bsd/netinet6/in6_pcb.c

   1 /*
   2  * Copyright (c) 2003-2017 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /*
  29  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
  30  * All rights reserved.
  31  *
  32  * Redistribution and use in source and binary forms, with or without
  33  * modification, are permitted provided that the following conditions
  34  * are met:
  35  * 1. Redistributions of source code must retain the above copyright
  36  *    notice, this list of conditions and the following disclaimer.
  37  * 2. Redistributions in binary form must reproduce the above copyright
  38  *    notice, this list of conditions and the following disclaimer in the
  39  *    documentation and/or other materials provided with the distribution.
  40  * 3. Neither the name of the project nor the names of its contributors
  41  *    may be used to endorse or promote products derived from this software
  42  *    without specific prior written permission.
  43  *
  44  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  45  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  46  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  47  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  48  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  49  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  50  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  51  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  52  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  53  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  54  * SUCH DAMAGE.
  55  *
  56  */
  57
  58 /*
  59  * Copyright (c) 1982, 1986, 1991, 1993
  60  *      The Regents of the University of California.  All rights reserved.
  61  *
  62  * Redistribution and use in source and binary forms, with or without
  63  * modification, are permitted provided that the following conditions
  64  * are met:
  65  * 1. Redistributions of source code must retain the above copyright
  66  *    notice, this list of conditions and the following disclaimer.
  67  * 2. Redistributions in binary form must reproduce the above copyright
  68  *    notice, this list of conditions and the following disclaimer in the
  69  *    documentation and/or other materials provided with the distribution.
  70  * 3. All advertising materials mentioning features or use of this software
  71  *    must display the following acknowledgement:
  72  *      This product includes software developed by the University of
  73  *      California, Berkeley and its contributors.
  74  * 4. Neither the name of the University nor the names of its contributors
  75  *    may be used to endorse or promote products derived from this software
  76  *    without specific prior written permission.
  77  *
  78  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  79  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  80  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  81  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  82  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  83  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  84  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  85  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  86  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  87  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  88  * SUCH DAMAGE.
  89  *
  90  *      @(#)in_pcb.c    8.2 (Berkeley) 1/4/94
  91  */
  92
  93 #include <sys/param.h>
  94 #include <sys/systm.h>
  95 #include <sys/malloc.h>
  96 #include <sys/mbuf.h>
  97 #include <sys/domain.h>
  98 #include <sys/protosw.h>
  99 #include <sys/socket.h>
 100 #include <sys/socketvar.h>
 101 #include <sys/sockio.h>
 102 #include <sys/errno.h>
 103 #include <sys/time.h>
 104 #include <sys/proc.h>
 105 #include <sys/kauth.h>
 106 #include <sys/priv.h>
 107
 108 #include <net/if.h>
 109 #include <net/if_types.h>
 110 #include <net/route.h>
 111 #include <net/ntstat.h>
 112
 113 #include <netinet/in.h>
 114 #include <netinet/in_var.h>
 115 #include <netinet/in_systm.h>
 116 #include <netinet/ip6.h>
 117 #include <netinet/ip_var.h>
 118 #include <netinet6/ip6_var.h>
 119 #include <netinet6/nd6.h>
 120 #include <netinet/in_pcb.h>
 121 #include <netinet6/in6_pcb.h>
 122 #include <net/if_types.h>
 123 #include <net/if_var.h>
 124
 125 #include <kern/kern_types.h>
 126 #include <kern/zalloc.h>
 127
 128 #if IPSEC
 129 #include <netinet6/ipsec.h>
 130 #if INET6
 131 #include <netinet6/ipsec6.h>
 132 #endif
 133 #include <netinet6/ah.h>
 134 #if INET6
 135 #include <netinet6/ah6.h>
 136 #endif
 137 #include <netkey/key.h>
 138 #endif /* IPSEC */
 139
 140 #if NECP
 141 #include <net/necp.h>
 142 #endif /* NECP */
 143
 144 /*
 145  * in6_pcblookup_local_and_cleanup does everything
 146  * in6_pcblookup_local does but it checks for a socket
 147  * that's going away. Since we know that the lock is
 148  * held read+write when this function is called, we
 149  * can safely dispose of this socket like the slow
 150  * timer would usually do and return NULL. This is
 151  * great for bind.
 152  */
 153 static struct inpcb *
 154 in6_pcblookup_local_and_cleanup(struct inpcbinfo *pcbinfo,
 155     struct in6_addr *laddr, u_int lport_arg, int wild_okay)
 156 {
 157         struct inpcb *inp;
 158
 159         /* Perform normal lookup */
 160         inp = in6_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay);
 161
 162         /* Check if we found a match but it's waiting to be disposed */
 163         if (inp != NULL && inp->inp_wantcnt == WNT_STOPUSING) {
 164                 struct socket *so = inp->inp_socket;
 165
 166                 socket_lock(so, 0);
 167
 168                 if (so->so_usecount == 0) {
 169                         if (inp->inp_state != INPCB_STATE_DEAD) {
 170                                 in6_pcbdetach(inp);
 171                         }
 172                         in_pcbdispose(inp);     /* will unlock & destroy */
 173                         inp = NULL;
 174                 } else {
 175                         socket_unlock(so, 0);
 176                 }
 177         }
 178
 179         return inp;
 180 }
 181
 182 /*
 183  * Bind an INPCB to an address and/or port.  This routine should not alter
 184  * the caller-supplied local address "nam".
 185  */
 186 int
 187 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct proc *p)
 188 {
 189         struct socket *so = inp->inp_socket;
 190         struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
 191         u_short lport = 0;
 192         int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
 193         struct ifnet *outif = NULL;
 194         struct sockaddr_in6 sin6;
 195 #if !CONFIG_EMBEDDED
 196         int error;
 197         kauth_cred_t cred;
 198 #endif /* !CONFIG_EMBEDDED */
 199
 200         if (!in6_ifaddrs) { /* XXX broken! */
 201                 return EADDRNOTAVAIL;
 202         }
 203         if (!(so->so_options & (SO_REUSEADDR | SO_REUSEPORT))) {
 204                 wild = 1;
 205         }
 206
 207         socket_unlock(so, 0); /* keep reference */
 208         lck_rw_lock_exclusive(pcbinfo->ipi_lock);
 209         if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
 210                 /* another thread completed the bind */
 211                 lck_rw_done(pcbinfo->ipi_lock);
 212                 socket_lock(so, 0);
 213                 return EINVAL;
 214         }
 215
 216         bzero(&sin6, sizeof(sin6));
 217         if (nam != NULL) {
 218                 if (nam->sa_len != sizeof(struct sockaddr_in6)) {
 219                         lck_rw_done(pcbinfo->ipi_lock);
 220                         socket_lock(so, 0);
 221                         return EINVAL;
 222                 }
 223                 /*
 224                  * family check.
 225                  */
 226                 if (nam->sa_family != AF_INET6) {
 227                         lck_rw_done(pcbinfo->ipi_lock);
 228                         socket_lock(so, 0);
 229                         return EAFNOSUPPORT;
 230                 }
 231                 lport = SIN6(nam)->sin6_port;
 232
 233                 *(&sin6) = *SIN6(nam);
 234
 235                 /* KAME hack: embed scopeid */
 236                 if (in6_embedscope(&sin6.sin6_addr, &sin6, inp, NULL,
 237                     NULL) != 0) {
 238                         lck_rw_done(pcbinfo->ipi_lock);
 239                         socket_lock(so, 0);
 240                         return EINVAL;
 241                 }
 242
 243                 /* Sanitize local copy for address searches */
 244                 sin6.sin6_flowinfo = 0;
 245                 sin6.sin6_scope_id = 0;
 246                 sin6.sin6_port = 0;
 247
 248                 if (IN6_IS_ADDR_MULTICAST(&sin6.sin6_addr)) {
 249                         /*
 250                          * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
 251                          * allow compepte duplication of binding if
 252                          * SO_REUSEPORT is set, or if SO_REUSEADDR is set
 253                          * and a multicast address is bound on both
 254                          * new and duplicated sockets.
 255                          */
 256                         if (so->so_options & SO_REUSEADDR) {
 257                                 reuseport = SO_REUSEADDR | SO_REUSEPORT;
 258                         }
 259                 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr)) {
 260                         struct ifaddr *ifa;
 261
 262                         ifa = ifa_ifwithaddr(SA(&sin6));
 263                         if (ifa == NULL) {
 264                                 lck_rw_done(pcbinfo->ipi_lock);
 265                                 socket_lock(so, 0);
 266                                 return EADDRNOTAVAIL;
 267                         } else {
 268                                 /*
 269                                  * XXX: bind to an anycast address might
 270                                  * accidentally cause sending a packet with
 271                                  * anycast source address.  We should allow
 272                                  * to bind to a deprecated address, since
 273                                  * the application dare to use it.
 274                                  */
 275                                 IFA_LOCK_SPIN(ifa);
 276                                 if (((struct in6_ifaddr *)ifa)->ia6_flags &
 277                                     (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY |
 278                                     IN6_IFF_DETACHED | IN6_IFF_CLAT46)) {
 279                                         IFA_UNLOCK(ifa);
 280                                         IFA_REMREF(ifa);
 281                                         lck_rw_done(pcbinfo->ipi_lock);
 282                                         socket_lock(so, 0);
 283                                         return EADDRNOTAVAIL;
 284                                 }
 285                                 /*
 286                                  * Opportunistically determine the outbound
 287                                  * interface that may be used; this may not
 288                                  * hold true if we end up using a route
 289                                  * going over a different interface, e.g.
 290                                  * when sending to a local address.  This
 291                                  * will get updated again after sending.
 292                                  */
 293                                 outif = ifa->ifa_ifp;
 294                                 IFA_UNLOCK(ifa);
 295                                 IFA_REMREF(ifa);
 296                         }
 297                 }
 298                 if (lport != 0) {
 299                         struct inpcb *t;
 300                         uid_t u;
 301
 302 #if !CONFIG_EMBEDDED
 303                         if (ntohs(lport) < IPV6PORT_RESERVED &&
 304                             !IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr)) {
 305                                 cred = kauth_cred_proc_ref(p);
 306                                 error = priv_check_cred(cred,
 307                                     PRIV_NETINET_RESERVEDPORT, 0);
 308                                 kauth_cred_unref(&cred);
 309                                 if (error != 0) {
 310                                         lck_rw_done(pcbinfo->ipi_lock);
 311                                         socket_lock(so, 0);
 312                                         return EACCES;
 313                                 }
 314                         }
 315 #endif /* !CONFIG_EMBEDDED */
 316                         if (!IN6_IS_ADDR_MULTICAST(&sin6.sin6_addr) &&
 317                             (u = kauth_cred_getuid(so->so_cred)) != 0) {
 318                                 t = in6_pcblookup_local_and_cleanup(pcbinfo,
 319                                     &sin6.sin6_addr, lport,
 320                                     INPLOOKUP_WILDCARD);
 321                                 if (t != NULL && (!IN6_IS_ADDR_UNSPECIFIED(
 322                                             &sin6.sin6_addr) ||
 323                                     !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
 324                                     !(t->inp_socket->so_options &
 325                                     SO_REUSEPORT)) && (u != kauth_cred_getuid(
 326                                             t->inp_socket->so_cred)) &&
 327                                     !(t->inp_socket->so_flags &
 328                                     SOF_REUSESHAREUID)) {
 329                                         lck_rw_done(pcbinfo->ipi_lock);
 330                                         socket_lock(so, 0);
 331                                         return EADDRINUSE;
 332                                 }
 333                                 if (!(inp->inp_flags & IN6P_IPV6_V6ONLY) &&
 334                                     IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr)) {
 335                                         struct sockaddr_in sin;
 336
 337                                         in6_sin6_2_sin(&sin, &sin6);
 338                                         t = in_pcblookup_local_and_cleanup(
 339                                                 pcbinfo, sin.sin_addr, lport,
 340                                                 INPLOOKUP_WILDCARD);
 341                                         if (t != NULL &&
 342                                             !(t->inp_socket->so_options &
 343                                             SO_REUSEPORT) &&
 344                                             (kauth_cred_getuid(so->so_cred) !=
 345                                             kauth_cred_getuid(t->inp_socket->
 346                                             so_cred)) && (t->inp_laddr.s_addr !=
 347                                             INADDR_ANY || SOCK_DOM(so) ==
 348                                             SOCK_DOM(t->inp_socket))) {
 349                                                 lck_rw_done(pcbinfo->ipi_lock);
 350                                                 socket_lock(so, 0);
 351                                                 return EADDRINUSE;
 352                                         }
 353                                 }
 354                         }
 355                         t = in6_pcblookup_local_and_cleanup(pcbinfo,
 356                             &sin6.sin6_addr, lport, wild);
 357                         if (t != NULL &&
 358                             (reuseport & t->inp_socket->so_options) == 0) {
 359                                 lck_rw_done(pcbinfo->ipi_lock);
 360                                 socket_lock(so, 0);
 361                                 return EADDRINUSE;
 362                         }
 363                         if (!(inp->inp_flags & IN6P_IPV6_V6ONLY) &&
 364                             IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr)) {
 365                                 struct sockaddr_in sin;
 366
 367                                 in6_sin6_2_sin(&sin, &sin6);
 368                                 t = in_pcblookup_local_and_cleanup(pcbinfo,
 369                                     sin.sin_addr, lport, wild);
 370                                 if (t != NULL && (reuseport &
 371                                     t->inp_socket->so_options) == 0 &&
 372                                     (t->inp_laddr.s_addr != INADDR_ANY ||
 373                                     SOCK_DOM(so) == SOCK_DOM(t->inp_socket))) {
 374                                         lck_rw_done(pcbinfo->ipi_lock);
 375                                         socket_lock(so, 0);
 376                                         return EADDRINUSE;
 377                                 }
 378                         }
 379                 }
 380         }
 381
 382         socket_lock(so, 0);
 383         /*
 384          * We unlocked socket's protocol lock for a long time.
 385          * The socket might have been dropped/defuncted.
 386          * Checking if world has changed since.
 387          */
 388         if (inp->inp_state == INPCB_STATE_DEAD) {
 389                 lck_rw_done(pcbinfo->ipi_lock);
 390                 return ECONNABORTED;
 391         }
 392
 393         /* check if the socket got bound when the lock was released */
 394         if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
 395                 lck_rw_done(pcbinfo->ipi_lock);
 396                 return EINVAL;
 397         }
 398
 399         if (!IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr)) {
 400                 inp->in6p_laddr = sin6.sin6_addr;
 401                 inp->in6p_last_outifp = outif;
 402         }
 403
 404         if (lport == 0) {
 405                 int e;
 406                 if ((e = in6_pcbsetport(&inp->in6p_laddr, inp, p, 1)) != 0) {
 407                         /* Undo any address bind from above. */
 408                         inp->in6p_laddr = in6addr_any;
 409                         inp->in6p_last_outifp = NULL;
 410                         lck_rw_done(pcbinfo->ipi_lock);
 411                         return e;
 412                 }
 413         } else {
 414                 inp->inp_lport = lport;
 415                 if (in_pcbinshash(inp, 1) != 0) {
 416                         inp->in6p_laddr = in6addr_any;
 417                         inp->inp_lport = 0;
 418                         inp->in6p_last_outifp = NULL;
 419                         lck_rw_done(pcbinfo->ipi_lock);
 420                         return EAGAIN;
 421                 }
 422         }
 423         lck_rw_done(pcbinfo->ipi_lock);
 424         sflt_notify(so, sock_evt_bound, NULL);
 425         return 0;
 426 }
 427
 428 /*
 429  * Transform old in6_pcbconnect() into an inner subroutine for new
 430  * in6_pcbconnect(); do some validity-checking on the remote address
 431  * (in "nam") and then determine local host address (i.e., which
 432  * interface) to use to access that remote host.
 433  *
 434  * This routine may alter the caller-supplied remote address "nam".
 435  *
 436  * This routine might return an ifp with a reference held if the caller
 437  * provides a non-NULL outif, even in the error case.  The caller is
 438  * responsible for releasing its reference.
 439  */
 440 int
 441 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
 442     struct in6_addr *plocal_addr6, struct ifnet **outif)
 443 {
 444         struct in6_addr *addr6 = NULL;
 445         struct in6_addr src_storage;
 446         int error = 0;
 447         unsigned int ifscope;
 448
 449         if (outif != NULL) {
 450                 *outif = NULL;
 451         }
 452         if (nam->sa_len != sizeof(struct sockaddr_in6)) {
 453                 return EINVAL;
 454         }
 455         if (SIN6(nam)->sin6_family != AF_INET6) {
 456                 return EAFNOSUPPORT;
 457         }
 458         if (SIN6(nam)->sin6_port == 0) {
 459                 return EADDRNOTAVAIL;
 460         }
 461
 462         /* KAME hack: embed scopeid */
 463         if (in6_embedscope(&SIN6(nam)->sin6_addr, SIN6(nam), inp, NULL, NULL) != 0) {
 464                 return EINVAL;
 465         }
 466
 467         if (in6_ifaddrs) {
 468                 /*
 469                  * If the destination address is UNSPECIFIED addr,
 470                  * use the loopback addr, e.g ::1.
 471                  */
 472                 if (IN6_IS_ADDR_UNSPECIFIED(&SIN6(nam)->sin6_addr)) {
 473                         SIN6(nam)->sin6_addr = in6addr_loopback;
 474                 }
 475         }
 476
 477         ifscope = (inp->inp_flags & INP_BOUND_IF) ?
 478             inp->inp_boundifp->if_index : IFSCOPE_NONE;
 479
 480         /*
 481          * XXX: in6_selectsrc might replace the bound local address
 482          * with the address specified by setsockopt(IPV6_PKTINFO).
 483          * Is it the intended behavior?
 484          *
 485          * in6_selectsrc() might return outif with its reference held
 486          * even in the error case; caller always needs to release it
 487          * if non-NULL.
 488          */
 489         addr6 = in6_selectsrc(SIN6(nam), inp->in6p_outputopts, inp,
 490             &inp->in6p_route, outif, &src_storage, ifscope, &error);
 491
 492         if (outif != NULL) {
 493                 struct rtentry *rt = inp->in6p_route.ro_rt;
 494                 /*
 495                  * If in6_selectsrc() returns a route, it should be one
 496                  * which points to the same ifp as outif.  Just in case
 497                  * it isn't, use the one from the route for consistency.
 498                  * Otherwise if there is no route, leave outif alone as
 499                  * it could still be useful to the caller.
 500                  */
 501                 if (rt != NULL && rt->rt_ifp != *outif) {
 502                         ifnet_reference(rt->rt_ifp);    /* for caller */
 503                         if (*outif != NULL) {
 504                                 ifnet_release(*outif);
 505                         }
 506                         *outif = rt->rt_ifp;
 507                 }
 508         }
 509
 510         if (addr6 == NULL) {
 511                 if (outif != NULL && (*outif) != NULL &&
 512                     inp_restricted_send(inp, *outif)) {
 513                         soevent(inp->inp_socket,
 514                             (SO_FILT_HINT_LOCKED | SO_FILT_HINT_IFDENIED));
 515                         error = EHOSTUNREACH;
 516                 }
 517                 if (error == 0) {
 518                         error = EADDRNOTAVAIL;
 519                 }
 520                 return error;
 521         }
 522
 523         *plocal_addr6 = *addr6;
 524         /*
 525          * Don't do pcblookup call here; return interface in
 526          * plocal_addr6 and exit to caller, that will do the lookup.
 527          */
 528         return 0;
 529 }
 530
 531 /*
 532  * Outer subroutine:
 533  * Connect from a socket to a specified address.
 534  * Both address and port must be specified in argument sin.
 535  * If don't have a local address for this socket yet,
 536  * then pick one.
 537  */
 538 int
 539 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct proc *p)
 540 {
 541         struct in6_addr addr6;
 542         struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)(void *)nam;
 543         struct inpcb *pcb;
 544         int error = 0;
 545         struct ifnet *outif = NULL;
 546         struct socket *so = inp->inp_socket;
 547
 548 #if CONTENT_FILTER
 549         if (so) {
 550                 so->so_state_change_cnt++;
 551         }
 552 #endif
 553
 554         if (so->so_proto->pr_protocol == IPPROTO_UDP &&
 555             sin6->sin6_port == htons(53) && !(so->so_flags1 & SOF1_DNS_COUNTED)) {
 556                 so->so_flags1 |= SOF1_DNS_COUNTED;
 557                 INC_ATOMIC_INT64_LIM(net_api_stats.nas_socket_inet_dgram_dns);
 558         }
 559
 560         /*
 561          * Call inner routine, to assign local interface address.
 562          * in6_pcbladdr() may automatically fill in sin6_scope_id.
 563          *
 564          * in6_pcbladdr() might return an ifp with its reference held
 565          * even in the error case, so make sure that it's released
 566          * whenever it's non-NULL.
 567          */
 568         if ((error = in6_pcbladdr(inp, nam, &addr6, &outif)) != 0) {
 569                 if (outif != NULL && inp_restricted_send(inp, outif)) {
 570                         soevent(so,
 571                             (SO_FILT_HINT_LOCKED | SO_FILT_HINT_IFDENIED));
 572                 }
 573                 goto done;
 574         }
 575         socket_unlock(so, 0);
 576         pcb = in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
 577             sin6->sin6_port, IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) ?
 578             &addr6 : &inp->in6p_laddr, inp->inp_lport, 0, NULL);
 579         socket_lock(so, 0);
 580         if (pcb != NULL) {
 581                 in_pcb_checkstate(pcb, WNT_RELEASE, pcb == inp ? 1 : 0);
 582                 error = EADDRINUSE;
 583                 goto done;
 584         }
 585         if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
 586                 if (inp->inp_lport == 0) {
 587                         error = in6_pcbbind(inp, NULL, p);
 588                         if (error) {
 589                                 goto done;
 590                         }
 591                 }
 592                 inp->in6p_laddr = addr6;
 593                 inp->in6p_last_outifp = outif;  /* no reference needed */
 594                 inp->in6p_flags |= INP_IN6ADDR_ANY;
 595         }
 596         if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->ipi_lock)) {
 597                 /* lock inversion issue, mostly with udp multicast packets */
 598                 socket_unlock(so, 0);
 599                 lck_rw_lock_exclusive(inp->inp_pcbinfo->ipi_lock);
 600                 socket_lock(so, 0);
 601         }
 602         inp->in6p_faddr = sin6->sin6_addr;
 603         inp->inp_fport = sin6->sin6_port;
 604         if (nstat_collect && SOCK_PROTO(so) == IPPROTO_UDP) {
 605                 nstat_pcb_invalidate_cache(inp);
 606         }
 607         in_pcbrehash(inp);
 608         lck_rw_done(inp->inp_pcbinfo->ipi_lock);
 609
 610 done:
 611         if (outif != NULL) {
 612                 ifnet_release(outif);
 613         }
 614
 615         return error;
 616 }
 617
 618 void
 619 in6_pcbdisconnect(struct inpcb *inp)
 620 {
 621         struct socket *so = inp->inp_socket;
 622
 623 #if CONTENT_FILTER
 624         if (so) {
 625                 so->so_state_change_cnt++;
 626         }
 627 #endif
 628
 629         if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->ipi_lock)) {
 630                 /* lock inversion issue, mostly with udp multicast packets */
 631                 socket_unlock(so, 0);
 632                 lck_rw_lock_exclusive(inp->inp_pcbinfo->ipi_lock);
 633                 socket_lock(so, 0);
 634         }
 635         if (nstat_collect && SOCK_PROTO(so) == IPPROTO_UDP) {
 636                 nstat_pcb_cache(inp);
 637         }
 638         bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
 639         inp->inp_fport = 0;
 640         /* clear flowinfo - RFC 6437 */
 641         inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
 642         in_pcbrehash(inp);
 643         lck_rw_done(inp->inp_pcbinfo->ipi_lock);
 644         /*
 645          * A multipath subflow socket would have its SS_NOFDREF set by default,
 646          * so check for SOF_MP_SUBFLOW socket flag before detaching the PCB;
 647          * when the socket is closed for real, SOF_MP_SUBFLOW would be cleared.
 648          */
 649         if (!(so->so_flags & SOF_MP_SUBFLOW) && (so->so_state & SS_NOFDREF)) {
 650                 in6_pcbdetach(inp);
 651         }
 652 }
 653
 654 void
 655 in6_pcbdetach(struct inpcb *inp)
 656 {
 657         struct socket *so = inp->inp_socket;
 658
 659         if (so->so_pcb == NULL) {
 660                 /* PCB has been disposed */
 661                 panic("%s: inp=%p so=%p proto=%d so_pcb is null!\n", __func__,
 662                     inp, so, SOCK_PROTO(so));
 663                 /* NOTREACHED */
 664         }
 665
 666 #if IPSEC
 667         if (inp->in6p_sp != NULL) {
 668                 (void) ipsec6_delete_pcbpolicy(inp);
 669         }
 670 #endif /* IPSEC */
 671
 672         if (inp->inp_stat != NULL && SOCK_PROTO(so) == IPPROTO_UDP) {
 673                 if (inp->inp_stat->rxpackets == 0 && inp->inp_stat->txpackets == 0) {
 674                         INC_ATOMIC_INT64_LIM(net_api_stats.nas_socket_inet6_dgram_no_data);
 675                 }
 676         }
 677
 678         /*
 679          * Let NetworkStatistics know this PCB is going away
 680          * before we detach it.
 681          */
 682         if (nstat_collect &&
 683             (SOCK_PROTO(so) == IPPROTO_TCP || SOCK_PROTO(so) == IPPROTO_UDP)) {
 684                 nstat_pcb_detach(inp);
 685         }
 686         /* mark socket state as dead */
 687         if (in_pcb_checkstate(inp, WNT_STOPUSING, 1) != WNT_STOPUSING) {
 688                 panic("%s: so=%p proto=%d couldn't set to STOPUSING\n",
 689                     __func__, so, SOCK_PROTO(so));
 690                 /* NOTREACHED */
 691         }
 692
 693         if (!(so->so_flags & SOF_PCBCLEARING)) {
 694                 struct ip_moptions *imo;
 695                 struct ip6_moptions *im6o;
 696
 697                 inp->inp_vflag = 0;
 698                 if (inp->in6p_options != NULL) {
 699                         m_freem(inp->in6p_options);
 700                         inp->in6p_options = NULL;
 701                 }
 702                 ip6_freepcbopts(inp->in6p_outputopts);
 703                 ROUTE_RELEASE(&inp->in6p_route);
 704                 /* free IPv4 related resources in case of mapped addr */
 705                 if (inp->inp_options != NULL) {
 706                         (void) m_free(inp->inp_options);
 707                         inp->inp_options = NULL;
 708                 }
 709                 im6o = inp->in6p_moptions;
 710                 inp->in6p_moptions = NULL;
 711
 712                 imo = inp->inp_moptions;
 713                 inp->inp_moptions = NULL;
 714
 715                 sofreelastref(so, 0);
 716                 inp->inp_state = INPCB_STATE_DEAD;
 717                 /* makes sure we're not called twice from so_close */
 718                 so->so_flags |= SOF_PCBCLEARING;
 719
 720                 inpcb_gc_sched(inp->inp_pcbinfo, INPCB_TIMER_FAST);
 721
 722                 /*
 723                  * See inp_join_group() for why we need to unlock
 724                  */
 725                 if (im6o != NULL || imo != NULL) {
 726                         socket_unlock(so, 0);
 727                         if (im6o != NULL) {
 728                                 IM6O_REMREF(im6o);
 729                         }
 730                         if (imo != NULL) {
 731                                 IMO_REMREF(imo);
 732                         }
 733                         socket_lock(so, 0);
 734                 }
 735         }
 736 }
 737
 738 struct sockaddr *
 739 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
 740 {
 741         struct sockaddr_in6 *sin6;
 742
 743         MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, M_WAITOK);
 744         if (sin6 == NULL) {
 745                 return NULL;
 746         }
 747         bzero(sin6, sizeof(*sin6));
 748         sin6->sin6_family = AF_INET6;
 749         sin6->sin6_len = sizeof(*sin6);
 750         sin6->sin6_port = port;
 751         sin6->sin6_addr = *addr_p;
 752
 753         /* would be good to use sa6_recoverscope(), except for locking  */
 754         if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
 755                 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
 756         } else {
 757                 sin6->sin6_scope_id = 0;        /* XXX */
 758         }
 759         if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
 760                 sin6->sin6_addr.s6_addr16[1] = 0;
 761         }
 762
 763         return (struct sockaddr *)sin6;
 764 }
 765
 766 void
 767 in6_sockaddr_s(in_port_t port, struct in6_addr *addr_p,
 768     struct sockaddr_in6 *sin6)
 769 {
 770         bzero(sin6, sizeof(*sin6));
 771         sin6->sin6_family = AF_INET6;
 772         sin6->sin6_len = sizeof(*sin6);
 773         sin6->sin6_port = port;
 774         sin6->sin6_addr = *addr_p;
 775
 776         /* would be good to use sa6_recoverscope(), except for locking  */
 777         if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
 778                 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
 779         } else {
 780                 sin6->sin6_scope_id = 0;        /* XXX */
 781         }
 782         if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
 783                 sin6->sin6_addr.s6_addr16[1] = 0;
 784         }
 785 }
 786
 787 /*
 788  * The calling convention of in6_getsockaddr() and in6_getpeeraddr() was
 789  * modified to match the pru_sockaddr() and pru_peeraddr() entry points
 790  * in struct pr_usrreqs, so that protocols can just reference then directly
 791  * without the need for a wrapper function.
 792  */
 793 int
 794 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
 795 {
 796         struct inpcb *inp;
 797         struct in6_addr addr;
 798         in_port_t port;
 799
 800         if ((inp = sotoinpcb(so)) == NULL) {
 801                 return EINVAL;
 802         }
 803
 804         port = inp->inp_lport;
 805         addr = inp->in6p_laddr;
 806
 807         *nam = in6_sockaddr(port, &addr);
 808         if (*nam == NULL) {
 809                 return ENOBUFS;
 810         }
 811         return 0;
 812 }
 813
 814 int
 815 in6_getsockaddr_s(struct socket *so, struct sockaddr_in6 *ss)
 816 {
 817         struct inpcb *inp;
 818         struct in6_addr addr;
 819         in_port_t port;
 820
 821         VERIFY(ss != NULL);
 822         bzero(ss, sizeof(*ss));
 823
 824         if ((inp = sotoinpcb(so)) == NULL) {
 825                 return EINVAL;
 826         }
 827
 828         port = inp->inp_lport;
 829         addr = inp->in6p_laddr;
 830
 831         in6_sockaddr_s(port, &addr, ss);
 832         return 0;
 833 }
 834
 835 int
 836 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
 837 {
 838         struct inpcb *inp;
 839         struct in6_addr addr;
 840         in_port_t port;
 841
 842         if ((inp = sotoinpcb(so)) == NULL) {
 843                 return EINVAL;
 844         }
 845
 846         port = inp->inp_fport;
 847         addr = inp->in6p_faddr;
 848
 849         *nam = in6_sockaddr(port, &addr);
 850         if (*nam == NULL) {
 851                 return ENOBUFS;
 852         }
 853         return 0;
 854 }
 855
 856 int
 857 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
 858 {
 859         struct  inpcb *inp = sotoinpcb(so);
 860         int     error;
 861
 862         if (inp == NULL) {
 863                 return EINVAL;
 864         }
 865         if (inp->inp_vflag & INP_IPV4) {
 866                 error = in_getsockaddr(so, nam);
 867                 if (error == 0) {
 868                         error = in6_sin_2_v4mapsin6_in_sock(nam);
 869                 }
 870         } else {
 871                 /* scope issues will be handled in in6_getsockaddr(). */
 872                 error = in6_getsockaddr(so, nam);
 873         }
 874         return error;
 875 }
 876
 877 int
 878 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
 879 {
 880         struct  inpcb *inp = sotoinpcb(so);
 881         int     error;
 882
 883         if (inp == NULL) {
 884                 return EINVAL;
 885         }
 886         if (inp->inp_vflag & INP_IPV4) {
 887                 error = in_getpeeraddr(so, nam);
 888                 if (error == 0) {
 889                         error = in6_sin_2_v4mapsin6_in_sock(nam);
 890                 }
 891         } else {
 892                 /* scope issues will be handled in in6_getpeeraddr(). */
 893                 error = in6_getpeeraddr(so, nam);
 894         }
 895         return error;
 896 }
 897
 898 /*
 899  * Pass some notification to all connections of a protocol
 900  * associated with address dst.  The local address and/or port numbers
 901  * may be specified to limit the search.  The "usual action" will be
 902  * taken, depending on the ctlinput cmd.  The caller must filter any
 903  * cmds that are uninteresting (e.g., no error in the map).
 904  * Call the protocol specific routine (if any) to report
 905  * any errors for each matching socket.
 906  */
 907 void
 908 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, u_int fport_arg,
 909     const struct sockaddr *src, u_int lport_arg, int cmd, void *cmdarg,
 910     void (*notify)(struct inpcb *, int))
 911 {
 912         struct inpcbhead *head = pcbinfo->ipi_listhead;
 913         struct inpcb *inp, *ninp;
 914         struct sockaddr_in6 sa6_src, *sa6_dst;
 915         u_short fport = fport_arg, lport = lport_arg;
 916         u_int32_t flowinfo;
 917         int errno;
 918
 919         if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6) {
 920                 return;
 921         }
 922
 923         sa6_dst = (struct sockaddr_in6 *)(void *)dst;
 924         if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) {
 925                 return;
 926         }
 927
 928         /*
 929          * note that src can be NULL when we get notify by local fragmentation.
 930          */
 931         sa6_src = (src == NULL) ?
 932             sa6_any : *(struct sockaddr_in6 *)(uintptr_t)(size_t)src;
 933         flowinfo = sa6_src.sin6_flowinfo;
 934
 935         /*
 936          * Redirects go to all references to the destination,
 937          * and use in6_rtchange to invalidate the route cache.
 938          * Dead host indications: also use in6_rtchange to invalidate
 939          * the cache, and deliver the error to all the sockets.
 940          * Otherwise, if we have knowledge of the local port and address,
 941          * deliver only to that socket.
 942          */
 943         if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
 944                 fport = 0;
 945                 lport = 0;
 946                 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
 947
 948                 if (cmd != PRC_HOSTDEAD) {
 949                         notify = in6_rtchange;
 950                 }
 951         }
 952         errno = inet6ctlerrmap[cmd];
 953         lck_rw_lock_shared(pcbinfo->ipi_lock);
 954         for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
 955                 ninp = LIST_NEXT(inp, inp_list);
 956
 957                 if (!(inp->inp_vflag & INP_IPV6)) {
 958                         continue;
 959                 }
 960
 961                 /*
 962                  * If the error designates a new path MTU for a destination
 963                  * and the application (associated with this socket) wanted to
 964                  * know the value, notify. Note that we notify for all
 965                  * disconnected sockets if the corresponding application
 966                  * wanted. This is because some UDP applications keep sending
 967                  * sockets disconnected.
 968                  * XXX: should we avoid to notify the value to TCP sockets?
 969                  */
 970                 if (cmd == PRC_MSGSIZE) {
 971                         ip6_notify_pmtu(inp, (struct sockaddr_in6 *)(void *)dst,
 972                             (u_int32_t *)cmdarg);
 973                 }
 974
 975                 /*
 976                  * Detect if we should notify the error. If no source and
 977                  * destination ports are specifed, but non-zero flowinfo and
 978                  * local address match, notify the error. This is the case
 979                  * when the error is delivered with an encrypted buffer
 980                  * by ESP. Otherwise, just compare addresses and ports
 981                  * as usual.
 982                  */
 983                 if (lport == 0 && fport == 0 && flowinfo &&
 984                     inp->inp_socket != NULL &&
 985                     flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
 986                     IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) {
 987                         goto do_notify;
 988                 } else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
 989                     &sa6_dst->sin6_addr) || inp->inp_socket == NULL ||
 990                     (lport && inp->inp_lport != lport) ||
 991                     (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
 992                     !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
 993                     &sa6_src.sin6_addr)) || (fport && inp->inp_fport != fport)) {
 994                         continue;
 995                 }
 996
 997 do_notify:
 998                 if (notify) {
 999                         if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) ==
1000                             WNT_STOPUSING) {
1001                                 continue;
1002                         }
1003                         socket_lock(inp->inp_socket, 1);
1004                         (*notify)(inp, errno);
1005                         (void) in_pcb_checkstate(inp, WNT_RELEASE, 1);
1006                         socket_unlock(inp->inp_socket, 1);
1007                 }
1008         }
1009         lck_rw_done(pcbinfo->ipi_lock);
1010 }
1011
1012 /*
1013  * Lookup a PCB based on the local address and port.
1014  */
1015 struct inpcb *
1016 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
1017     u_int lport_arg, int wild_okay)
1018 {
1019         struct inpcb *inp;
1020         int matchwild = 3, wildcard;
1021         u_short lport = lport_arg;
1022         struct inpcbporthead *porthash;
1023         struct inpcb *match = NULL;
1024         struct inpcbport *phd;
1025
1026         if (!wild_okay) {
1027                 struct inpcbhead *head;
1028                 /*
1029                  * Look for an unconnected (wildcard foreign addr) PCB that
1030                  * matches the local address and port we're looking for.
1031                  */
1032                 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
1033                     pcbinfo->ipi_hashmask)];
1034                 LIST_FOREACH(inp, head, inp_hash) {
1035                         if (!(inp->inp_vflag & INP_IPV6)) {
1036                                 continue;
1037                         }
1038                         if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1039                             IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1040                             inp->inp_lport == lport) {
1041                                 /*
1042                                  * Found.
1043                                  */
1044                                 return inp;
1045                         }
1046                 }
1047                 /*
1048                  * Not found.
1049                  */
1050                 return NULL;
1051         }
1052         /*
1053          * Best fit PCB lookup.
1054          *
1055          * First see if this local port is in use by looking on the
1056          * port hash list.
1057          */
1058         porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
1059             pcbinfo->ipi_porthashmask)];
1060         LIST_FOREACH(phd, porthash, phd_hash) {
1061                 if (phd->phd_port == lport) {
1062                         break;
1063                 }
1064         }
1065         if (phd != NULL) {
1066                 /*
1067                  * Port is in use by one or more PCBs. Look for best
1068                  * fit.
1069                  */
1070                 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
1071                         wildcard = 0;
1072                         if (!(inp->inp_vflag & INP_IPV6)) {
1073                                 continue;
1074                         }
1075                         if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
1076                                 wildcard++;
1077                         }
1078                         if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1079                                 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) {
1080                                         wildcard++;
1081                                 } else if (!IN6_ARE_ADDR_EQUAL(
1082                                             &inp->in6p_laddr, laddr)) {
1083                                         continue;
1084                                 }
1085                         } else {
1086                                 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) {
1087                                         wildcard++;
1088                                 }
1089                         }
1090                         if (wildcard < matchwild) {
1091                                 match = inp;
1092                                 matchwild = wildcard;
1093                                 if (matchwild == 0) {
1094                                         break;
1095                                 }
1096                         }
1097                 }
1098         }
1099         return match;
1100 }
1101
1102 /*
1103  * Check for alternatives when higher level complains
1104  * about service problems.  For now, invalidate cached
1105  * routing information.  If the route was created dynamically
1106  * (by a redirect), time to try a default gateway again.
1107  */
1108 void
1109 in6_losing(struct inpcb *in6p)
1110 {
1111         struct rtentry *rt;
1112
1113         if ((rt = in6p->in6p_route.ro_rt) != NULL) {
1114                 RT_LOCK(rt);
1115                 if (rt->rt_flags & RTF_DYNAMIC) {
1116                         /*
1117                          * Prevent another thread from modifying rt_key,
1118                          * rt_gateway via rt_setgate() after the rt_lock
1119                          * is dropped by marking the route as defunct.
1120                          */
1121                         rt->rt_flags |= RTF_CONDEMNED;
1122                         RT_UNLOCK(rt);
1123                         (void) rtrequest(RTM_DELETE, rt_key(rt),
1124                             rt->rt_gateway, rt_mask(rt), rt->rt_flags, NULL);
1125                 } else {
1126                         RT_UNLOCK(rt);
1127                 }
1128                 /*
1129                  * A new route can be allocated
1130                  * the next time output is attempted.
1131                  */
1132         }
1133         ROUTE_RELEASE(&in6p->in6p_route);
1134 }
1135
1136 /*
1137  * After a routing change, flush old routing
1138  * and allocate a (hopefully) better one.
1139  */
1140 void
1141 in6_rtchange(struct inpcb *inp, int errno)
1142 {
1143 #pragma unused(errno)
1144         /*
1145          * A new route can be allocated the next time
1146          * output is attempted.
1147          */
1148         ROUTE_RELEASE(&inp->in6p_route);
1149 }
1150
1151 /*
1152  * Check if PCB exists hash list. Also returns uid and gid of socket
1153  */
1154 int
1155 in6_pcblookup_hash_exists(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1156     u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard,
1157     uid_t *uid, gid_t *gid, struct ifnet *ifp)
1158 {
1159         struct inpcbhead *head;
1160         struct inpcb *inp;
1161         u_short fport = fport_arg, lport = lport_arg;
1162         int found;
1163
1164         *uid = UID_MAX;
1165         *gid = GID_MAX;
1166
1167         lck_rw_lock_shared(pcbinfo->ipi_lock);
1168
1169         /*
1170          * First look for an exact match.
1171          */
1172         head = &pcbinfo->ipi_hashbase[INP_PCBHASH(faddr->s6_addr32[3] /* XXX */,
1173             lport, fport, pcbinfo->ipi_hashmask)];
1174         LIST_FOREACH(inp, head, inp_hash) {
1175                 if (!(inp->inp_vflag & INP_IPV6)) {
1176                         continue;
1177                 }
1178
1179                 if (inp_restricted_recv(inp, ifp)) {
1180                         continue;
1181                 }
1182
1183                 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1184                     IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1185                     inp->inp_fport == fport &&
1186                     inp->inp_lport == lport) {
1187                         if ((found = (inp->inp_socket != NULL))) {
1188                                 /*
1189                                  * Found. Check if pcb is still valid
1190                                  */
1191                                 *uid = kauth_cred_getuid(
1192                                         inp->inp_socket->so_cred);
1193                                 *gid = kauth_cred_getgid(
1194                                         inp->inp_socket->so_cred);
1195                         }
1196                         lck_rw_done(pcbinfo->ipi_lock);
1197                         return found;
1198                 }
1199         }
1200         if (wildcard) {
1201                 struct inpcb *local_wild = NULL;
1202
1203                 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
1204                     pcbinfo->ipi_hashmask)];
1205                 LIST_FOREACH(inp, head, inp_hash) {
1206                         if (!(inp->inp_vflag & INP_IPV6)) {
1207                                 continue;
1208                         }
1209
1210                         if (inp_restricted_recv(inp, ifp)) {
1211                                 continue;
1212                         }
1213
1214                         if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1215                             inp->inp_lport == lport) {
1216                                 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1217                                     laddr)) {
1218                                         found = (inp->inp_socket != NULL);
1219                                         if (found) {
1220                                                 *uid = kauth_cred_getuid(
1221                                                         inp->inp_socket->so_cred);
1222                                                 *gid = kauth_cred_getgid(
1223                                                         inp->inp_socket->so_cred);
1224                                         }
1225                                         lck_rw_done(pcbinfo->ipi_lock);
1226                                         return found;
1227                                 } else if (IN6_IS_ADDR_UNSPECIFIED(
1228                                             &inp->in6p_laddr)) {
1229                                         local_wild = inp;
1230                                 }
1231                         }
1232                 }
1233                 if (local_wild) {
1234                         if ((found = (local_wild->inp_socket != NULL))) {
1235                                 *uid = kauth_cred_getuid(
1236                                         local_wild->inp_socket->so_cred);
1237                                 *gid = kauth_cred_getgid(
1238                                         local_wild->inp_socket->so_cred);
1239                         }
1240                         lck_rw_done(pcbinfo->ipi_lock);
1241                         return found;
1242                 }
1243         }
1244
1245         /*
1246          * Not found.
1247          */
1248         lck_rw_done(pcbinfo->ipi_lock);
1249         return 0;
1250 }
1251
1252 /*
1253  * Lookup PCB in hash list.
1254  */
1255 struct inpcb *
1256 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1257     u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard,
1258     struct ifnet *ifp)
1259 {
1260         struct inpcbhead *head;
1261         struct inpcb *inp;
1262         u_short fport = fport_arg, lport = lport_arg;
1263
1264         lck_rw_lock_shared(pcbinfo->ipi_lock);
1265
1266         /*
1267          * First look for an exact match.
1268          */
1269         head = &pcbinfo->ipi_hashbase[INP_PCBHASH(faddr->s6_addr32[3] /* XXX */,
1270             lport, fport, pcbinfo->ipi_hashmask)];
1271         LIST_FOREACH(inp, head, inp_hash) {
1272                 if (!(inp->inp_vflag & INP_IPV6)) {
1273                         continue;
1274                 }
1275
1276                 if (inp_restricted_recv(inp, ifp)) {
1277                         continue;
1278                 }
1279
1280                 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1281                     IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1282                     inp->inp_fport == fport &&
1283                     inp->inp_lport == lport) {
1284                         /*
1285                          * Found. Check if pcb is still valid
1286                          */
1287                         if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) !=
1288                             WNT_STOPUSING) {
1289                                 lck_rw_done(pcbinfo->ipi_lock);
1290                                 return inp;
1291                         } else {
1292                                 /* it's there but dead, say it isn't found */
1293                                 lck_rw_done(pcbinfo->ipi_lock);
1294                                 return NULL;
1295                         }
1296                 }
1297         }
1298         if (wildcard) {
1299                 struct inpcb *local_wild = NULL;
1300
1301                 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
1302                     pcbinfo->ipi_hashmask)];
1303                 LIST_FOREACH(inp, head, inp_hash) {
1304                         if (!(inp->inp_vflag & INP_IPV6)) {
1305                                 continue;
1306                         }
1307
1308                         if (inp_restricted_recv(inp, ifp)) {
1309                                 continue;
1310                         }
1311
1312                         if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1313                             inp->inp_lport == lport) {
1314                                 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1315                                     laddr)) {
1316                                         if (in_pcb_checkstate(inp, WNT_ACQUIRE,
1317                                             0) != WNT_STOPUSING) {
1318                                                 lck_rw_done(pcbinfo->ipi_lock);
1319                                                 return inp;
1320                                         } else {
1321                                                 /* dead; say it isn't found */
1322                                                 lck_rw_done(pcbinfo->ipi_lock);
1323                                                 return NULL;
1324                                         }
1325                                 } else if (IN6_IS_ADDR_UNSPECIFIED(
1326                                             &inp->in6p_laddr)) {
1327                                         local_wild = inp;
1328                                 }
1329                         }
1330                 }
1331                 if (local_wild && in_pcb_checkstate(local_wild,
1332                     WNT_ACQUIRE, 0) != WNT_STOPUSING) {
1333                         lck_rw_done(pcbinfo->ipi_lock);
1334                         return local_wild;
1335                 } else {
1336                         lck_rw_done(pcbinfo->ipi_lock);
1337                         return NULL;
1338                 }
1339         }
1340
1341         /*
1342          * Not found.
1343          */
1344         lck_rw_done(pcbinfo->ipi_lock);
1345         return NULL;
1346 }
1347
1348 void
1349 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1350 {
1351         struct ip6_hdr *ip;
1352
1353         ip = mtod(m, struct ip6_hdr *);
1354         bzero(sin6, sizeof(*sin6));
1355         sin6->sin6_len = sizeof(*sin6);
1356         sin6->sin6_family = AF_INET6;
1357         sin6->sin6_addr = ip->ip6_src;
1358         if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
1359                 sin6->sin6_addr.s6_addr16[1] = 0;
1360                 if ((m->m_pkthdr.pkt_flags & (PKTF_LOOP | PKTF_IFAINFO)) ==
1361                     (PKTF_LOOP | PKTF_IFAINFO)) {
1362                         sin6->sin6_scope_id = m->m_pkthdr.src_ifindex;
1363                 } else if (m->m_pkthdr.rcvif != NULL) {
1364                         sin6->sin6_scope_id = m->m_pkthdr.rcvif->if_index;
1365                 }
1366         }
1367 }
1368
1369 /*
1370  * The following routines implement this scheme:
1371  *
1372  * Callers of ip6_output() that intend to cache the route in the inpcb pass
1373  * a local copy of the struct route to ip6_output().  Using a local copy of
1374  * the cached route significantly simplifies things as IP no longer has to
1375  * worry about having exclusive access to the passed in struct route, since
1376  * it's defined in the caller's stack; in essence, this allows for a lock-
1377  * less operation when updating the struct route at the IP level and below,
1378  * whenever necessary. The scheme works as follows:
1379  *
1380  * Prior to dropping the socket's lock and calling ip6_output(), the caller
1381  * copies the struct route from the inpcb into its stack, and adds a reference
1382  * to the cached route entry, if there was any.  The socket's lock is then
1383  * dropped and ip6_output() is called with a pointer to the copy of struct
1384  * route defined on the stack (not to the one in the inpcb.)
1385  *
1386  * Upon returning from ip6_output(), the caller then acquires the socket's
1387  * lock and synchronizes the cache; if there is no route cached in the inpcb,
1388  * it copies the local copy of struct route (which may or may not contain any
1389  * route) back into the cache; otherwise, if the inpcb has a route cached in
1390  * it, the one in the local copy will be freed, if there's any.  Trashing the
1391  * cached route in the inpcb can be avoided because ip6_output() is single-
1392  * threaded per-PCB (i.e. multiple transmits on a PCB are always serialized
1393  * by the socket/transport layer.)
1394  */
1395 void
1396 in6p_route_copyout(struct inpcb *inp, struct route_in6 *dst)
1397 {
1398         struct route_in6 *src = &inp->in6p_route;
1399
1400         socket_lock_assert_owned(inp->inp_socket);
1401
1402         /* Minor sanity check */
1403         if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET6) {
1404                 panic("%s: wrong or corrupted route: %p", __func__, src);
1405         }
1406
1407         route_copyout((struct route *)dst, (struct route *)src, sizeof(*dst));
1408 }
1409
1410 void
1411 in6p_route_copyin(struct inpcb *inp, struct route_in6 *src)
1412 {
1413         struct route_in6 *dst = &inp->in6p_route;
1414
1415         socket_lock_assert_owned(inp->inp_socket);
1416
1417         /* Minor sanity check */
1418         if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET6) {
1419                 panic("%s: wrong or corrupted route: %p", __func__, src);
1420         }
1421
1422         route_copyin((struct route *)src, (struct route *)dst, sizeof(*src));
1423 }