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
;
168 if (so
->so_usecount
== 0) {
169 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
;
197 #endif /* !CONFIG_EMBEDDED */
199 if (!in6_ifaddrs
) /* XXX broken! */
200 return (EADDRNOTAVAIL
);
201 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
))
203 if (!(so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)))
206 socket_unlock(so
, 0); /* keep reference */
207 lck_rw_lock_exclusive(pcbinfo
->ipi_lock
);
209 bzero(&sin6
, sizeof (sin6
));
211 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
)) {
212 lck_rw_done(pcbinfo
->ipi_lock
);
219 if (nam
->sa_family
!= AF_INET6
) {
220 lck_rw_done(pcbinfo
->ipi_lock
);
222 return (EAFNOSUPPORT
);
224 lport
= SIN6(nam
)->sin6_port
;
226 *(&sin6
) = *SIN6(nam
);
228 /* KAME hack: embed scopeid */
229 if (in6_embedscope(&sin6
.sin6_addr
, &sin6
, inp
, NULL
,
231 lck_rw_done(pcbinfo
->ipi_lock
);
236 /* Sanitize local copy for address searches */
237 sin6
.sin6_flowinfo
= 0;
238 sin6
.sin6_scope_id
= 0;
241 if (IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
)) {
243 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
244 * allow compepte duplication of binding if
245 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
246 * and a multicast address is bound on both
247 * new and duplicated sockets.
249 if (so
->so_options
& SO_REUSEADDR
)
250 reuseport
= SO_REUSEADDR
|SO_REUSEPORT
;
251 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
254 ifa
= ifa_ifwithaddr(SA(&sin6
));
256 lck_rw_done(pcbinfo
->ipi_lock
);
258 return (EADDRNOTAVAIL
);
261 * XXX: bind to an anycast address might
262 * accidentally cause sending a packet with
263 * anycast source address. We should allow
264 * to bind to a deprecated address, since
265 * the application dare to use it.
268 if (((struct in6_ifaddr
*)ifa
)->ia6_flags
&
269 (IN6_IFF_ANYCAST
|IN6_IFF_NOTREADY
|
273 lck_rw_done(pcbinfo
->ipi_lock
);
275 return (EADDRNOTAVAIL
);
278 * Opportunistically determine the outbound
279 * interface that may be used; this may not
280 * hold true if we end up using a route
281 * going over a different interface, e.g.
282 * when sending to a local address. This
283 * will get updated again after sending.
285 outif
= ifa
->ifa_ifp
;
296 if (ntohs(lport
) < IPV6PORT_RESERVED
) {
297 cred
= kauth_cred_proc_ref(p
);
298 error
= priv_check_cred(cred
,
299 PRIV_NETINET_RESERVEDPORT
, 0);
300 kauth_cred_unref(&cred
);
302 lck_rw_done(pcbinfo
->ipi_lock
);
307 #endif /* !CONFIG_EMBEDDED */
308 if (!IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
) &&
309 (u
= kauth_cred_getuid(so
->so_cred
)) != 0) {
310 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
311 &sin6
.sin6_addr
, lport
,
313 if (t
!= NULL
&& (!IN6_IS_ADDR_UNSPECIFIED(
315 !IN6_IS_ADDR_UNSPECIFIED(&t
->in6p_laddr
) ||
316 !(t
->inp_socket
->so_options
&
317 SO_REUSEPORT
)) && (u
!= kauth_cred_getuid(
318 t
->inp_socket
->so_cred
)) &&
319 !(t
->inp_socket
->so_flags
&
320 SOF_REUSESHAREUID
)) {
321 lck_rw_done(pcbinfo
->ipi_lock
);
325 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
326 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
327 struct sockaddr_in sin
;
329 in6_sin6_2_sin(&sin
, &sin6
);
330 t
= in_pcblookup_local_and_cleanup(
331 pcbinfo
, sin
.sin_addr
, lport
,
334 !(t
->inp_socket
->so_options
&
336 (kauth_cred_getuid(so
->so_cred
) !=
337 kauth_cred_getuid(t
->inp_socket
->
338 so_cred
)) && (t
->inp_laddr
.s_addr
!=
339 INADDR_ANY
|| SOCK_DOM(so
) ==
340 SOCK_DOM(t
->inp_socket
))) {
341 lck_rw_done(pcbinfo
->ipi_lock
);
347 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
348 &sin6
.sin6_addr
, lport
, wild
);
350 (reuseport
& t
->inp_socket
->so_options
) == 0) {
351 lck_rw_done(pcbinfo
->ipi_lock
);
355 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
356 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
357 struct sockaddr_in sin
;
359 in6_sin6_2_sin(&sin
, &sin6
);
360 t
= in_pcblookup_local_and_cleanup(pcbinfo
,
361 sin
.sin_addr
, lport
, wild
);
362 if (t
!= NULL
&& (reuseport
&
363 t
->inp_socket
->so_options
) == 0 &&
364 (t
->inp_laddr
.s_addr
!= INADDR_ANY
||
365 SOCK_DOM(so
) == SOCK_DOM(t
->inp_socket
))) {
366 lck_rw_done(pcbinfo
->ipi_lock
);
376 * We unlocked socket's protocol lock for a long time.
377 * The socket might have been dropped/defuncted.
378 * Checking if world has changed since.
380 if (inp
->inp_state
== INPCB_STATE_DEAD
) {
381 lck_rw_done(pcbinfo
->ipi_lock
);
382 return (ECONNABORTED
);
385 /* check if the socket got bound when the lock was released */
386 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
387 lck_rw_done(pcbinfo
->ipi_lock
);
391 if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
392 inp
->in6p_laddr
= sin6
.sin6_addr
;
393 inp
->in6p_last_outifp
= outif
;
398 if ((e
= in6_pcbsetport(&inp
->in6p_laddr
, inp
, p
, 1)) != 0) {
399 /* Undo any address bind from above. */
400 inp
->in6p_laddr
= in6addr_any
;
401 inp
->in6p_last_outifp
= NULL
;
402 lck_rw_done(pcbinfo
->ipi_lock
);
406 inp
->inp_lport
= lport
;
407 if (in_pcbinshash(inp
, 1) != 0) {
408 inp
->in6p_laddr
= in6addr_any
;
410 inp
->in6p_last_outifp
= NULL
;
411 lck_rw_done(pcbinfo
->ipi_lock
);
415 lck_rw_done(pcbinfo
->ipi_lock
);
416 sflt_notify(so
, sock_evt_bound
, NULL
);
421 * Transform old in6_pcbconnect() into an inner subroutine for new
422 * in6_pcbconnect(); do some validity-checking on the remote address
423 * (in "nam") and then determine local host address (i.e., which
424 * interface) to use to access that remote host.
426 * This routine may alter the caller-supplied remote address "nam".
428 * This routine might return an ifp with a reference held if the caller
429 * provides a non-NULL outif, even in the error case. The caller is
430 * responsible for releasing its reference.
433 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
434 struct in6_addr
*plocal_addr6
, struct ifnet
**outif
)
436 struct in6_addr
*addr6
= NULL
;
437 struct in6_addr src_storage
;
439 unsigned int ifscope
;
443 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
))
445 if (SIN6(nam
)->sin6_family
!= AF_INET6
)
446 return (EAFNOSUPPORT
);
447 if (SIN6(nam
)->sin6_port
== 0)
448 return (EADDRNOTAVAIL
);
450 /* KAME hack: embed scopeid */
451 if (in6_embedscope(&SIN6(nam
)->sin6_addr
, SIN6(nam
), inp
, NULL
, NULL
) != 0)
456 * If the destination address is UNSPECIFIED addr,
457 * use the loopback addr, e.g ::1.
459 if (IN6_IS_ADDR_UNSPECIFIED(&SIN6(nam
)->sin6_addr
))
460 SIN6(nam
)->sin6_addr
= in6addr_loopback
;
463 ifscope
= (inp
->inp_flags
& INP_BOUND_IF
) ?
464 inp
->inp_boundifp
->if_index
: IFSCOPE_NONE
;
467 * XXX: in6_selectsrc might replace the bound local address
468 * with the address specified by setsockopt(IPV6_PKTINFO).
469 * Is it the intended behavior?
471 * in6_selectsrc() might return outif with its reference held
472 * even in the error case; caller always needs to release it
475 addr6
= in6_selectsrc(SIN6(nam
), inp
->in6p_outputopts
, inp
,
476 &inp
->in6p_route
, outif
, &src_storage
, ifscope
, &error
);
479 struct rtentry
*rt
= inp
->in6p_route
.ro_rt
;
481 * If in6_selectsrc() returns a route, it should be one
482 * which points to the same ifp as outif. Just in case
483 * it isn't, use the one from the route for consistency.
484 * Otherwise if there is no route, leave outif alone as
485 * it could still be useful to the caller.
487 if (rt
!= NULL
&& rt
->rt_ifp
!= *outif
) {
488 ifnet_reference(rt
->rt_ifp
); /* for caller */
490 ifnet_release(*outif
);
496 if (outif
!= NULL
&& (*outif
) != NULL
&&
497 inp_restricted_send(inp
, *outif
)) {
498 soevent(inp
->inp_socket
,
499 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
500 error
= EHOSTUNREACH
;
503 error
= EADDRNOTAVAIL
;
507 *plocal_addr6
= *addr6
;
509 * Don't do pcblookup call here; return interface in
510 * plocal_addr6 and exit to caller, that will do the lookup.
517 * Connect from a socket to a specified address.
518 * Both address and port must be specified in argument sin.
519 * If don't have a local address for this socket yet,
523 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct proc
*p
)
525 struct in6_addr addr6
;
526 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)(void *)nam
;
529 struct ifnet
*outif
= NULL
;
530 struct socket
*so
= inp
->inp_socket
;
532 if (so
->so_proto
->pr_protocol
== IPPROTO_UDP
&&
533 sin6
->sin6_port
== htons(53) && !(so
->so_flags1
& SOF1_DNS_COUNTED
)) {
534 so
->so_flags1
|= SOF1_DNS_COUNTED
;
535 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet_dgram_dns
);
539 * Call inner routine, to assign local interface address.
540 * in6_pcbladdr() may automatically fill in sin6_scope_id.
542 * in6_pcbladdr() might return an ifp with its reference held
543 * even in the error case, so make sure that it's released
544 * whenever it's non-NULL.
546 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, &outif
)) != 0) {
547 if (outif
!= NULL
&& inp_restricted_send(inp
, outif
))
549 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
552 socket_unlock(so
, 0);
553 pcb
= in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
554 sin6
->sin6_port
, IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
) ?
555 &addr6
: &inp
->in6p_laddr
, inp
->inp_lport
, 0, NULL
);
558 in_pcb_checkstate(pcb
, WNT_RELEASE
, pcb
== inp
? 1 : 0);
562 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
563 if (inp
->inp_lport
== 0) {
564 error
= in6_pcbbind(inp
, NULL
, p
);
568 inp
->in6p_laddr
= addr6
;
569 inp
->in6p_last_outifp
= outif
; /* no reference needed */
570 inp
->in6p_flags
|= INP_IN6ADDR_ANY
;
572 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
573 /* lock inversion issue, mostly with udp multicast packets */
574 socket_unlock(so
, 0);
575 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
578 inp
->in6p_faddr
= sin6
->sin6_addr
;
579 inp
->inp_fport
= sin6
->sin6_port
;
580 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
581 nstat_pcb_invalidate_cache(inp
);
583 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
587 ifnet_release(outif
);
593 in6_pcbdisconnect(struct inpcb
*inp
)
595 struct socket
*so
= inp
->inp_socket
;
597 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
598 /* lock inversion issue, mostly with udp multicast packets */
599 socket_unlock(so
, 0);
600 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
603 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
604 nstat_pcb_cache(inp
);
605 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof (inp
->in6p_faddr
));
607 /* clear flowinfo - RFC 6437 */
608 inp
->inp_flow
&= ~IPV6_FLOWLABEL_MASK
;
610 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
612 * A multipath subflow socket would have its SS_NOFDREF set by default,
613 * so check for SOF_MP_SUBFLOW socket flag before detaching the PCB;
614 * when the socket is closed for real, SOF_MP_SUBFLOW would be cleared.
616 if (!(so
->so_flags
& SOF_MP_SUBFLOW
) && (so
->so_state
& SS_NOFDREF
))
621 in6_pcbdetach(struct inpcb
*inp
)
623 struct socket
*so
= inp
->inp_socket
;
625 if (so
->so_pcb
== NULL
) {
626 /* PCB has been disposed */
627 panic("%s: inp=%p so=%p proto=%d so_pcb is null!\n", __func__
,
628 inp
, so
, SOCK_PROTO(so
));
633 if (inp
->in6p_sp
!= NULL
) {
634 (void) ipsec6_delete_pcbpolicy(inp
);
638 if (inp
->inp_stat
!= NULL
&& SOCK_PROTO(so
) == IPPROTO_UDP
) {
639 if (inp
->inp_stat
->rxpackets
== 0 && inp
->inp_stat
->txpackets
== 0) {
640 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet6_dgram_no_data
);
645 * Let NetworkStatistics know this PCB is going away
646 * before we detach it.
649 (SOCK_PROTO(so
) == IPPROTO_TCP
|| SOCK_PROTO(so
) == IPPROTO_UDP
))
650 nstat_pcb_detach(inp
);
651 /* mark socket state as dead */
652 if (in_pcb_checkstate(inp
, WNT_STOPUSING
, 1) != WNT_STOPUSING
) {
653 panic("%s: so=%p proto=%d couldn't set to STOPUSING\n",
654 __func__
, so
, SOCK_PROTO(so
));
658 if (!(so
->so_flags
& SOF_PCBCLEARING
)) {
659 struct ip_moptions
*imo
;
660 struct ip6_moptions
*im6o
;
663 if (inp
->in6p_options
!= NULL
) {
664 m_freem(inp
->in6p_options
);
665 inp
->in6p_options
= NULL
;
667 ip6_freepcbopts(inp
->in6p_outputopts
);
668 ROUTE_RELEASE(&inp
->in6p_route
);
669 /* free IPv4 related resources in case of mapped addr */
670 if (inp
->inp_options
!= NULL
) {
671 (void) m_free(inp
->inp_options
);
672 inp
->inp_options
= NULL
;
674 im6o
= inp
->in6p_moptions
;
675 inp
->in6p_moptions
= NULL
;
677 imo
= inp
->inp_moptions
;
678 inp
->inp_moptions
= NULL
;
680 sofreelastref(so
, 0);
681 inp
->inp_state
= INPCB_STATE_DEAD
;
682 /* makes sure we're not called twice from so_close */
683 so
->so_flags
|= SOF_PCBCLEARING
;
685 inpcb_gc_sched(inp
->inp_pcbinfo
, INPCB_TIMER_FAST
);
688 * See inp_join_group() for why we need to unlock
690 if (im6o
!= NULL
|| imo
!= NULL
) {
691 socket_unlock(so
, 0);
702 in6_sockaddr(in_port_t port
, struct in6_addr
*addr_p
)
704 struct sockaddr_in6
*sin6
;
706 MALLOC(sin6
, struct sockaddr_in6
*, sizeof (*sin6
), M_SONAME
, M_WAITOK
);
709 bzero(sin6
, sizeof (*sin6
));
710 sin6
->sin6_family
= AF_INET6
;
711 sin6
->sin6_len
= sizeof (*sin6
);
712 sin6
->sin6_port
= port
;
713 sin6
->sin6_addr
= *addr_p
;
715 /* would be good to use sa6_recoverscope(), except for locking */
716 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
717 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
719 sin6
->sin6_scope_id
= 0; /* XXX */
720 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
721 sin6
->sin6_addr
.s6_addr16
[1] = 0;
723 return ((struct sockaddr
*)sin6
);
727 in6_sockaddr_s(in_port_t port
, struct in6_addr
*addr_p
,
728 struct sockaddr_in6
*sin6
)
730 bzero(sin6
, sizeof (*sin6
));
731 sin6
->sin6_family
= AF_INET6
;
732 sin6
->sin6_len
= sizeof (*sin6
);
733 sin6
->sin6_port
= port
;
734 sin6
->sin6_addr
= *addr_p
;
736 /* would be good to use sa6_recoverscope(), except for locking */
737 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
738 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
740 sin6
->sin6_scope_id
= 0; /* XXX */
741 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
742 sin6
->sin6_addr
.s6_addr16
[1] = 0;
746 * The calling convention of in6_getsockaddr() and in6_getpeeraddr() was
747 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
748 * in struct pr_usrreqs, so that protocols can just reference then directly
749 * without the need for a wrapper function.
752 in6_getsockaddr(struct socket
*so
, struct sockaddr
**nam
)
755 struct in6_addr addr
;
758 if ((inp
= sotoinpcb(so
)) == NULL
)
761 port
= inp
->inp_lport
;
762 addr
= inp
->in6p_laddr
;
764 *nam
= in6_sockaddr(port
, &addr
);
771 in6_getsockaddr_s(struct socket
*so
, struct sockaddr_in6
*ss
)
774 struct in6_addr addr
;
778 bzero(ss
, sizeof (*ss
));
780 if ((inp
= sotoinpcb(so
)) == NULL
)
783 port
= inp
->inp_lport
;
784 addr
= inp
->in6p_laddr
;
786 in6_sockaddr_s(port
, &addr
, ss
);
791 in6_getpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
794 struct in6_addr addr
;
797 if ((inp
= sotoinpcb(so
)) == NULL
)
800 port
= inp
->inp_fport
;
801 addr
= inp
->in6p_faddr
;
803 *nam
= in6_sockaddr(port
, &addr
);
810 in6_mapped_sockaddr(struct socket
*so
, struct sockaddr
**nam
)
812 struct inpcb
*inp
= sotoinpcb(so
);
817 if (inp
->inp_vflag
& INP_IPV4
) {
818 error
= in_getsockaddr(so
, nam
);
820 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
822 /* scope issues will be handled in in6_getsockaddr(). */
823 error
= in6_getsockaddr(so
, nam
);
829 in6_mapped_peeraddr(struct socket
*so
, struct sockaddr
**nam
)
831 struct inpcb
*inp
= sotoinpcb(so
);
836 if (inp
->inp_vflag
& INP_IPV4
) {
837 error
= in_getpeeraddr(so
, nam
);
839 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
841 /* scope issues will be handled in in6_getpeeraddr(). */
842 error
= in6_getpeeraddr(so
, nam
);
848 * Pass some notification to all connections of a protocol
849 * associated with address dst. The local address and/or port numbers
850 * may be specified to limit the search. The "usual action" will be
851 * taken, depending on the ctlinput cmd. The caller must filter any
852 * cmds that are uninteresting (e.g., no error in the map).
853 * Call the protocol specific routine (if any) to report
854 * any errors for each matching socket.
857 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, u_int fport_arg
,
858 const struct sockaddr
*src
, u_int lport_arg
, int cmd
, void *cmdarg
,
859 void (*notify
)(struct inpcb
*, int))
861 struct inpcbhead
*head
= pcbinfo
->ipi_listhead
;
862 struct inpcb
*inp
, *ninp
;
863 struct sockaddr_in6 sa6_src
, *sa6_dst
;
864 u_short fport
= fport_arg
, lport
= lport_arg
;
868 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
)
871 sa6_dst
= (struct sockaddr_in6
*)(void *)dst
;
872 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
))
876 * note that src can be NULL when we get notify by local fragmentation.
878 sa6_src
= (src
== NULL
) ?
879 sa6_any
: *(struct sockaddr_in6
*)(uintptr_t)(size_t)src
;
880 flowinfo
= sa6_src
.sin6_flowinfo
;
883 * Redirects go to all references to the destination,
884 * and use in6_rtchange to invalidate the route cache.
885 * Dead host indications: also use in6_rtchange to invalidate
886 * the cache, and deliver the error to all the sockets.
887 * Otherwise, if we have knowledge of the local port and address,
888 * deliver only to that socket.
890 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
893 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof (sa6_src
.sin6_addr
));
895 if (cmd
!= PRC_HOSTDEAD
)
896 notify
= in6_rtchange
;
898 errno
= inet6ctlerrmap
[cmd
];
899 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
900 for (inp
= LIST_FIRST(head
); inp
!= NULL
; inp
= ninp
) {
901 ninp
= LIST_NEXT(inp
, inp_list
);
903 if (!(inp
->inp_vflag
& INP_IPV6
))
907 * If the error designates a new path MTU for a destination
908 * and the application (associated with this socket) wanted to
909 * know the value, notify. Note that we notify for all
910 * disconnected sockets if the corresponding application
911 * wanted. This is because some UDP applications keep sending
912 * sockets disconnected.
913 * XXX: should we avoid to notify the value to TCP sockets?
915 if (cmd
== PRC_MSGSIZE
)
916 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)(void *)dst
,
917 (u_int32_t
*)cmdarg
);
920 * Detect if we should notify the error. If no source and
921 * destination ports are specifed, but non-zero flowinfo and
922 * local address match, notify the error. This is the case
923 * when the error is delivered with an encrypted buffer
924 * by ESP. Otherwise, just compare addresses and ports
927 if (lport
== 0 && fport
== 0 && flowinfo
&&
928 inp
->inp_socket
!= NULL
&&
929 flowinfo
== (inp
->inp_flow
& IPV6_FLOWLABEL_MASK
) &&
930 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
))
932 else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
933 &sa6_dst
->sin6_addr
) || inp
->inp_socket
== NULL
||
934 (lport
&& inp
->inp_lport
!= lport
) ||
935 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
936 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
937 &sa6_src
.sin6_addr
)) || (fport
&& inp
->inp_fport
!= fport
))
942 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) ==
945 socket_lock(inp
->inp_socket
, 1);
946 (*notify
)(inp
, errno
);
947 (void) in_pcb_checkstate(inp
, WNT_RELEASE
, 1);
948 socket_unlock(inp
->inp_socket
, 1);
951 lck_rw_done(pcbinfo
->ipi_lock
);
955 * Lookup a PCB based on the local address and port.
958 in6_pcblookup_local(struct inpcbinfo
*pcbinfo
, struct in6_addr
*laddr
,
959 u_int lport_arg
, int wild_okay
)
962 int matchwild
= 3, wildcard
;
963 u_short lport
= lport_arg
;
964 struct inpcbporthead
*porthash
;
965 struct inpcb
*match
= NULL
;
966 struct inpcbport
*phd
;
969 struct inpcbhead
*head
;
971 * Look for an unconnected (wildcard foreign addr) PCB that
972 * matches the local address and port we're looking for.
974 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
975 pcbinfo
->ipi_hashmask
)];
976 LIST_FOREACH(inp
, head
, inp_hash
) {
977 if (!(inp
->inp_vflag
& INP_IPV6
))
979 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
980 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
981 inp
->inp_lport
== lport
) {
994 * Best fit PCB lookup.
996 * First see if this local port is in use by looking on the
999 porthash
= &pcbinfo
->ipi_porthashbase
[INP_PCBPORTHASH(lport
,
1000 pcbinfo
->ipi_porthashmask
)];
1001 LIST_FOREACH(phd
, porthash
, phd_hash
) {
1002 if (phd
->phd_port
== lport
)
1007 * Port is in use by one or more PCBs. Look for best
1010 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
1012 if (!(inp
->inp_vflag
& INP_IPV6
))
1014 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
1016 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
1017 if (IN6_IS_ADDR_UNSPECIFIED(laddr
))
1019 else if (!IN6_ARE_ADDR_EQUAL(
1020 &inp
->in6p_laddr
, laddr
))
1023 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
))
1026 if (wildcard
< matchwild
) {
1028 matchwild
= wildcard
;
1029 if (matchwild
== 0) {
1039 * Check for alternatives when higher level complains
1040 * about service problems. For now, invalidate cached
1041 * routing information. If the route was created dynamically
1042 * (by a redirect), time to try a default gateway again.
1045 in6_losing(struct inpcb
*in6p
)
1049 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
1051 if (rt
->rt_flags
& RTF_DYNAMIC
) {
1053 * Prevent another thread from modifying rt_key,
1054 * rt_gateway via rt_setgate() after the rt_lock
1055 * is dropped by marking the route as defunct.
1057 rt
->rt_flags
|= RTF_CONDEMNED
;
1059 (void) rtrequest(RTM_DELETE
, rt_key(rt
),
1060 rt
->rt_gateway
, rt_mask(rt
), rt
->rt_flags
, NULL
);
1065 * A new route can be allocated
1066 * the next time output is attempted.
1069 ROUTE_RELEASE(&in6p
->in6p_route
);
1073 * After a routing change, flush old routing
1074 * and allocate a (hopefully) better one.
1077 in6_rtchange(struct inpcb
*inp
, int errno
)
1079 #pragma unused(errno)
1081 * A new route can be allocated the next time
1082 * output is attempted.
1084 ROUTE_RELEASE(&inp
->in6p_route
);
1088 * Check if PCB exists hash list. Also returns uid and gid of socket
1091 in6_pcblookup_hash_exists(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1092 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1093 uid_t
*uid
, gid_t
*gid
, struct ifnet
*ifp
)
1095 struct inpcbhead
*head
;
1097 u_short fport
= fport_arg
, lport
= lport_arg
;
1103 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1106 * First look for an exact match.
1108 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1109 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1110 LIST_FOREACH(inp
, head
, inp_hash
) {
1111 if (!(inp
->inp_vflag
& INP_IPV6
))
1114 if (inp_restricted_recv(inp
, ifp
))
1117 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1118 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1119 inp
->inp_fport
== fport
&&
1120 inp
->inp_lport
== lport
) {
1121 if ((found
= (inp
->inp_socket
!= NULL
))) {
1123 * Found. Check if pcb is still valid
1125 *uid
= kauth_cred_getuid(
1126 inp
->inp_socket
->so_cred
);
1127 *gid
= kauth_cred_getgid(
1128 inp
->inp_socket
->so_cred
);
1130 lck_rw_done(pcbinfo
->ipi_lock
);
1135 struct inpcb
*local_wild
= NULL
;
1137 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1138 pcbinfo
->ipi_hashmask
)];
1139 LIST_FOREACH(inp
, head
, inp_hash
) {
1140 if (!(inp
->inp_vflag
& INP_IPV6
))
1143 if (inp_restricted_recv(inp
, ifp
))
1146 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1147 inp
->inp_lport
== lport
) {
1148 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1150 found
= (inp
->inp_socket
!= NULL
);
1152 *uid
= kauth_cred_getuid(
1153 inp
->inp_socket
->so_cred
);
1154 *gid
= kauth_cred_getgid(
1155 inp
->inp_socket
->so_cred
);
1157 lck_rw_done(pcbinfo
->ipi_lock
);
1159 } else if (IN6_IS_ADDR_UNSPECIFIED(
1160 &inp
->in6p_laddr
)) {
1166 if ((found
= (local_wild
->inp_socket
!= NULL
))) {
1167 *uid
= kauth_cred_getuid(
1168 local_wild
->inp_socket
->so_cred
);
1169 *gid
= kauth_cred_getgid(
1170 local_wild
->inp_socket
->so_cred
);
1172 lck_rw_done(pcbinfo
->ipi_lock
);
1180 lck_rw_done(pcbinfo
->ipi_lock
);
1185 * Lookup PCB in hash list.
1188 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1189 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1192 struct inpcbhead
*head
;
1194 u_short fport
= fport_arg
, lport
= lport_arg
;
1196 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1199 * First look for an exact match.
1201 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1202 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1203 LIST_FOREACH(inp
, head
, inp_hash
) {
1204 if (!(inp
->inp_vflag
& INP_IPV6
))
1207 if (inp_restricted_recv(inp
, ifp
))
1210 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1211 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1212 inp
->inp_fport
== fport
&&
1213 inp
->inp_lport
== lport
) {
1215 * Found. Check if pcb is still valid
1217 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) !=
1219 lck_rw_done(pcbinfo
->ipi_lock
);
1222 /* it's there but dead, say it isn't found */
1223 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
))
1237 if (inp_restricted_recv(inp
, ifp
))
1240 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1241 inp
->inp_lport
== lport
) {
1242 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1244 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
,
1245 0) != WNT_STOPUSING
) {
1246 lck_rw_done(pcbinfo
->ipi_lock
);
1249 /* dead; say it isn't found */
1250 lck_rw_done(pcbinfo
->ipi_lock
);
1253 } else if (IN6_IS_ADDR_UNSPECIFIED(
1254 &inp
->in6p_laddr
)) {
1259 if (local_wild
&& in_pcb_checkstate(local_wild
,
1260 WNT_ACQUIRE
, 0) != WNT_STOPUSING
) {
1261 lck_rw_done(pcbinfo
->ipi_lock
);
1262 return (local_wild
);
1264 lck_rw_done(pcbinfo
->ipi_lock
);
1272 lck_rw_done(pcbinfo
->ipi_lock
);
1277 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
1281 ip
= mtod(m
, struct ip6_hdr
*);
1282 bzero(sin6
, sizeof (*sin6
));
1283 sin6
->sin6_len
= sizeof (*sin6
);
1284 sin6
->sin6_family
= AF_INET6
;
1285 sin6
->sin6_addr
= ip
->ip6_src
;
1286 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
1287 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1288 if ((m
->m_pkthdr
.pkt_flags
& (PKTF_LOOP
|PKTF_IFAINFO
)) ==
1289 (PKTF_LOOP
|PKTF_IFAINFO
))
1290 sin6
->sin6_scope_id
= m
->m_pkthdr
.src_ifindex
;
1291 else if (m
->m_pkthdr
.rcvif
!= NULL
)
1292 sin6
->sin6_scope_id
= m
->m_pkthdr
.rcvif
->if_index
;
1297 * The following routines implement this scheme:
1299 * Callers of ip6_output() that intend to cache the route in the inpcb pass
1300 * a local copy of the struct route to ip6_output(). Using a local copy of
1301 * the cached route significantly simplifies things as IP no longer has to
1302 * worry about having exclusive access to the passed in struct route, since
1303 * it's defined in the caller's stack; in essence, this allows for a lock-
1304 * less operation when updating the struct route at the IP level and below,
1305 * whenever necessary. The scheme works as follows:
1307 * Prior to dropping the socket's lock and calling ip6_output(), the caller
1308 * copies the struct route from the inpcb into its stack, and adds a reference
1309 * to the cached route entry, if there was any. The socket's lock is then
1310 * dropped and ip6_output() is called with a pointer to the copy of struct
1311 * route defined on the stack (not to the one in the inpcb.)
1313 * Upon returning from ip6_output(), the caller then acquires the socket's
1314 * lock and synchronizes the cache; if there is no route cached in the inpcb,
1315 * it copies the local copy of struct route (which may or may not contain any
1316 * route) back into the cache; otherwise, if the inpcb has a route cached in
1317 * it, the one in the local copy will be freed, if there's any. Trashing the
1318 * cached route in the inpcb can be avoided because ip6_output() is single-
1319 * threaded per-PCB (i.e. multiple transmits on a PCB are always serialized
1320 * by the socket/transport layer.)
1323 in6p_route_copyout(struct inpcb
*inp
, struct route_in6
*dst
)
1325 struct route_in6
*src
= &inp
->in6p_route
;
1327 socket_lock_assert_owned(inp
->inp_socket
);
1329 /* Minor sanity check */
1330 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
)
1331 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1333 route_copyout((struct route
*)dst
, (struct route
*)src
, sizeof (*dst
));
1337 in6p_route_copyin(struct inpcb
*inp
, struct route_in6
*src
)
1339 struct route_in6
*dst
= &inp
->in6p_route
;
1341 socket_lock_assert_owned(inp
->inp_socket
);
1343 /* Minor sanity check */
1344 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
)
1345 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1347 route_copyin((struct route
*)src
, (struct route
*)dst
, sizeof (*src
));