2 * Copyright (c) 2003-2013 Apple Inc. All rights reserved.
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
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.
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
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.
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
29 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
30 * All rights reserved.
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
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.
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
59 * Copyright (c) 1982, 1986, 1991, 1993
60 * The Regents of the University of California. All rights reserved.
62 * Redistribution and use in source and binary forms, with or without
63 * modification, are permitted provided that the following conditions
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.
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
90 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
93 #include <sys/param.h>
94 #include <sys/systm.h>
95 #include <sys/malloc.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>
109 #include <net/if_types.h>
110 #include <net/route.h>
111 #include <net/ntstat.h>
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>
125 #include <kern/kern_types.h>
126 #include <kern/zalloc.h>
129 #include <netinet6/ipsec.h>
131 #include <netinet6/ipsec6.h>
133 #include <netinet6/ah.h>
135 #include <netinet6/ah6.h>
137 #include <netkey/key.h>
141 #include <net/necp.h>
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
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
)
159 /* Perform normal lookup */
160 inp
= in6_pcblookup_local(pcbinfo
, laddr
, lport_arg
, wild_okay
);
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
;
166 lck_mtx_lock(&inp
->inpcb_mtx
);
168 if (so
->so_usecount
== 0) {
169 if (inp
->inp_state
!= INPCB_STATE_DEAD
)
171 in_pcbdispose(inp
); /* will unlock & destroy */
174 lck_mtx_unlock(&inp
->inpcb_mtx
);
182 * Bind an INPCB to an address and/or port. This routine should not alter
183 * the caller-supplied local address "nam".
186 in6_pcbbind(struct inpcb
*inp
, struct sockaddr
*nam
, struct proc
*p
)
188 struct socket
*so
= inp
->inp_socket
;
189 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
191 int wild
= 0, reuseport
= (so
->so_options
& SO_REUSEPORT
);
192 struct ifnet
*outif
= NULL
;
193 struct sockaddr_in6 sin6
;
197 if (!in6_ifaddrs
) /* XXX broken! */
198 return (EADDRNOTAVAIL
);
199 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
))
201 if (!(so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)))
204 socket_unlock(so
, 0); /* keep reference */
205 lck_rw_lock_exclusive(pcbinfo
->ipi_lock
);
207 bzero(&sin6
, sizeof (sin6
));
209 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
)) {
210 lck_rw_done(pcbinfo
->ipi_lock
);
217 if (nam
->sa_family
!= AF_INET6
) {
218 lck_rw_done(pcbinfo
->ipi_lock
);
220 return (EAFNOSUPPORT
);
222 lport
= SIN6(nam
)->sin6_port
;
224 *(&sin6
) = *SIN6(nam
);
226 /* KAME hack: embed scopeid */
227 if (in6_embedscope(&sin6
.sin6_addr
, &sin6
, inp
, NULL
,
229 lck_rw_done(pcbinfo
->ipi_lock
);
234 /* Sanitize local copy for address searches */
235 sin6
.sin6_flowinfo
= 0;
236 sin6
.sin6_scope_id
= 0;
239 if (IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
)) {
241 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
242 * allow compepte duplication of binding if
243 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
244 * and a multicast address is bound on both
245 * new and duplicated sockets.
247 if (so
->so_options
& SO_REUSEADDR
)
248 reuseport
= SO_REUSEADDR
|SO_REUSEPORT
;
249 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
252 ifa
= ifa_ifwithaddr(SA(&sin6
));
254 lck_rw_done(pcbinfo
->ipi_lock
);
256 return (EADDRNOTAVAIL
);
259 * XXX: bind to an anycast address might
260 * accidentally cause sending a packet with
261 * anycast source address. We should allow
262 * to bind to a deprecated address, since
263 * the application dare to use it.
266 if (((struct in6_ifaddr
*)ifa
)->ia6_flags
&
267 (IN6_IFF_ANYCAST
|IN6_IFF_NOTREADY
|
271 lck_rw_done(pcbinfo
->ipi_lock
);
273 return (EADDRNOTAVAIL
);
276 * Opportunistically determine the outbound
277 * interface that may be used; this may not
278 * hold true if we end up using a route
279 * going over a different interface, e.g.
280 * when sending to a local address. This
281 * will get updated again after sending.
283 outif
= ifa
->ifa_ifp
;
293 if (ntohs(lport
) < IPV6PORT_RESERVED
) {
294 cred
= kauth_cred_proc_ref(p
);
295 error
= priv_check_cred(cred
,
296 PRIV_NETINET_RESERVEDPORT
, 0);
297 kauth_cred_unref(&cred
);
299 lck_rw_done(pcbinfo
->ipi_lock
);
304 if (!IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
) &&
305 (u
= kauth_cred_getuid(so
->so_cred
)) != 0) {
306 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
307 &sin6
.sin6_addr
, lport
,
309 if (t
!= NULL
&& (!IN6_IS_ADDR_UNSPECIFIED(
311 !IN6_IS_ADDR_UNSPECIFIED(&t
->in6p_laddr
) ||
312 !(t
->inp_socket
->so_options
&
313 SO_REUSEPORT
)) && (u
!= kauth_cred_getuid(
314 t
->inp_socket
->so_cred
)) &&
315 !(t
->inp_socket
->so_flags
&
316 SOF_REUSESHAREUID
)) {
317 lck_rw_done(pcbinfo
->ipi_lock
);
321 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
322 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
323 struct sockaddr_in sin
;
325 in6_sin6_2_sin(&sin
, &sin6
);
326 t
= in_pcblookup_local_and_cleanup(
327 pcbinfo
, sin
.sin_addr
, lport
,
330 !(t
->inp_socket
->so_options
&
332 (kauth_cred_getuid(so
->so_cred
) !=
333 kauth_cred_getuid(t
->inp_socket
->
334 so_cred
)) && (t
->inp_laddr
.s_addr
!=
335 INADDR_ANY
|| SOCK_DOM(so
) ==
336 SOCK_DOM(t
->inp_socket
))) {
337 lck_rw_done(pcbinfo
->ipi_lock
);
343 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
344 &sin6
.sin6_addr
, lport
, wild
);
346 (reuseport
& t
->inp_socket
->so_options
) == 0) {
347 lck_rw_done(pcbinfo
->ipi_lock
);
351 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
352 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
353 struct sockaddr_in sin
;
355 in6_sin6_2_sin(&sin
, &sin6
);
356 t
= in_pcblookup_local_and_cleanup(pcbinfo
,
357 sin
.sin_addr
, lport
, wild
);
358 if (t
!= NULL
&& (reuseport
&
359 t
->inp_socket
->so_options
) == 0 &&
360 (t
->inp_laddr
.s_addr
!= INADDR_ANY
||
361 SOCK_DOM(so
) == SOCK_DOM(t
->inp_socket
))) {
362 lck_rw_done(pcbinfo
->ipi_lock
);
372 * We unlocked socket's protocol lock for a long time.
373 * The socket might have been dropped/defuncted.
374 * Checking if world has changed since.
376 if (inp
->inp_state
== INPCB_STATE_DEAD
) {
377 lck_rw_done(pcbinfo
->ipi_lock
);
378 return (ECONNABORTED
);
381 /* check if the socket got bound when the lock was released */
382 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
383 lck_rw_done(pcbinfo
->ipi_lock
);
387 if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
388 inp
->in6p_laddr
= sin6
.sin6_addr
;
389 inp
->in6p_last_outifp
= outif
;
394 if ((e
= in6_pcbsetport(&inp
->in6p_laddr
, inp
, p
, 1)) != 0) {
395 /* Undo any address bind from above. */
396 inp
->in6p_laddr
= in6addr_any
;
397 inp
->in6p_last_outifp
= NULL
;
398 lck_rw_done(pcbinfo
->ipi_lock
);
402 inp
->inp_lport
= lport
;
403 if (in_pcbinshash(inp
, 1) != 0) {
404 inp
->in6p_laddr
= in6addr_any
;
406 inp
->in6p_last_outifp
= NULL
;
407 lck_rw_done(pcbinfo
->ipi_lock
);
411 lck_rw_done(pcbinfo
->ipi_lock
);
412 sflt_notify(so
, sock_evt_bound
, NULL
);
417 * Transform old in6_pcbconnect() into an inner subroutine for new
418 * in6_pcbconnect(); do some validity-checking on the remote address
419 * (in "nam") and then determine local host address (i.e., which
420 * interface) to use to access that remote host.
422 * This routine may alter the caller-supplied remote address "nam".
424 * This routine might return an ifp with a reference held if the caller
425 * provides a non-NULL outif, even in the error case. The caller is
426 * responsible for releasing its reference.
429 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
430 struct in6_addr
*plocal_addr6
, struct ifnet
**outif
)
432 struct in6_addr
*addr6
= NULL
;
433 struct in6_addr src_storage
;
435 unsigned int ifscope
;
439 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
))
441 if (SIN6(nam
)->sin6_family
!= AF_INET6
)
442 return (EAFNOSUPPORT
);
443 if (SIN6(nam
)->sin6_port
== 0)
444 return (EADDRNOTAVAIL
);
446 /* KAME hack: embed scopeid */
447 if (in6_embedscope(&SIN6(nam
)->sin6_addr
, SIN6(nam
), inp
, NULL
, NULL
) != 0)
452 * If the destination address is UNSPECIFIED addr,
453 * use the loopback addr, e.g ::1.
455 if (IN6_IS_ADDR_UNSPECIFIED(&SIN6(nam
)->sin6_addr
))
456 SIN6(nam
)->sin6_addr
= in6addr_loopback
;
459 ifscope
= (inp
->inp_flags
& INP_BOUND_IF
) ?
460 inp
->inp_boundifp
->if_index
: IFSCOPE_NONE
;
463 * XXX: in6_selectsrc might replace the bound local address
464 * with the address specified by setsockopt(IPV6_PKTINFO).
465 * Is it the intended behavior?
467 * in6_selectsrc() might return outif with its reference held
468 * even in the error case; caller always needs to release it
471 addr6
= in6_selectsrc(SIN6(nam
), inp
->in6p_outputopts
, inp
,
472 &inp
->in6p_route
, outif
, &src_storage
, ifscope
, &error
);
475 struct rtentry
*rt
= inp
->in6p_route
.ro_rt
;
477 * If in6_selectsrc() returns a route, it should be one
478 * which points to the same ifp as outif. Just in case
479 * it isn't, use the one from the route for consistency.
480 * Otherwise if there is no route, leave outif alone as
481 * it could still be useful to the caller.
483 if (rt
!= NULL
&& rt
->rt_ifp
!= *outif
) {
484 ifnet_reference(rt
->rt_ifp
); /* for caller */
486 ifnet_release(*outif
);
492 if (outif
!= NULL
&& (*outif
) != NULL
&&
493 inp_restricted_send(inp
, *outif
)) {
494 soevent(inp
->inp_socket
,
495 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
496 error
= EHOSTUNREACH
;
499 error
= EADDRNOTAVAIL
;
503 *plocal_addr6
= *addr6
;
505 * Don't do pcblookup call here; return interface in
506 * plocal_addr6 and exit to caller, that will do the lookup.
513 * Connect from a socket to a specified address.
514 * Both address and port must be specified in argument sin.
515 * If don't have a local address for this socket yet,
519 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct proc
*p
)
521 struct in6_addr addr6
;
522 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)(void *)nam
;
525 struct ifnet
*outif
= NULL
;
526 struct socket
*so
= inp
->inp_socket
;
529 * Call inner routine, to assign local interface address.
530 * in6_pcbladdr() may automatically fill in sin6_scope_id.
532 * in6_pcbladdr() might return an ifp with its reference held
533 * even in the error case, so make sure that it's released
534 * whenever it's non-NULL.
536 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, &outif
)) != 0) {
537 if (outif
!= NULL
&& inp_restricted_send(inp
, outif
))
539 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
542 socket_unlock(so
, 0);
543 pcb
= in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
544 sin6
->sin6_port
, IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
) ?
545 &addr6
: &inp
->in6p_laddr
, inp
->inp_lport
, 0, NULL
);
548 in_pcb_checkstate(pcb
, WNT_RELEASE
, pcb
== inp
? 1 : 0);
552 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
553 if (inp
->inp_lport
== 0) {
554 error
= in6_pcbbind(inp
, NULL
, p
);
558 inp
->in6p_laddr
= addr6
;
559 inp
->in6p_last_outifp
= outif
; /* no reference needed */
560 inp
->in6p_flags
|= INP_IN6ADDR_ANY
;
562 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
563 /* lock inversion issue, mostly with udp multicast packets */
564 socket_unlock(so
, 0);
565 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
568 inp
->in6p_faddr
= sin6
->sin6_addr
;
569 inp
->inp_fport
= sin6
->sin6_port
;
570 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
571 nstat_pcb_invalidate_cache(inp
);
573 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
577 ifnet_release(outif
);
583 in6_pcbdisconnect(struct inpcb
*inp
)
585 struct socket
*so
= inp
->inp_socket
;
587 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
588 /* lock inversion issue, mostly with udp multicast packets */
589 socket_unlock(so
, 0);
590 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
593 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
594 nstat_pcb_cache(inp
);
595 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof (inp
->in6p_faddr
));
597 /* clear flowinfo - RFC 6437 */
598 inp
->inp_flow
&= ~IPV6_FLOWLABEL_MASK
;
600 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
602 * A multipath subflow socket would have its SS_NOFDREF set by default,
603 * so check for SOF_MP_SUBFLOW socket flag before detaching the PCB;
604 * when the socket is closed for real, SOF_MP_SUBFLOW would be cleared.
606 if (!(so
->so_flags
& SOF_MP_SUBFLOW
) && (so
->so_state
& SS_NOFDREF
))
611 in6_pcbdetach(struct inpcb
*inp
)
613 struct socket
*so
= inp
->inp_socket
;
615 if (so
->so_pcb
== NULL
) {
616 /* PCB has been disposed */
617 panic("%s: inp=%p so=%p proto=%d so_pcb is null!\n", __func__
,
618 inp
, so
, SOCK_PROTO(so
));
623 if (inp
->in6p_sp
!= NULL
) {
624 (void) ipsec6_delete_pcbpolicy(inp
);
629 * Let NetworkStatistics know this PCB is going away
630 * before we detach it.
633 (SOCK_PROTO(so
) == IPPROTO_TCP
|| SOCK_PROTO(so
) == IPPROTO_UDP
))
634 nstat_pcb_detach(inp
);
635 /* mark socket state as dead */
636 if (in_pcb_checkstate(inp
, WNT_STOPUSING
, 1) != WNT_STOPUSING
) {
637 panic("%s: so=%p proto=%d couldn't set to STOPUSING\n",
638 __func__
, so
, SOCK_PROTO(so
));
642 if (!(so
->so_flags
& SOF_PCBCLEARING
)) {
643 struct ip_moptions
*imo
;
644 struct ip6_moptions
*im6o
;
647 if (inp
->in6p_options
!= NULL
) {
648 m_freem(inp
->in6p_options
);
649 inp
->in6p_options
= NULL
;
651 ip6_freepcbopts(inp
->in6p_outputopts
);
652 ROUTE_RELEASE(&inp
->in6p_route
);
653 /* free IPv4 related resources in case of mapped addr */
654 if (inp
->inp_options
!= NULL
) {
655 (void) m_free(inp
->inp_options
);
656 inp
->inp_options
= NULL
;
658 im6o
= inp
->in6p_moptions
;
659 inp
->in6p_moptions
= NULL
;
661 imo
= inp
->inp_moptions
;
662 inp
->inp_moptions
= NULL
;
664 sofreelastref(so
, 0);
665 inp
->inp_state
= INPCB_STATE_DEAD
;
666 /* makes sure we're not called twice from so_close */
667 so
->so_flags
|= SOF_PCBCLEARING
;
669 inpcb_gc_sched(inp
->inp_pcbinfo
, INPCB_TIMER_FAST
);
672 * See inp_join_group() for why we need to unlock
674 if (im6o
!= NULL
|| imo
!= NULL
) {
675 socket_unlock(so
, 0);
686 in6_sockaddr(in_port_t port
, struct in6_addr
*addr_p
)
688 struct sockaddr_in6
*sin6
;
690 MALLOC(sin6
, struct sockaddr_in6
*, sizeof (*sin6
), M_SONAME
, M_WAITOK
);
693 bzero(sin6
, sizeof (*sin6
));
694 sin6
->sin6_family
= AF_INET6
;
695 sin6
->sin6_len
= sizeof (*sin6
);
696 sin6
->sin6_port
= port
;
697 sin6
->sin6_addr
= *addr_p
;
699 /* would be good to use sa6_recoverscope(), except for locking */
700 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
701 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
703 sin6
->sin6_scope_id
= 0; /* XXX */
704 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
705 sin6
->sin6_addr
.s6_addr16
[1] = 0;
707 return ((struct sockaddr
*)sin6
);
711 in6_sockaddr_s(in_port_t port
, struct in6_addr
*addr_p
,
712 struct sockaddr_in6
*sin6
)
714 bzero(sin6
, sizeof (*sin6
));
715 sin6
->sin6_family
= AF_INET6
;
716 sin6
->sin6_len
= sizeof (*sin6
);
717 sin6
->sin6_port
= port
;
718 sin6
->sin6_addr
= *addr_p
;
720 /* would be good to use sa6_recoverscope(), except for locking */
721 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
722 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
724 sin6
->sin6_scope_id
= 0; /* XXX */
725 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
726 sin6
->sin6_addr
.s6_addr16
[1] = 0;
730 * The calling convention of in6_getsockaddr() and in6_getpeeraddr() was
731 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
732 * in struct pr_usrreqs, so that protocols can just reference then directly
733 * without the need for a wrapper function.
736 in6_getsockaddr(struct socket
*so
, struct sockaddr
**nam
)
739 struct in6_addr addr
;
742 if ((inp
= sotoinpcb(so
)) == NULL
)
745 port
= inp
->inp_lport
;
746 addr
= inp
->in6p_laddr
;
748 *nam
= in6_sockaddr(port
, &addr
);
755 in6_getsockaddr_s(struct socket
*so
, struct sockaddr_storage
*ss
)
758 struct in6_addr addr
;
762 bzero(ss
, sizeof (*ss
));
764 if ((inp
= sotoinpcb(so
)) == NULL
766 || (necp_socket_should_use_flow_divert(inp
))
769 return (inp
== NULL
? EINVAL
: EPROTOTYPE
);
771 port
= inp
->inp_lport
;
772 addr
= inp
->in6p_laddr
;
774 in6_sockaddr_s(port
, &addr
, SIN6(ss
));
779 in6_getpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
782 struct in6_addr addr
;
785 if ((inp
= sotoinpcb(so
)) == NULL
)
788 port
= inp
->inp_fport
;
789 addr
= inp
->in6p_faddr
;
791 *nam
= in6_sockaddr(port
, &addr
);
798 in6_getpeeraddr_s(struct socket
*so
, struct sockaddr_storage
*ss
)
801 struct in6_addr addr
;
805 bzero(ss
, sizeof (*ss
));
807 if ((inp
= sotoinpcb(so
)) == NULL
809 || (necp_socket_should_use_flow_divert(inp
))
812 return (inp
== NULL
? EINVAL
: EPROTOTYPE
);
814 port
= inp
->inp_fport
;
815 addr
= inp
->in6p_faddr
;
817 in6_sockaddr_s(port
, &addr
, SIN6(ss
));
822 in6_mapped_sockaddr(struct socket
*so
, struct sockaddr
**nam
)
824 struct inpcb
*inp
= sotoinpcb(so
);
829 if (inp
->inp_vflag
& INP_IPV4
) {
830 error
= in_getsockaddr(so
, nam
);
832 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
834 /* scope issues will be handled in in6_getsockaddr(). */
835 error
= in6_getsockaddr(so
, nam
);
841 in6_mapped_peeraddr(struct socket
*so
, struct sockaddr
**nam
)
843 struct inpcb
*inp
= sotoinpcb(so
);
848 if (inp
->inp_vflag
& INP_IPV4
) {
849 error
= in_getpeeraddr(so
, nam
);
851 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
853 /* scope issues will be handled in in6_getpeeraddr(). */
854 error
= in6_getpeeraddr(so
, nam
);
860 * Pass some notification to all connections of a protocol
861 * associated with address dst. The local address and/or port numbers
862 * may be specified to limit the search. The "usual action" will be
863 * taken, depending on the ctlinput cmd. The caller must filter any
864 * cmds that are uninteresting (e.g., no error in the map).
865 * Call the protocol specific routine (if any) to report
866 * any errors for each matching socket.
869 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, u_int fport_arg
,
870 const struct sockaddr
*src
, u_int lport_arg
, int cmd
, void *cmdarg
,
871 void (*notify
)(struct inpcb
*, int))
873 struct inpcbhead
*head
= pcbinfo
->ipi_listhead
;
874 struct inpcb
*inp
, *ninp
;
875 struct sockaddr_in6 sa6_src
, *sa6_dst
;
876 u_short fport
= fport_arg
, lport
= lport_arg
;
880 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
)
883 sa6_dst
= (struct sockaddr_in6
*)(void *)dst
;
884 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
))
888 * note that src can be NULL when we get notify by local fragmentation.
890 sa6_src
= (src
== NULL
) ?
891 sa6_any
: *(struct sockaddr_in6
*)(uintptr_t)(size_t)src
;
892 flowinfo
= sa6_src
.sin6_flowinfo
;
895 * Redirects go to all references to the destination,
896 * and use in6_rtchange to invalidate the route cache.
897 * Dead host indications: also use in6_rtchange to invalidate
898 * the cache, and deliver the error to all the sockets.
899 * Otherwise, if we have knowledge of the local port and address,
900 * deliver only to that socket.
902 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
905 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof (sa6_src
.sin6_addr
));
907 if (cmd
!= PRC_HOSTDEAD
)
908 notify
= in6_rtchange
;
910 errno
= inet6ctlerrmap
[cmd
];
911 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
912 for (inp
= LIST_FIRST(head
); inp
!= NULL
; inp
= ninp
) {
913 ninp
= LIST_NEXT(inp
, inp_list
);
915 if (!(inp
->inp_vflag
& INP_IPV6
))
919 * If the error designates a new path MTU for a destination
920 * and the application (associated with this socket) wanted to
921 * know the value, notify. Note that we notify for all
922 * disconnected sockets if the corresponding application
923 * wanted. This is because some UDP applications keep sending
924 * sockets disconnected.
925 * XXX: should we avoid to notify the value to TCP sockets?
927 if (cmd
== PRC_MSGSIZE
&& (inp
->inp_flags
& IN6P_MTU
) != 0 &&
928 (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) ||
929 IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
930 &sa6_dst
->sin6_addr
))) {
931 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)(void *)dst
,
932 (u_int32_t
*)cmdarg
);
936 * Detect if we should notify the error. If no source and
937 * destination ports are specifed, but non-zero flowinfo and
938 * local address match, notify the error. This is the case
939 * when the error is delivered with an encrypted buffer
940 * by ESP. Otherwise, just compare addresses and ports
943 if (lport
== 0 && fport
== 0 && flowinfo
&&
944 inp
->inp_socket
!= NULL
&&
945 flowinfo
== (inp
->inp_flow
& IPV6_FLOWLABEL_MASK
) &&
946 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
))
948 else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
949 &sa6_dst
->sin6_addr
) || inp
->inp_socket
== NULL
||
950 (lport
&& inp
->inp_lport
!= lport
) ||
951 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
952 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
953 &sa6_src
.sin6_addr
)) || (fport
&& inp
->inp_fport
!= fport
))
958 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) ==
961 socket_lock(inp
->inp_socket
, 1);
962 (*notify
)(inp
, errno
);
963 (void) in_pcb_checkstate(inp
, WNT_RELEASE
, 1);
964 socket_unlock(inp
->inp_socket
, 1);
967 lck_rw_done(pcbinfo
->ipi_lock
);
971 * Lookup a PCB based on the local address and port.
974 in6_pcblookup_local(struct inpcbinfo
*pcbinfo
, struct in6_addr
*laddr
,
975 u_int lport_arg
, int wild_okay
)
978 int matchwild
= 3, wildcard
;
979 u_short lport
= lport_arg
;
980 struct inpcbporthead
*porthash
;
981 struct inpcb
*match
= NULL
;
982 struct inpcbport
*phd
;
985 struct inpcbhead
*head
;
987 * Look for an unconnected (wildcard foreign addr) PCB that
988 * matches the local address and port we're looking for.
990 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
991 pcbinfo
->ipi_hashmask
)];
992 LIST_FOREACH(inp
, head
, inp_hash
) {
993 if (!(inp
->inp_vflag
& INP_IPV6
))
995 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
996 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
997 inp
->inp_lport
== lport
) {
1010 * Best fit PCB lookup.
1012 * First see if this local port is in use by looking on the
1015 porthash
= &pcbinfo
->ipi_porthashbase
[INP_PCBPORTHASH(lport
,
1016 pcbinfo
->ipi_porthashmask
)];
1017 LIST_FOREACH(phd
, porthash
, phd_hash
) {
1018 if (phd
->phd_port
== lport
)
1023 * Port is in use by one or more PCBs. Look for best
1026 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
1028 if (!(inp
->inp_vflag
& INP_IPV6
))
1030 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
1032 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
1033 if (IN6_IS_ADDR_UNSPECIFIED(laddr
))
1035 else if (!IN6_ARE_ADDR_EQUAL(
1036 &inp
->in6p_laddr
, laddr
))
1039 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
))
1042 if (wildcard
< matchwild
) {
1044 matchwild
= wildcard
;
1045 if (matchwild
== 0) {
1055 * Check for alternatives when higher level complains
1056 * about service problems. For now, invalidate cached
1057 * routing information. If the route was created dynamically
1058 * (by a redirect), time to try a default gateway again.
1061 in6_losing(struct inpcb
*in6p
)
1065 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
1067 if (rt
->rt_flags
& RTF_DYNAMIC
) {
1069 * Prevent another thread from modifying rt_key,
1070 * rt_gateway via rt_setgate() after the rt_lock
1071 * is dropped by marking the route as defunct.
1073 rt
->rt_flags
|= RTF_CONDEMNED
;
1075 (void) rtrequest(RTM_DELETE
, rt_key(rt
),
1076 rt
->rt_gateway
, rt_mask(rt
), rt
->rt_flags
, NULL
);
1081 * A new route can be allocated
1082 * the next time output is attempted.
1085 ROUTE_RELEASE(&in6p
->in6p_route
);
1089 * After a routing change, flush old routing
1090 * and allocate a (hopefully) better one.
1093 in6_rtchange(struct inpcb
*inp
, int errno
)
1095 #pragma unused(errno)
1097 * A new route can be allocated the next time
1098 * output is attempted.
1100 ROUTE_RELEASE(&inp
->in6p_route
);
1104 * Check if PCB exists hash list. Also returns uid and gid of socket
1107 in6_pcblookup_hash_exists(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1108 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1109 uid_t
*uid
, gid_t
*gid
, struct ifnet
*ifp
)
1111 struct inpcbhead
*head
;
1113 u_short fport
= fport_arg
, lport
= lport_arg
;
1119 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1122 * First look for an exact match.
1124 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1125 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1126 LIST_FOREACH(inp
, head
, inp_hash
) {
1127 if (!(inp
->inp_vflag
& INP_IPV6
))
1130 if (inp_restricted_recv(inp
, ifp
))
1133 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1134 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1135 inp
->inp_fport
== fport
&&
1136 inp
->inp_lport
== lport
) {
1137 if ((found
= (inp
->inp_socket
!= NULL
))) {
1139 * Found. Check if pcb is still valid
1141 *uid
= kauth_cred_getuid(
1142 inp
->inp_socket
->so_cred
);
1143 *gid
= kauth_cred_getgid(
1144 inp
->inp_socket
->so_cred
);
1146 lck_rw_done(pcbinfo
->ipi_lock
);
1151 struct inpcb
*local_wild
= NULL
;
1153 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1154 pcbinfo
->ipi_hashmask
)];
1155 LIST_FOREACH(inp
, head
, inp_hash
) {
1156 if (!(inp
->inp_vflag
& INP_IPV6
))
1159 if (inp_restricted_recv(inp
, ifp
))
1162 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1163 inp
->inp_lport
== lport
) {
1164 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1166 found
= (inp
->inp_socket
!= NULL
);
1168 *uid
= kauth_cred_getuid(
1169 inp
->inp_socket
->so_cred
);
1170 *gid
= kauth_cred_getgid(
1171 inp
->inp_socket
->so_cred
);
1173 lck_rw_done(pcbinfo
->ipi_lock
);
1175 } else if (IN6_IS_ADDR_UNSPECIFIED(
1176 &inp
->in6p_laddr
)) {
1182 if ((found
= (local_wild
->inp_socket
!= NULL
))) {
1183 *uid
= kauth_cred_getuid(
1184 local_wild
->inp_socket
->so_cred
);
1185 *gid
= kauth_cred_getgid(
1186 local_wild
->inp_socket
->so_cred
);
1188 lck_rw_done(pcbinfo
->ipi_lock
);
1196 lck_rw_done(pcbinfo
->ipi_lock
);
1201 * Lookup PCB in hash list.
1204 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1205 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1208 struct inpcbhead
*head
;
1210 u_short fport
= fport_arg
, lport
= lport_arg
;
1212 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1215 * First look for an exact match.
1217 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1218 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1219 LIST_FOREACH(inp
, head
, inp_hash
) {
1220 if (!(inp
->inp_vflag
& INP_IPV6
))
1223 if (inp_restricted_recv(inp
, ifp
))
1226 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1227 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1228 inp
->inp_fport
== fport
&&
1229 inp
->inp_lport
== lport
) {
1231 * Found. Check if pcb is still valid
1233 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) !=
1235 lck_rw_done(pcbinfo
->ipi_lock
);
1238 /* it's there but dead, say it isn't found */
1239 lck_rw_done(pcbinfo
->ipi_lock
);
1245 struct inpcb
*local_wild
= NULL
;
1247 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1248 pcbinfo
->ipi_hashmask
)];
1249 LIST_FOREACH(inp
, head
, inp_hash
) {
1250 if (!(inp
->inp_vflag
& INP_IPV6
))
1253 if (inp_restricted_recv(inp
, ifp
))
1256 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1257 inp
->inp_lport
== lport
) {
1258 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1260 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
,
1261 0) != WNT_STOPUSING
) {
1262 lck_rw_done(pcbinfo
->ipi_lock
);
1265 /* dead; say it isn't found */
1266 lck_rw_done(pcbinfo
->ipi_lock
);
1269 } else if (IN6_IS_ADDR_UNSPECIFIED(
1270 &inp
->in6p_laddr
)) {
1275 if (local_wild
&& in_pcb_checkstate(local_wild
,
1276 WNT_ACQUIRE
, 0) != WNT_STOPUSING
) {
1277 lck_rw_done(pcbinfo
->ipi_lock
);
1278 return (local_wild
);
1280 lck_rw_done(pcbinfo
->ipi_lock
);
1288 lck_rw_done(pcbinfo
->ipi_lock
);
1293 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
1297 ip
= mtod(m
, struct ip6_hdr
*);
1298 bzero(sin6
, sizeof (*sin6
));
1299 sin6
->sin6_len
= sizeof (*sin6
);
1300 sin6
->sin6_family
= AF_INET6
;
1301 sin6
->sin6_addr
= ip
->ip6_src
;
1302 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
1303 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1304 if ((m
->m_pkthdr
.pkt_flags
& (PKTF_LOOP
|PKTF_IFAINFO
)) ==
1305 (PKTF_LOOP
|PKTF_IFAINFO
))
1306 sin6
->sin6_scope_id
= m
->m_pkthdr
.src_ifindex
;
1307 else if (m
->m_pkthdr
.rcvif
!= NULL
)
1308 sin6
->sin6_scope_id
= m
->m_pkthdr
.rcvif
->if_index
;
1313 * The following routines implement this scheme:
1315 * Callers of ip6_output() that intend to cache the route in the inpcb pass
1316 * a local copy of the struct route to ip6_output(). Using a local copy of
1317 * the cached route significantly simplifies things as IP no longer has to
1318 * worry about having exclusive access to the passed in struct route, since
1319 * it's defined in the caller's stack; in essence, this allows for a lock-
1320 * less operation when updating the struct route at the IP level and below,
1321 * whenever necessary. The scheme works as follows:
1323 * Prior to dropping the socket's lock and calling ip6_output(), the caller
1324 * copies the struct route from the inpcb into its stack, and adds a reference
1325 * to the cached route entry, if there was any. The socket's lock is then
1326 * dropped and ip6_output() is called with a pointer to the copy of struct
1327 * route defined on the stack (not to the one in the inpcb.)
1329 * Upon returning from ip6_output(), the caller then acquires the socket's
1330 * lock and synchronizes the cache; if there is no route cached in the inpcb,
1331 * it copies the local copy of struct route (which may or may not contain any
1332 * route) back into the cache; otherwise, if the inpcb has a route cached in
1333 * it, the one in the local copy will be freed, if there's any. Trashing the
1334 * cached route in the inpcb can be avoided because ip6_output() is single-
1335 * threaded per-PCB (i.e. multiple transmits on a PCB are always serialized
1336 * by the socket/transport layer.)
1339 in6p_route_copyout(struct inpcb
*inp
, struct route_in6
*dst
)
1341 struct route_in6
*src
= &inp
->in6p_route
;
1343 lck_mtx_assert(&inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1345 /* Minor sanity check */
1346 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
)
1347 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1349 route_copyout((struct route
*)dst
, (struct route
*)src
, sizeof (*dst
));
1353 in6p_route_copyin(struct inpcb
*inp
, struct route_in6
*src
)
1355 struct route_in6
*dst
= &inp
->in6p_route
;
1357 lck_mtx_assert(&inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1359 /* Minor sanity check */
1360 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
)
1361 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1363 route_copyin((struct route
*)src
, (struct route
*)dst
, sizeof (*src
));