, 319 * use the loopback addr, e.g ::1. 320 */ 321 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 322 sin6->sin6_addr = in6addr_loopback; 323 } 324 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 325 return (error); 326 327 error = in6_selectsrc(sin6, inp->in6p_outputopts, 328 inp, NULL, inp->inp_cred, &ifp, &in6a); 329 if (error) 330 return (error); 331 332 if (ifp && scope_ambiguous && 333 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) { 334 return(error); 335 } 336 337 /* 338 * Do not update this earlier, in case we return with an error. 339 * 340 * XXX: this in6_selectsrc result might replace the bound local 341 * aaddress with the address specified by setsockopt(IPV6_PKTINFO). 342 * Is it the intended behavior? 343 */ 344 *plocal_addr6 = in6a; 345 346 /* 347 * Don't do pcblookup call here; return interface in 348 * plocal_addr6 349 * and exit to caller, that will do the lookup. 350 */ 351 352 return (0); 353 } 354 355 /* 356 * Outer subroutine: 357 * Connect from a socket to a specified address. 358 * Both address and port must be specified in argument sin. 359 * If don't have a local address for this socket yet, 360 * then pick one. 361 */ 362 int 363 in6_pcbconnect(register struct inpcb *inp, struct sockaddr *nam, 364 struct ucred *cred) 365 { 366 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 367 struct in6_addr addr6; 368 int error; 369 370 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 371 INP_WLOCK_ASSERT(inp); 372 373 /* 374 * Call inner routine, to assign local interface address. 375 * in6_pcbladdr() may automatically fill in sin6_scope_id. 376 */ 377 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0) 378 return (error); 379 380 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr, 381 sin6->sin6_port, 382 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 383 ? &addr6 : &inp->in6p_laddr, 384 inp->inp_lport, 0, NULL) != NULL) { 385 return (EADDRINUSE); 386 } 387 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 388 if (inp->inp_lport == 0) { 389 error = in6_pcbbind(inp, (struct sockaddr *)0, cred); 390 if (error) 391 return (error); 392 } 393 inp->in6p_laddr = addr6; 394 } 395 inp->in6p_faddr = sin6->sin6_addr; 396 inp->inp_fport = sin6->sin6_port; 397 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 398 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 399 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 400 inp->inp_flow |= 401 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 402 403 in_pcbrehash(inp); 404 405 return (0); 406 } 407 408 void 409 in6_pcbdisconnect(struct inpcb *inp) 410 { 411 412 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 413 INP_WLOCK_ASSERT(inp); 414 415 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr)); 416 inp->inp_fport = 0; 417 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 418 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 419 in_pcbrehash(inp); 420 } 421 422 struct sockaddr * 423 in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 424 { 425 struct sockaddr_in6 *sin6; 426 427 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 428 bzero(sin6, sizeof *sin6); 429 sin6->sin6_family = AF_INET6; 430 sin6->sin6_len = sizeof(*sin6); 431 sin6->sin6_port = port; 432 sin6->sin6_addr = *addr_p; 433 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 434 435 return (struct sockaddr *)sin6; 436 } 437 438 struct sockaddr * 439 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 440 { 441 struct sockaddr_in sin; 442 struct sockaddr_in6 *sin6_p; 443 444 bzero(&sin, sizeof sin); 445 sin.sin_family = AF_INET; 446 sin.sin_len = sizeof(sin); 447 sin.sin_port = port; 448 sin.sin_addr = *addr_p; 449 450 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 451 M_WAITOK); 452 in6_sin_2_v4mapsin6(&sin, sin6_p); 453 454 return (struct sockaddr *)sin6_p; 455 } 456 457 int 458 in6_getsockaddr(struct socket *so, struct sockaddr **nam) 459 { 460 register struct inpcb *inp; 461 struct in6_addr addr; 462 in_port_t port; 463 464 inp = sotoinpcb(so); 465 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); 466 467 INP_RLOCK(inp); 468 port = inp->inp_lport; 469 addr = inp->in6p_laddr; 470 INP_RUNLOCK(inp); 471 472 *nam = in6_sockaddr(port, &addr); 473 return 0; 474 } 475 476 int 477 in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 478 { 479 struct inpcb *inp; 480 struct in6_addr addr; 481 in_port_t port; 482 483 inp = sotoinpcb(so); 484 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 485 486 INP_RLOCK(inp); 487 port = inp->inp_fport; 488 addr = inp->in6p_faddr; 489 INP_RUNLOCK(inp); 490 491 *nam = in6_sockaddr(port, &addr); 492 return 0; 493 } 494 495 int 496 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 497 { 498 struct inpcb *inp; 499 int error; 500 501 inp = sotoinpcb(so); 502 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 503 504 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 /* scope issues will be handled in in6_getsockaddr(). */ 510 error = in6_getsockaddr(so, nam); 511 } 512 513 return error; 514 } 515 516 int 517 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 518 { 519 struct inpcb *inp; 520 int error; 521 522 inp = sotoinpcb(so); 523 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 524 525 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 526 error = in_getpeeraddr(so, nam); 527 if (error == 0) 528 in6_sin_2_v4mapsin6_in_sock(nam); 529 } else 530 /* scope issues will be handled in in6_getpeeraddr(). */ 531 error = in6_getpeeraddr(so, nam); 532 533 return error; 534 } 535 536 /* 537 * Pass some notification to all connections of a protocol 538 * associated with address dst. The local address and/or port numbers 539 * may be specified to limit the search. The "usual action" will be 540 * taken, depending on the ctlinput cmd. The caller must filter any 541 * cmds that are uninteresting (e.g., no error in the map). 542 * Call the protocol specific routine (if any) to report 543 * any errors for each matching socket. 544 */ 545 void 546 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, 547 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, 548 int cmd, void *cmdarg, 549 struct inpcb *(*notify)(struct inpcb *, int)) 550 { 551 struct inpcb *inp, *inp_temp; 552 struct sockaddr_in6 sa6_src, *sa6_dst; 553 u_short fport = fport_arg, lport = lport_arg; 554 u_int32_t flowinfo; 555 int errno; 556 557 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) 558 return; 559 560 sa6_dst = (struct sockaddr_in6 *)dst; 561 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 562 return; 563 564 /* 565 * note that src can be NULL when we get notify by local fragmentation. 566 */ 567 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src; 568 flowinfo = sa6_src.sin6_flowinfo; 569 570 /* 571 * Redirects go to all references to the destination, 572 * and use in6_rtchange to invalidate the route cache. 573 * Dead host indications: also use in6_rtchange to invalidate 574 * the cache, and deliver the error to all the sockets. 575 * Otherwise, if we have knowledge of the local port and address, 576 * deliver only to that socket. 577 */ 578 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 579 fport = 0; 580 lport = 0; 581 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 582 583 if (cmd != PRC_HOSTDEAD) 584 notify = in6_rtchange; 585 } 586 errno = inet6ctlerrmap[cmd]; 587 INP_INFO_WLOCK(pcbinfo); 588 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { 589 INP_WLOCK(inp); 590 if ((inp->inp_vflag & INP_IPV6) == 0) { 591 INP_WUNLOCK(inp); 592 continue; 593 } 594 595 /* 596 * If the error designates a new path MTU for a destination 597 * and the application (associated with this socket) wanted to 598 * know the value, notify. Note that we notify for all 599 * disconnected sockets if the corresponding application 600 * wanted. This is because some UDP applications keep sending 601 * sockets disconnected. 602 * XXX: should we avoid to notify the value to TCP sockets? 603 */ 604 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 && 605 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 606 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) { 607 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, 608 (u_int32_t *)cmdarg); 609 } 610 611 /* 612 * Detect if we should notify the error. If no source and 613 * destination ports are specifed, but non-zero flowinfo and 614 * local address match, notify the error. This is the case 615 * when the error is delivered with an encrypted buffer 616 * by ESP. Otherwise, just compare addresses and ports 617 * as usual. 618 */ 619 if (lport == 0 && fport == 0 && flowinfo && 620 inp->inp_socket != NULL && 621 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 622 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 623 goto do_notify; 624 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 625 &sa6_dst->sin6_addr) || 626 inp->inp_socket == 0 || 627 (lport && inp->inp_lport != lport) || 628 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 629 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 630 &sa6_src.sin6_addr)) || 631 (fport && inp->inp_fport != fport)) { 632 INP_WUNLOCK(inp); 633 continue; 634 } 635 636 do_notify: 637 if (notify) { 638 if ((*notify)(inp, errno)) 639 INP_WUNLOCK(inp); 640 } else 641 INP_WUNLOCK(inp); 642 } 643 INP_INFO_WUNLOCK(pcbinfo); 644 } 645 646 /* 647 * Lookup a PCB based on the local address and port. 648 */ 649 struct inpcb * 650 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 651 u_short lport, int wild_okay, struct ucred *cred) 652 { 653 register struct inpcb *inp; 654 int matchwild = 3, wildcard; 655 656 INP_INFO_WLOCK_ASSERT(pcbinfo); 657 658 if (!wild_okay) { 659 struct inpcbhead *head; 660 /* 661 * Look for an unconnected (wildcard foreign addr) PCB that 662 * matches the local address and port we're looking for. 663 */ 664 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 665 0, pcbinfo->ipi_hashmask)]; 666 LIST_FOREACH(inp, head, inp_hash) { 667 /* XXX inp locking */ 668 if ((inp->inp_vflag & INP_IPV6) == 0) 669 continue; 670 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 671 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 672 inp->inp_lport == lport) { 673 /* Found. */ 674 if (cred == NULL || 675 prison_equal_ip6(cred->cr_prison, 676 inp->inp_cred->cr_prison)) 677 return (inp); 678 } 679 } 680 /* 681 * Not found. 682 */ 683 return (NULL); 684 } else { 685 struct inpcbporthead *porthash; 686 struct inpcbport *phd; 687 struct inpcb *match = NULL; 688 /* 689 * Best fit PCB lookup. 690 * 691 * First see if this local port is in use by looking on the 692 * port hash list. 693 */ 694 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 695 pcbinfo->ipi_porthashmask)]; 696 LIST_FOREACH(phd, porthash, phd_hash) { 697 if (phd->phd_port == lport) 698 break; 699 } 700 if (phd != NULL) { 701 /* 702 * Port is in use by one or more PCBs. Look for best 703 * fit. 704 */ 705 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 706 wildcard = 0; 707 if (cred != NULL && 708 !prison_equal_ip6(cred->cr_prison, 709 inp->inp_cred->cr_prison)) 710 continue; 711 /* XXX inp locking */ 712 if ((inp->inp_vflag & INP_IPV6) == 0) 713 continue; 714 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 715 wildcard++; 716 if (!IN6_IS_ADDR_UNSPECIFIED( 717 &inp->in6p_laddr)) { 718 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 719 wildcard++; 720 else if (!IN6_ARE_ADDR_EQUAL( 721 &inp->in6p_laddr, laddr)) 722 continue; 723 } else { 724 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 725 wildcard++; 726 } 727 if (wildcard < matchwild) { 728 match = inp; 729 matchwild = wildcard; 730 if (matchwild == 0) 731 break; 732 } 733 } 734 } 735 return (match); 736 } 737 } 738 739 void 740 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 741 { 742 struct inpcb *in6p; 743 struct ip6_moptions *im6o; 744 int i, gap; 745 746 INP_INFO_RLOCK(pcbinfo); 747 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) { 748 INP_WLOCK(in6p); 749 im6o = in6p->in6p_moptions; 750 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) { 751 /* 752 * Unselect the outgoing ifp for multicast if it 753 * is being detached. 754 */ 755 if (im6o->im6o_multicast_ifp == ifp) 756 im6o->im6o_multicast_ifp = NULL; 757 /* 758 * Drop multicast group membership if we joined 759 * through the interface being detached. 760 */ 761 gap = 0; 762 for (i = 0; i < im6o->im6o_num_memberships; i++) { 763 if (im6o->im6o_membership[i]->in6m_ifp == 764 ifp) { 765 in6_mc_leave(im6o->im6o_membership[i], 766 NULL); 767 gap++; 768 } else if (gap != 0) { 769 im6o->im6o_membership[i - gap] = 770 im6o->im6o_membership[i]; 771 } 772 } 773 im6o->im6o_num_memberships -= gap; 774 } 775 INP_WUNLOCK(in6p); 776 } 777 INP_INFO_RUNLOCK(pcbinfo); 778 } 779 780 /* 781 * Check for alternatives when higher level complains 782 * about service problems. For now, invalidate cached 783 * routing information. If the route was created dynamically 784 * (by a redirect), time to try a default gateway again. 785 */ 786 void 787 in6_losing(struct inpcb *in6p) 788 { 789 790 /* 791 * We don't store route pointers in the routing table anymore 792 */ 793 return; 794 } 795 796 /* 797 * After a routing change, flush old routing 798 * and allocate a (hopefully) better one. 799 */ 800 struct inpcb * 801 in6_rtchange(struct inpcb *inp, int errno) 802 { 803 /* 804 * We don't store route pointers in the routing table anymore 805 */ 806 return inp; 807 } 808 809 /* 810 * Lookup PCB in hash list. 811 */ 812 struct inpcb * 813 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 814 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard, 815 struct ifnet *ifp) 816 { 817 struct inpcbhead *head; 818 struct inpcb *inp, *tmpinp; 819 u_short fport = fport_arg, lport = lport_arg; 820 int faith; 821 822 INP_INFO_LOCK_ASSERT(pcbinfo); 823 824 if (faithprefix_p != NULL) 825 faith = (*faithprefix_p)(laddr); 826 else 827 faith = 0; 828 829 /* 830 * First look for an exact match. 831 */ 832 tmpinp = NULL; 833 head = &pcbinfo->ipi_hashbase[ 834 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport, 835 pcbinfo->ipi_hashmask)]; 836 LIST_FOREACH(inp, head, inp_hash) { 837 /* XXX inp locking */ 838 if ((inp->inp_vflag & INP_IPV6) == 0) 839 continue; 840 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 841 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 842 inp->inp_fport == fport && 843 inp->inp_lport == lport) { 844 /* 845 * XXX We should be able to directly return 846 * the inp here, without any checks. 847 * Well unless both bound with SO_REUSEPORT? 848 */ 849 if (prison_flag(inp->inp_cred, PR_IP6)) 850 return (inp); 851 if (tmpinp == NULL) 852 tmpinp = inp; 853 } 854 } 855 if (tmpinp != NULL) 856 return (tmpinp); 857 858 /* 859 * Then look for a wildcard match, if requested. 860 */ 861 if (wildcard == INPLOOKUP_WILDCARD) { 862 struct inpcb *local_wild = NULL, *local_exact = NULL; 863 struct inpcb *jail_wild = NULL; 864 int injail; 865 866 /* 867 * Order of socket selection - we always prefer jails. 868 * 1. jailed, non-wild. 869 * 2. jailed, wild. 870 * 3. non-jailed, non-wild. 871 * 4. non-jailed, wild. 872 */ 873 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 874 0, pcbinfo->ipi_hashmask)]; 875 LIST_FOREACH(inp, head, inp_hash) { 876 /* XXX inp locking */ 877 if ((inp->inp_vflag & INP_IPV6) == 0) 878 continue; 879 880 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 881 inp->inp_lport != lport) { 882 continue; 883 } 884 885 /* XXX inp locking */ 886 if (faith && (inp->inp_flags & INP_FAITH) == 0) 887 continue; 888 889 injail = prison_flag(inp->inp_cred, PR_IP6); 890 if (injail) { 891 if (prison_check_ip6(inp->inp_cred, 892 laddr) != 0) 893 continue; 894 } else { 895 if (local_exact != NULL) 896 continue; 897 } 898 899 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 900 if (injail) 901 return (inp); 902 else 903 local_exact = inp; 904 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 905 if (injail) 906 jail_wild = inp; 907 else 908 local_wild = inp; 909 } 910 } /* LIST_FOREACH */ 911 912 if (jail_wild != NULL) 913 return (jail_wild); 914 if (local_exact != NULL) 915 return (local_exact); 916 if (local_wild != NULL) 917 return (local_wild); 918 } /* if (wildcard == INPLOOKUP_WILDCARD) */ 919 920 /* 921 * Not found. 922 */ 923 return (NULL); 924 } 925 926 void 927 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m) 928 { 929 struct ip6_hdr *ip; 930 931 ip = mtod(m, struct ip6_hdr *); 932 bzero(sin6, sizeof(*sin6)); 933 sin6->sin6_len = sizeof(*sin6); 934 sin6->sin6_family = AF_INET6; 935 sin6->sin6_addr = ip->ip6_src; 936 937 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 938 939 return; 940 } 941