2 * Copyright (c) 2003-2020 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>
112 #include <net/restricted_in_port.h>
114 #include <netinet/in.h>
115 #include <netinet/in_var.h>
116 #include <netinet/in_systm.h>
117 #include <netinet/ip6.h>
118 #include <netinet/ip_var.h>
120 #include <netinet6/ip6_var.h>
121 #include <netinet6/nd6.h>
122 #include <netinet/in_pcb.h>
123 #include <netinet6/in6_pcb.h>
125 #include <net/if_types.h>
126 #include <net/if_var.h>
128 #include <kern/kern_types.h>
129 #include <kern/zalloc.h>
132 #include <netinet6/ipsec.h>
133 #include <netinet6/ipsec6.h>
134 #include <netinet6/ah.h>
135 #include <netinet6/ah6.h>
136 #include <netkey/key.h>
140 #include <net/necp.h>
144 * in6_pcblookup_local_and_cleanup does everything
145 * in6_pcblookup_local does but it checks for a socket
146 * that's going away. Since we know that the lock is
147 * held read+write when this function is called, we
148 * can safely dispose of this socket like the slow
149 * timer would usually do and return NULL. This is
152 static struct inpcb
*
153 in6_pcblookup_local_and_cleanup(struct inpcbinfo
*pcbinfo
,
154 struct in6_addr
*laddr
, u_int lport_arg
, int wild_okay
)
158 /* Perform normal lookup */
159 inp
= in6_pcblookup_local(pcbinfo
, laddr
, lport_arg
, wild_okay
);
161 /* Check if we found a match but it's waiting to be disposed */
162 if (inp
!= NULL
&& inp
->inp_wantcnt
== WNT_STOPUSING
) {
163 struct socket
*so
= inp
->inp_socket
;
167 if (so
->so_usecount
== 0) {
168 if (inp
->inp_state
!= INPCB_STATE_DEAD
) {
171 in_pcbdispose(inp
); /* will unlock & destroy */
174 socket_unlock(so
, 0);
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
;
194 #if XNU_TARGET_OS_OSX
197 #endif /* XNU_TARGET_OS_OSX */
199 if (TAILQ_EMPTY(&in6_ifaddrhead
)) { /* XXX broken! */
200 return EADDRNOTAVAIL
;
202 if (!(so
->so_options
& (SO_REUSEADDR
| SO_REUSEPORT
))) {
206 socket_unlock(so
, 0); /* keep reference */
207 lck_rw_lock_exclusive(pcbinfo
->ipi_lock
);
208 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
209 /* another thread completed the bind */
210 lck_rw_done(pcbinfo
->ipi_lock
);
215 bzero(&sin6
, sizeof(sin6
));
217 if (nam
->sa_len
!= sizeof(struct sockaddr_in6
)) {
218 lck_rw_done(pcbinfo
->ipi_lock
);
225 if (nam
->sa_family
!= AF_INET6
) {
226 lck_rw_done(pcbinfo
->ipi_lock
);
230 lport
= SIN6(nam
)->sin6_port
;
232 *(&sin6
) = *SIN6(nam
);
234 /* KAME hack: embed scopeid */
235 if (in6_embedscope(&sin6
.sin6_addr
, &sin6
, inp
, NULL
,
237 lck_rw_done(pcbinfo
->ipi_lock
);
242 /* Sanitize local copy for address searches */
243 sin6
.sin6_flowinfo
= 0;
244 sin6
.sin6_scope_id
= 0;
247 if (IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
)) {
249 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
250 * allow compepte duplication of binding if
251 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
252 * and a multicast address is bound on both
253 * new and duplicated sockets.
255 if (so
->so_options
& SO_REUSEADDR
) {
256 reuseport
= SO_REUSEADDR
| SO_REUSEPORT
;
258 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
261 ifa
= ifa_ifwithaddr(SA(&sin6
));
263 lck_rw_done(pcbinfo
->ipi_lock
);
265 return EADDRNOTAVAIL
;
268 * XXX: bind to an anycast address might
269 * accidentally cause sending a packet with
270 * anycast source address. We should allow
271 * to bind to a deprecated address, since
272 * the application dare to use it.
275 if (((struct in6_ifaddr
*)ifa
)->ia6_flags
&
276 (IN6_IFF_ANYCAST
| IN6_IFF_NOTREADY
|
277 IN6_IFF_DETACHED
| IN6_IFF_CLAT46
)) {
280 lck_rw_done(pcbinfo
->ipi_lock
);
282 return EADDRNOTAVAIL
;
285 * Opportunistically determine the outbound
286 * interface that may be used; this may not
287 * hold true if we end up using a route
288 * going over a different interface, e.g.
289 * when sending to a local address. This
290 * will get updated again after sending.
292 outif
= ifa
->ifa_ifp
;
303 #if XNU_TARGET_OS_OSX
304 if (ntohs(lport
) < IPV6PORT_RESERVED
&&
305 !IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
) &&
306 !(inp
->inp_flags2
& INP2_EXTERNAL_PORT
)) {
307 cred
= kauth_cred_proc_ref(p
);
308 error
= priv_check_cred(cred
,
309 PRIV_NETINET_RESERVEDPORT
, 0);
310 kauth_cred_unref(&cred
);
312 lck_rw_done(pcbinfo
->ipi_lock
);
317 #endif /* XNU_TARGET_OS_OSX */
319 * Check wether the process is allowed to bind to a restricted port
321 if (!current_task_can_use_restricted_in_port(lport
,
322 (uint8_t)SOCK_PROTO(so
), PORT_FLAGS_BSD
)) {
323 lck_rw_done(pcbinfo
->ipi_lock
);
328 if (!IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
) &&
329 (u
= kauth_cred_getuid(so
->so_cred
)) != 0) {
330 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
331 &sin6
.sin6_addr
, lport
,
334 (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
) ||
335 !IN6_IS_ADDR_UNSPECIFIED(&t
->in6p_laddr
) ||
336 !(t
->inp_socket
->so_options
& SO_REUSEPORT
)) &&
337 (u
!= kauth_cred_getuid(t
->inp_socket
->so_cred
)) &&
338 !(t
->inp_socket
->so_flags
& SOF_REUSESHAREUID
) &&
339 (!(t
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
340 !(inp
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
341 uuid_compare(t
->necp_client_uuid
, inp
->necp_client_uuid
) != 0)) {
342 lck_rw_done(pcbinfo
->ipi_lock
);
346 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
347 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
348 struct sockaddr_in sin
;
350 in6_sin6_2_sin(&sin
, &sin6
);
351 t
= in_pcblookup_local_and_cleanup(
352 pcbinfo
, sin
.sin_addr
, lport
,
355 !(t
->inp_socket
->so_options
& SO_REUSEPORT
) &&
356 (kauth_cred_getuid(so
->so_cred
) !=
357 kauth_cred_getuid(t
->inp_socket
->so_cred
)) &&
358 (t
->inp_laddr
.s_addr
!= INADDR_ANY
||
359 SOCK_DOM(so
) == SOCK_DOM(t
->inp_socket
)) &&
360 (!(t
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
361 !(inp
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
362 uuid_compare(t
->necp_client_uuid
, inp
->necp_client_uuid
) != 0)) {
363 lck_rw_done(pcbinfo
->ipi_lock
);
370 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
371 &sin6
.sin6_addr
, lport
, wild
);
373 (reuseport
& t
->inp_socket
->so_options
) == 0 &&
374 (!(t
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
375 !(inp
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
376 uuid_compare(t
->necp_client_uuid
, inp
->necp_client_uuid
) != 0)) {
377 lck_rw_done(pcbinfo
->ipi_lock
);
381 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
382 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
383 struct sockaddr_in sin
;
385 in6_sin6_2_sin(&sin
, &sin6
);
386 t
= in_pcblookup_local_and_cleanup(pcbinfo
,
387 sin
.sin_addr
, lport
, wild
);
388 if (t
!= NULL
&& (reuseport
&
389 t
->inp_socket
->so_options
) == 0 &&
390 (t
->inp_laddr
.s_addr
!= INADDR_ANY
||
391 SOCK_DOM(so
) == SOCK_DOM(t
->inp_socket
)) &&
392 (!(t
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
393 !(inp
->inp_flags2
& INP2_EXTERNAL_PORT
) ||
394 uuid_compare(t
->necp_client_uuid
, inp
->necp_client_uuid
) != 0)) {
395 lck_rw_done(pcbinfo
->ipi_lock
);
405 * We unlocked socket's protocol lock for a long time.
406 * The socket might have been dropped/defuncted.
407 * Checking if world has changed since.
409 if (inp
->inp_state
== INPCB_STATE_DEAD
) {
410 lck_rw_done(pcbinfo
->ipi_lock
);
414 /* check if the socket got bound when the lock was released */
415 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
416 lck_rw_done(pcbinfo
->ipi_lock
);
420 if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
421 inp
->in6p_laddr
= sin6
.sin6_addr
;
422 inp
->in6p_last_outifp
= outif
;
427 if ((e
= in6_pcbsetport(&inp
->in6p_laddr
, inp
, p
, 1)) != 0) {
428 /* Undo any address bind from above. */
429 inp
->in6p_laddr
= in6addr_any
;
430 inp
->in6p_last_outifp
= NULL
;
431 lck_rw_done(pcbinfo
->ipi_lock
);
435 inp
->inp_lport
= lport
;
436 if (in_pcbinshash(inp
, 1) != 0) {
437 inp
->in6p_laddr
= in6addr_any
;
439 inp
->in6p_last_outifp
= NULL
;
440 lck_rw_done(pcbinfo
->ipi_lock
);
444 lck_rw_done(pcbinfo
->ipi_lock
);
445 sflt_notify(so
, sock_evt_bound
, NULL
);
450 * Transform old in6_pcbconnect() into an inner subroutine for new
451 * in6_pcbconnect(); do some validity-checking on the remote address
452 * (in "nam") and then determine local host address (i.e., which
453 * interface) to use to access that remote host.
455 * This routine may alter the caller-supplied remote address "nam".
457 * This routine might return an ifp with a reference held if the caller
458 * provides a non-NULL outif, even in the error case. The caller is
459 * responsible for releasing its reference.
462 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
463 struct in6_addr
*plocal_addr6
, struct ifnet
**outif
)
465 struct in6_addr
*addr6
= NULL
;
466 struct in6_addr src_storage
;
468 unsigned int ifscope
;
473 if (nam
->sa_len
!= sizeof(struct sockaddr_in6
)) {
476 if (SIN6(nam
)->sin6_family
!= AF_INET6
) {
479 if (SIN6(nam
)->sin6_port
== 0) {
480 return EADDRNOTAVAIL
;
483 /* KAME hack: embed scopeid */
484 if (in6_embedscope(&SIN6(nam
)->sin6_addr
, SIN6(nam
), inp
, NULL
, NULL
) != 0) {
488 if (!TAILQ_EMPTY(&in6_ifaddrhead
)) {
490 * If the destination address is UNSPECIFIED addr,
491 * use the loopback addr, e.g ::1.
493 if (IN6_IS_ADDR_UNSPECIFIED(&SIN6(nam
)->sin6_addr
)) {
494 SIN6(nam
)->sin6_addr
= in6addr_loopback
;
498 ifscope
= (inp
->inp_flags
& INP_BOUND_IF
) ?
499 inp
->inp_boundifp
->if_index
: IFSCOPE_NONE
;
502 * XXX: in6_selectsrc might replace the bound local address
503 * with the address specified by setsockopt(IPV6_PKTINFO).
504 * Is it the intended behavior?
506 * in6_selectsrc() might return outif with its reference held
507 * even in the error case; caller always needs to release it
510 addr6
= in6_selectsrc(SIN6(nam
), inp
->in6p_outputopts
, inp
,
511 &inp
->in6p_route
, outif
, &src_storage
, ifscope
, &error
);
514 struct rtentry
*rt
= inp
->in6p_route
.ro_rt
;
516 * If in6_selectsrc() returns a route, it should be one
517 * which points to the same ifp as outif. Just in case
518 * it isn't, use the one from the route for consistency.
519 * Otherwise if there is no route, leave outif alone as
520 * it could still be useful to the caller.
522 if (rt
!= NULL
&& rt
->rt_ifp
!= *outif
) {
523 ifnet_reference(rt
->rt_ifp
); /* for caller */
524 if (*outif
!= NULL
) {
525 ifnet_release(*outif
);
532 if (outif
!= NULL
&& (*outif
) != NULL
&&
533 inp_restricted_send(inp
, *outif
)) {
534 soevent(inp
->inp_socket
,
535 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
536 error
= EHOSTUNREACH
;
539 error
= EADDRNOTAVAIL
;
544 *plocal_addr6
= *addr6
;
546 * Don't do pcblookup call here; return interface in
547 * plocal_addr6 and exit to caller, that will do the lookup.
554 * Connect from a socket to a specified address.
555 * Both address and port must be specified in argument sin.
556 * If don't have a local address for this socket yet,
560 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct proc
*p
)
562 struct in6_addr addr6
;
563 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)(void *)nam
;
566 struct ifnet
*outif
= NULL
;
567 struct socket
*so
= inp
->inp_socket
;
570 so
->so_state_change_cnt
++;
573 if (so
->so_proto
->pr_protocol
== IPPROTO_UDP
&&
574 sin6
->sin6_port
== htons(53) && !(so
->so_flags1
& SOF1_DNS_COUNTED
)) {
575 so
->so_flags1
|= SOF1_DNS_COUNTED
;
576 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet_dgram_dns
);
580 * Call inner routine, to assign local interface address.
581 * in6_pcbladdr() may automatically fill in sin6_scope_id.
583 * in6_pcbladdr() might return an ifp with its reference held
584 * even in the error case, so make sure that it's released
585 * whenever it's non-NULL.
587 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, &outif
)) != 0) {
588 if (outif
!= NULL
&& inp_restricted_send(inp
, outif
)) {
590 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
594 socket_unlock(so
, 0);
595 pcb
= in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
596 sin6
->sin6_port
, IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
) ?
597 &addr6
: &inp
->in6p_laddr
, inp
->inp_lport
, 0, NULL
);
600 in_pcb_checkstate(pcb
, WNT_RELEASE
, pcb
== inp
? 1 : 0);
604 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
605 if (inp
->inp_lport
== 0) {
606 error
= in6_pcbbind(inp
, NULL
, p
);
611 inp
->in6p_laddr
= addr6
;
612 inp
->in6p_last_outifp
= outif
; /* no reference needed */
613 inp
->in6p_flags
|= INP_IN6ADDR_ANY
;
615 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
616 /* lock inversion issue, mostly with udp multicast packets */
617 socket_unlock(so
, 0);
618 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
621 inp
->in6p_faddr
= sin6
->sin6_addr
;
622 inp
->inp_fport
= sin6
->sin6_port
;
623 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
) {
624 nstat_pcb_invalidate_cache(inp
);
627 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
631 ifnet_release(outif
);
638 in6_pcbdisconnect(struct inpcb
*inp
)
640 struct socket
*so
= inp
->inp_socket
;
644 so
->so_state_change_cnt
++;
648 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
649 /* lock inversion issue, mostly with udp multicast packets */
650 socket_unlock(so
, 0);
651 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
654 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
) {
655 nstat_pcb_cache(inp
);
657 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof(inp
->in6p_faddr
));
659 /* clear flowinfo - RFC 6437 */
660 inp
->inp_flow
&= ~IPV6_FLOWLABEL_MASK
;
662 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
664 * A multipath subflow socket would have its SS_NOFDREF set by default,
665 * so check for SOF_MP_SUBFLOW socket flag before detaching the PCB;
666 * when the socket is closed for real, SOF_MP_SUBFLOW would be cleared.
668 if (!(so
->so_flags
& SOF_MP_SUBFLOW
) && (so
->so_state
& SS_NOFDREF
)) {
674 in6_pcbdetach(struct inpcb
*inp
)
676 struct socket
*so
= inp
->inp_socket
;
678 if (so
->so_pcb
== NULL
) {
679 /* PCB has been disposed */
680 panic("%s: inp=%p so=%p proto=%d so_pcb is null!\n", __func__
,
681 inp
, so
, SOCK_PROTO(so
));
686 if (inp
->in6p_sp
!= NULL
) {
687 (void) ipsec6_delete_pcbpolicy(inp
);
691 if (inp
->inp_stat
!= NULL
&& SOCK_PROTO(so
) == IPPROTO_UDP
) {
692 if (inp
->inp_stat
->rxpackets
== 0 && inp
->inp_stat
->txpackets
== 0) {
693 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet6_dgram_no_data
);
698 * Let NetworkStatistics know this PCB is going away
699 * before we detach it.
702 (SOCK_PROTO(so
) == IPPROTO_TCP
|| SOCK_PROTO(so
) == IPPROTO_UDP
)) {
703 nstat_pcb_detach(inp
);
705 /* mark socket state as dead */
706 if (in_pcb_checkstate(inp
, WNT_STOPUSING
, 1) != WNT_STOPUSING
) {
707 panic("%s: so=%p proto=%d couldn't set to STOPUSING\n",
708 __func__
, so
, SOCK_PROTO(so
));
712 if (!(so
->so_flags
& SOF_PCBCLEARING
)) {
713 struct ip_moptions
*imo
;
714 struct ip6_moptions
*im6o
;
717 if (inp
->in6p_options
!= NULL
) {
718 m_freem(inp
->in6p_options
);
719 inp
->in6p_options
= NULL
;
721 ip6_freepcbopts(inp
->in6p_outputopts
);
722 inp
->in6p_outputopts
= NULL
;
723 ROUTE_RELEASE(&inp
->in6p_route
);
724 /* free IPv4 related resources in case of mapped addr */
725 if (inp
->inp_options
!= NULL
) {
726 (void) m_free(inp
->inp_options
);
727 inp
->inp_options
= NULL
;
729 im6o
= inp
->in6p_moptions
;
730 inp
->in6p_moptions
= NULL
;
732 imo
= inp
->inp_moptions
;
733 inp
->inp_moptions
= NULL
;
735 sofreelastref(so
, 0);
736 inp
->inp_state
= INPCB_STATE_DEAD
;
737 /* makes sure we're not called twice from so_close */
738 so
->so_flags
|= SOF_PCBCLEARING
;
740 inpcb_gc_sched(inp
->inp_pcbinfo
, INPCB_TIMER_FAST
);
743 * See inp_join_group() for why we need to unlock
745 if (im6o
!= NULL
|| imo
!= NULL
) {
746 socket_unlock(so
, 0);
759 in6_sockaddr(in_port_t port
, struct in6_addr
*addr_p
)
761 struct sockaddr_in6
*sin6
;
763 MALLOC(sin6
, struct sockaddr_in6
*, sizeof(*sin6
), M_SONAME
, M_WAITOK
);
767 bzero(sin6
, sizeof(*sin6
));
768 sin6
->sin6_family
= AF_INET6
;
769 sin6
->sin6_len
= sizeof(*sin6
);
770 sin6
->sin6_port
= port
;
771 sin6
->sin6_addr
= *addr_p
;
773 /* would be good to use sa6_recoverscope(), except for locking */
774 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
775 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
777 sin6
->sin6_scope_id
= 0; /* XXX */
779 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
780 sin6
->sin6_addr
.s6_addr16
[1] = 0;
783 return (struct sockaddr
*)sin6
;
787 in6_sockaddr_s(in_port_t port
, struct in6_addr
*addr_p
,
788 struct sockaddr_in6
*sin6
)
790 bzero(sin6
, sizeof(*sin6
));
791 sin6
->sin6_family
= AF_INET6
;
792 sin6
->sin6_len
= sizeof(*sin6
);
793 sin6
->sin6_port
= port
;
794 sin6
->sin6_addr
= *addr_p
;
796 /* would be good to use sa6_recoverscope(), except for locking */
797 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
798 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
800 sin6
->sin6_scope_id
= 0; /* XXX */
802 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
803 sin6
->sin6_addr
.s6_addr16
[1] = 0;
808 * The calling convention of in6_getsockaddr() and in6_getpeeraddr() was
809 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
810 * in struct pr_usrreqs, so that protocols can just reference then directly
811 * without the need for a wrapper function.
814 in6_getsockaddr(struct socket
*so
, struct sockaddr
**nam
)
817 struct in6_addr addr
;
820 if ((inp
= sotoinpcb(so
)) == NULL
) {
824 port
= inp
->inp_lport
;
825 addr
= inp
->in6p_laddr
;
827 *nam
= in6_sockaddr(port
, &addr
);
835 in6_getsockaddr_s(struct socket
*so
, struct sockaddr_in6
*ss
)
838 struct in6_addr addr
;
842 bzero(ss
, sizeof(*ss
));
844 if ((inp
= sotoinpcb(so
)) == NULL
) {
848 port
= inp
->inp_lport
;
849 addr
= inp
->in6p_laddr
;
851 in6_sockaddr_s(port
, &addr
, ss
);
856 in6_getpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
859 struct in6_addr addr
;
862 if ((inp
= sotoinpcb(so
)) == NULL
) {
866 port
= inp
->inp_fport
;
867 addr
= inp
->in6p_faddr
;
869 *nam
= in6_sockaddr(port
, &addr
);
877 in6_mapped_sockaddr(struct socket
*so
, struct sockaddr
**nam
)
879 struct inpcb
*inp
= sotoinpcb(so
);
885 if (inp
->inp_vflag
& INP_IPV4
) {
886 error
= in_getsockaddr(so
, nam
);
888 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
891 /* scope issues will be handled in in6_getsockaddr(). */
892 error
= in6_getsockaddr(so
, nam
);
898 in6_mapped_peeraddr(struct socket
*so
, struct sockaddr
**nam
)
900 struct inpcb
*inp
= sotoinpcb(so
);
906 if (inp
->inp_vflag
& INP_IPV4
) {
907 error
= in_getpeeraddr(so
, nam
);
909 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
912 /* scope issues will be handled in in6_getpeeraddr(). */
913 error
= in6_getpeeraddr(so
, nam
);
919 * Pass some notification to all connections of a protocol
920 * associated with address dst. The local address and/or port numbers
921 * may be specified to limit the search. The "usual action" will be
922 * taken, depending on the ctlinput cmd. The caller must filter any
923 * cmds that are uninteresting (e.g., no error in the map).
924 * Call the protocol specific routine (if any) to report
925 * any errors for each matching socket.
928 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, u_int fport_arg
,
929 const struct sockaddr
*src
, u_int lport_arg
, int cmd
, void *cmdarg
,
930 void (*notify
)(struct inpcb
*, int))
932 struct inpcbhead
*head
= pcbinfo
->ipi_listhead
;
933 struct inpcb
*inp
, *ninp
;
934 struct sockaddr_in6 sa6_src
, *sa6_dst
;
935 uint16_t fport
= (uint16_t)fport_arg
, lport
= (uint16_t)lport_arg
;
939 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
) {
943 sa6_dst
= (struct sockaddr_in6
*)(void *)dst
;
944 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
)) {
949 * note that src can be NULL when we get notify by local fragmentation.
951 sa6_src
= (src
== NULL
) ?
952 sa6_any
: *(struct sockaddr_in6
*)(uintptr_t)(size_t)src
;
953 flowinfo
= sa6_src
.sin6_flowinfo
;
956 * Redirects go to all references to the destination,
957 * and use in6_rtchange to invalidate the route cache.
958 * Dead host indications: also use in6_rtchange to invalidate
959 * the cache, and deliver the error to all the sockets.
960 * Otherwise, if we have knowledge of the local port and address,
961 * deliver only to that socket.
963 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
966 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof(sa6_src
.sin6_addr
));
968 if (cmd
!= PRC_HOSTDEAD
) {
969 notify
= in6_rtchange
;
972 errno
= inet6ctlerrmap
[cmd
];
973 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
974 for (inp
= LIST_FIRST(head
); inp
!= NULL
; inp
= ninp
) {
975 ninp
= LIST_NEXT(inp
, inp_list
);
977 if (!(inp
->inp_vflag
& INP_IPV6
)) {
982 * If the error designates a new path MTU for a destination
983 * and the application (associated with this socket) wanted to
984 * know the value, notify. Note that we notify for all
985 * disconnected sockets if the corresponding application
986 * wanted. This is because some UDP applications keep sending
987 * sockets disconnected.
988 * XXX: should we avoid to notify the value to TCP sockets?
990 if (cmd
== PRC_MSGSIZE
&& cmdarg
!= NULL
) {
991 socket_lock(inp
->inp_socket
, 1);
992 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)(void *)dst
,
993 (u_int32_t
*)cmdarg
);
994 socket_unlock(inp
->inp_socket
, 1);
998 * Detect if we should notify the error. If no source and
999 * destination ports are specifed, but non-zero flowinfo and
1000 * local address match, notify the error. This is the case
1001 * when the error is delivered with an encrypted buffer
1002 * by ESP. Otherwise, just compare addresses and ports
1005 if (lport
== 0 && fport
== 0 && flowinfo
&&
1006 inp
->inp_socket
!= NULL
&&
1007 flowinfo
== (inp
->inp_flow
& IPV6_FLOWLABEL_MASK
) &&
1008 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
)) {
1010 } else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
1011 &sa6_dst
->sin6_addr
) || inp
->inp_socket
== NULL
||
1012 (lport
&& inp
->inp_lport
!= lport
) ||
1013 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
1014 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1015 &sa6_src
.sin6_addr
)) || (fport
&& inp
->inp_fport
!= fport
)) {
1021 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) ==
1025 socket_lock(inp
->inp_socket
, 1);
1026 (*notify
)(inp
, errno
);
1027 (void) in_pcb_checkstate(inp
, WNT_RELEASE
, 1);
1028 socket_unlock(inp
->inp_socket
, 1);
1031 lck_rw_done(pcbinfo
->ipi_lock
);
1035 * Lookup a PCB based on the local address and port.
1038 in6_pcblookup_local(struct inpcbinfo
*pcbinfo
, struct in6_addr
*laddr
,
1039 u_int lport_arg
, int wild_okay
)
1042 int matchwild
= 3, wildcard
;
1043 uint16_t lport
= (uint16_t)lport_arg
;
1044 struct inpcbporthead
*porthash
;
1045 struct inpcb
*match
= NULL
;
1046 struct inpcbport
*phd
;
1049 struct inpcbhead
*head
;
1051 * Look for an unconnected (wildcard foreign addr) PCB that
1052 * matches the local address and port we're looking for.
1054 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1055 pcbinfo
->ipi_hashmask
)];
1056 LIST_FOREACH(inp
, head
, inp_hash
) {
1057 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1060 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1061 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1062 inp
->inp_lport
== lport
) {
1075 * Best fit PCB lookup.
1077 * First see if this local port is in use by looking on the
1080 porthash
= &pcbinfo
->ipi_porthashbase
[INP_PCBPORTHASH(lport
,
1081 pcbinfo
->ipi_porthashmask
)];
1082 LIST_FOREACH(phd
, porthash
, phd_hash
) {
1083 if (phd
->phd_port
== lport
) {
1089 * Port is in use by one or more PCBs. Look for best
1092 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
1094 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1097 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
)) {
1100 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
1101 if (IN6_IS_ADDR_UNSPECIFIED(laddr
)) {
1103 } else if (!IN6_ARE_ADDR_EQUAL(
1104 &inp
->in6p_laddr
, laddr
)) {
1108 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
)) {
1112 if (wildcard
< matchwild
) {
1114 matchwild
= wildcard
;
1115 if (matchwild
== 0) {
1125 * Check for alternatives when higher level complains
1126 * about service problems. For now, invalidate cached
1127 * routing information. If the route was created dynamically
1128 * (by a redirect), time to try a default gateway again.
1131 in6_losing(struct inpcb
*in6p
)
1135 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
1137 if (rt
->rt_flags
& RTF_DYNAMIC
) {
1139 * Prevent another thread from modifying rt_key,
1140 * rt_gateway via rt_setgate() after the rt_lock
1141 * is dropped by marking the route as defunct.
1143 rt
->rt_flags
|= RTF_CONDEMNED
;
1145 (void) rtrequest(RTM_DELETE
, rt_key(rt
),
1146 rt
->rt_gateway
, rt_mask(rt
), rt
->rt_flags
, NULL
);
1151 * A new route can be allocated
1152 * the next time output is attempted.
1155 ROUTE_RELEASE(&in6p
->in6p_route
);
1159 * After a routing change, flush old routing
1160 * and allocate a (hopefully) better one.
1163 in6_rtchange(struct inpcb
*inp
, int errno
)
1165 #pragma unused(errno)
1167 * A new route can be allocated the next time
1168 * output is attempted.
1170 ROUTE_RELEASE(&inp
->in6p_route
);
1174 * Check if PCB exists hash list. Also returns uid and gid of socket
1177 in6_pcblookup_hash_exists(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1178 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1179 uid_t
*uid
, gid_t
*gid
, struct ifnet
*ifp
)
1181 struct inpcbhead
*head
;
1183 uint16_t fport
= (uint16_t)fport_arg
, lport
= (uint16_t)lport_arg
;
1189 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1192 * First look for an exact match.
1194 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1195 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1196 LIST_FOREACH(inp
, head
, inp_hash
) {
1197 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1201 if (inp_restricted_recv(inp
, ifp
)) {
1206 if (!necp_socket_is_allowed_to_recv_on_interface(inp
, ifp
)) {
1211 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1212 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1213 inp
->inp_fport
== fport
&&
1214 inp
->inp_lport
== lport
) {
1215 if ((found
= (inp
->inp_socket
!= NULL
))) {
1217 * Found. Check if pcb is still valid
1219 *uid
= kauth_cred_getuid(
1220 inp
->inp_socket
->so_cred
);
1221 *gid
= kauth_cred_getgid(
1222 inp
->inp_socket
->so_cred
);
1224 lck_rw_done(pcbinfo
->ipi_lock
);
1229 struct inpcb
*local_wild
= NULL
;
1231 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1232 pcbinfo
->ipi_hashmask
)];
1233 LIST_FOREACH(inp
, head
, inp_hash
) {
1234 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1238 if (inp_restricted_recv(inp
, ifp
)) {
1243 if (!necp_socket_is_allowed_to_recv_on_interface(inp
, ifp
)) {
1248 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1249 inp
->inp_lport
== lport
) {
1250 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1252 found
= (inp
->inp_socket
!= NULL
);
1254 *uid
= kauth_cred_getuid(
1255 inp
->inp_socket
->so_cred
);
1256 *gid
= kauth_cred_getgid(
1257 inp
->inp_socket
->so_cred
);
1259 lck_rw_done(pcbinfo
->ipi_lock
);
1261 } else if (IN6_IS_ADDR_UNSPECIFIED(
1262 &inp
->in6p_laddr
)) {
1268 if ((found
= (local_wild
->inp_socket
!= NULL
))) {
1269 *uid
= kauth_cred_getuid(
1270 local_wild
->inp_socket
->so_cred
);
1271 *gid
= kauth_cred_getgid(
1272 local_wild
->inp_socket
->so_cred
);
1274 lck_rw_done(pcbinfo
->ipi_lock
);
1282 lck_rw_done(pcbinfo
->ipi_lock
);
1287 * Lookup PCB in hash list.
1290 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1291 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1294 struct inpcbhead
*head
;
1296 uint16_t fport
= (uint16_t)fport_arg
, lport
= (uint16_t)lport_arg
;
1298 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1301 * First look for an exact match.
1303 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1304 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1305 LIST_FOREACH(inp
, head
, inp_hash
) {
1306 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1310 if (inp_restricted_recv(inp
, ifp
)) {
1315 if (!necp_socket_is_allowed_to_recv_on_interface(inp
, ifp
)) {
1320 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1321 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1322 inp
->inp_fport
== fport
&&
1323 inp
->inp_lport
== lport
) {
1325 * Found. Check if pcb is still valid
1327 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) !=
1329 lck_rw_done(pcbinfo
->ipi_lock
);
1332 /* it's there but dead, say it isn't found */
1333 lck_rw_done(pcbinfo
->ipi_lock
);
1339 struct inpcb
*local_wild
= NULL
;
1341 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1342 pcbinfo
->ipi_hashmask
)];
1343 LIST_FOREACH(inp
, head
, inp_hash
) {
1344 if (!(inp
->inp_vflag
& INP_IPV6
)) {
1348 if (inp_restricted_recv(inp
, ifp
)) {
1353 if (!necp_socket_is_allowed_to_recv_on_interface(inp
, ifp
)) {
1358 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1359 inp
->inp_lport
== lport
) {
1360 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1362 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
,
1363 0) != WNT_STOPUSING
) {
1364 lck_rw_done(pcbinfo
->ipi_lock
);
1367 /* dead; say it isn't found */
1368 lck_rw_done(pcbinfo
->ipi_lock
);
1371 } else if (IN6_IS_ADDR_UNSPECIFIED(
1372 &inp
->in6p_laddr
)) {
1377 if (local_wild
&& in_pcb_checkstate(local_wild
,
1378 WNT_ACQUIRE
, 0) != WNT_STOPUSING
) {
1379 lck_rw_done(pcbinfo
->ipi_lock
);
1382 lck_rw_done(pcbinfo
->ipi_lock
);
1390 lck_rw_done(pcbinfo
->ipi_lock
);
1395 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
1399 ip
= mtod(m
, struct ip6_hdr
*);
1400 bzero(sin6
, sizeof(*sin6
));
1401 sin6
->sin6_len
= sizeof(*sin6
);
1402 sin6
->sin6_family
= AF_INET6
;
1403 sin6
->sin6_addr
= ip
->ip6_src
;
1404 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
1405 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1406 if ((m
->m_pkthdr
.pkt_flags
& (PKTF_LOOP
| PKTF_IFAINFO
)) ==
1407 (PKTF_LOOP
| PKTF_IFAINFO
)) {
1408 sin6
->sin6_scope_id
= m
->m_pkthdr
.src_ifindex
;
1409 } else if (m
->m_pkthdr
.rcvif
!= NULL
) {
1410 sin6
->sin6_scope_id
= m
->m_pkthdr
.rcvif
->if_index
;
1416 * The following routines implement this scheme:
1418 * Callers of ip6_output() that intend to cache the route in the inpcb pass
1419 * a local copy of the struct route to ip6_output(). Using a local copy of
1420 * the cached route significantly simplifies things as IP no longer has to
1421 * worry about having exclusive access to the passed in struct route, since
1422 * it's defined in the caller's stack; in essence, this allows for a lock-
1423 * less operation when updating the struct route at the IP level and below,
1424 * whenever necessary. The scheme works as follows:
1426 * Prior to dropping the socket's lock and calling ip6_output(), the caller
1427 * copies the struct route from the inpcb into its stack, and adds a reference
1428 * to the cached route entry, if there was any. The socket's lock is then
1429 * dropped and ip6_output() is called with a pointer to the copy of struct
1430 * route defined on the stack (not to the one in the inpcb.)
1432 * Upon returning from ip6_output(), the caller then acquires the socket's
1433 * lock and synchronizes the cache; if there is no route cached in the inpcb,
1434 * it copies the local copy of struct route (which may or may not contain any
1435 * route) back into the cache; otherwise, if the inpcb has a route cached in
1436 * it, the one in the local copy will be freed, if there's any. Trashing the
1437 * cached route in the inpcb can be avoided because ip6_output() is single-
1438 * threaded per-PCB (i.e. multiple transmits on a PCB are always serialized
1439 * by the socket/transport layer.)
1442 in6p_route_copyout(struct inpcb
*inp
, struct route_in6
*dst
)
1444 struct route_in6
*src
= &inp
->in6p_route
;
1446 socket_lock_assert_owned(inp
->inp_socket
);
1448 /* Minor sanity check */
1449 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
) {
1450 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1453 route_copyout((struct route
*)dst
, (struct route
*)src
, sizeof(*dst
));
1457 in6p_route_copyin(struct inpcb
*inp
, struct route_in6
*src
)
1459 struct route_in6
*dst
= &inp
->in6p_route
;
1461 socket_lock_assert_owned(inp
->inp_socket
);
1463 /* Minor sanity check */
1464 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
) {
1465 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1468 route_copyin((struct route
*)src
, (struct route
*)dst
, sizeof(*src
));