2 * Copyright (c) 2003-2017 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 (!(so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)))
204 socket_unlock(so
, 0); /* keep reference */
205 lck_rw_lock_exclusive(pcbinfo
->ipi_lock
);
206 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
207 /* another thread completed the bind */
208 lck_rw_done(pcbinfo
->ipi_lock
);
213 bzero(&sin6
, sizeof (sin6
));
215 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
)) {
216 lck_rw_done(pcbinfo
->ipi_lock
);
223 if (nam
->sa_family
!= AF_INET6
) {
224 lck_rw_done(pcbinfo
->ipi_lock
);
226 return (EAFNOSUPPORT
);
228 lport
= SIN6(nam
)->sin6_port
;
230 *(&sin6
) = *SIN6(nam
);
232 /* KAME hack: embed scopeid */
233 if (in6_embedscope(&sin6
.sin6_addr
, &sin6
, inp
, NULL
,
235 lck_rw_done(pcbinfo
->ipi_lock
);
240 /* Sanitize local copy for address searches */
241 sin6
.sin6_flowinfo
= 0;
242 sin6
.sin6_scope_id
= 0;
245 if (IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
)) {
247 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
248 * allow compepte duplication of binding if
249 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
250 * and a multicast address is bound on both
251 * new and duplicated sockets.
253 if (so
->so_options
& SO_REUSEADDR
)
254 reuseport
= SO_REUSEADDR
|SO_REUSEPORT
;
255 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
258 ifa
= ifa_ifwithaddr(SA(&sin6
));
260 lck_rw_done(pcbinfo
->ipi_lock
);
262 return (EADDRNOTAVAIL
);
265 * XXX: bind to an anycast address might
266 * accidentally cause sending a packet with
267 * anycast source address. We should allow
268 * to bind to a deprecated address, since
269 * the application dare to use it.
272 if (((struct in6_ifaddr
*)ifa
)->ia6_flags
&
273 (IN6_IFF_ANYCAST
| IN6_IFF_NOTREADY
|
274 IN6_IFF_DETACHED
| IN6_IFF_CLAT46
)) {
277 lck_rw_done(pcbinfo
->ipi_lock
);
279 return (EADDRNOTAVAIL
);
282 * Opportunistically determine the outbound
283 * interface that may be used; this may not
284 * hold true if we end up using a route
285 * going over a different interface, e.g.
286 * when sending to a local address. This
287 * will get updated again after sending.
289 outif
= ifa
->ifa_ifp
;
299 if (ntohs(lport
) < IPV6PORT_RESERVED
&&
300 !IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
301 cred
= kauth_cred_proc_ref(p
);
302 error
= priv_check_cred(cred
,
303 PRIV_NETINET_RESERVEDPORT
, 0);
304 kauth_cred_unref(&cred
);
306 lck_rw_done(pcbinfo
->ipi_lock
);
311 #endif /* !CONFIG_EMBEDDED */
312 if (!IN6_IS_ADDR_MULTICAST(&sin6
.sin6_addr
) &&
313 (u
= kauth_cred_getuid(so
->so_cred
)) != 0) {
314 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
315 &sin6
.sin6_addr
, lport
,
317 if (t
!= NULL
&& (!IN6_IS_ADDR_UNSPECIFIED(
319 !IN6_IS_ADDR_UNSPECIFIED(&t
->in6p_laddr
) ||
320 !(t
->inp_socket
->so_options
&
321 SO_REUSEPORT
)) && (u
!= kauth_cred_getuid(
322 t
->inp_socket
->so_cred
)) &&
323 !(t
->inp_socket
->so_flags
&
324 SOF_REUSESHAREUID
)) {
325 lck_rw_done(pcbinfo
->ipi_lock
);
329 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
330 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
331 struct sockaddr_in sin
;
333 in6_sin6_2_sin(&sin
, &sin6
);
334 t
= in_pcblookup_local_and_cleanup(
335 pcbinfo
, sin
.sin_addr
, lport
,
338 !(t
->inp_socket
->so_options
&
340 (kauth_cred_getuid(so
->so_cred
) !=
341 kauth_cred_getuid(t
->inp_socket
->
342 so_cred
)) && (t
->inp_laddr
.s_addr
!=
343 INADDR_ANY
|| SOCK_DOM(so
) ==
344 SOCK_DOM(t
->inp_socket
))) {
345 lck_rw_done(pcbinfo
->ipi_lock
);
351 t
= in6_pcblookup_local_and_cleanup(pcbinfo
,
352 &sin6
.sin6_addr
, lport
, wild
);
354 (reuseport
& t
->inp_socket
->so_options
) == 0) {
355 lck_rw_done(pcbinfo
->ipi_lock
);
359 if (!(inp
->inp_flags
& IN6P_IPV6_V6ONLY
) &&
360 IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
361 struct sockaddr_in sin
;
363 in6_sin6_2_sin(&sin
, &sin6
);
364 t
= in_pcblookup_local_and_cleanup(pcbinfo
,
365 sin
.sin_addr
, lport
, wild
);
366 if (t
!= NULL
&& (reuseport
&
367 t
->inp_socket
->so_options
) == 0 &&
368 (t
->inp_laddr
.s_addr
!= INADDR_ANY
||
369 SOCK_DOM(so
) == SOCK_DOM(t
->inp_socket
))) {
370 lck_rw_done(pcbinfo
->ipi_lock
);
380 * We unlocked socket's protocol lock for a long time.
381 * The socket might have been dropped/defuncted.
382 * Checking if world has changed since.
384 if (inp
->inp_state
== INPCB_STATE_DEAD
) {
385 lck_rw_done(pcbinfo
->ipi_lock
);
386 return (ECONNABORTED
);
389 /* check if the socket got bound when the lock was released */
390 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
391 lck_rw_done(pcbinfo
->ipi_lock
);
395 if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
.sin6_addr
)) {
396 inp
->in6p_laddr
= sin6
.sin6_addr
;
397 inp
->in6p_last_outifp
= outif
;
402 if ((e
= in6_pcbsetport(&inp
->in6p_laddr
, inp
, p
, 1)) != 0) {
403 /* Undo any address bind from above. */
404 inp
->in6p_laddr
= in6addr_any
;
405 inp
->in6p_last_outifp
= NULL
;
406 lck_rw_done(pcbinfo
->ipi_lock
);
410 inp
->inp_lport
= lport
;
411 if (in_pcbinshash(inp
, 1) != 0) {
412 inp
->in6p_laddr
= in6addr_any
;
414 inp
->in6p_last_outifp
= NULL
;
415 lck_rw_done(pcbinfo
->ipi_lock
);
419 lck_rw_done(pcbinfo
->ipi_lock
);
420 sflt_notify(so
, sock_evt_bound
, NULL
);
425 * Transform old in6_pcbconnect() into an inner subroutine for new
426 * in6_pcbconnect(); do some validity-checking on the remote address
427 * (in "nam") and then determine local host address (i.e., which
428 * interface) to use to access that remote host.
430 * This routine may alter the caller-supplied remote address "nam".
432 * This routine might return an ifp with a reference held if the caller
433 * provides a non-NULL outif, even in the error case. The caller is
434 * responsible for releasing its reference.
437 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
438 struct in6_addr
*plocal_addr6
, struct ifnet
**outif
)
440 struct in6_addr
*addr6
= NULL
;
441 struct in6_addr src_storage
;
443 unsigned int ifscope
;
447 if (nam
->sa_len
!= sizeof (struct sockaddr_in6
))
449 if (SIN6(nam
)->sin6_family
!= AF_INET6
)
450 return (EAFNOSUPPORT
);
451 if (SIN6(nam
)->sin6_port
== 0)
452 return (EADDRNOTAVAIL
);
454 /* KAME hack: embed scopeid */
455 if (in6_embedscope(&SIN6(nam
)->sin6_addr
, SIN6(nam
), inp
, NULL
, NULL
) != 0)
460 * If the destination address is UNSPECIFIED addr,
461 * use the loopback addr, e.g ::1.
463 if (IN6_IS_ADDR_UNSPECIFIED(&SIN6(nam
)->sin6_addr
))
464 SIN6(nam
)->sin6_addr
= in6addr_loopback
;
467 ifscope
= (inp
->inp_flags
& INP_BOUND_IF
) ?
468 inp
->inp_boundifp
->if_index
: IFSCOPE_NONE
;
471 * XXX: in6_selectsrc might replace the bound local address
472 * with the address specified by setsockopt(IPV6_PKTINFO).
473 * Is it the intended behavior?
475 * in6_selectsrc() might return outif with its reference held
476 * even in the error case; caller always needs to release it
479 addr6
= in6_selectsrc(SIN6(nam
), inp
->in6p_outputopts
, inp
,
480 &inp
->in6p_route
, outif
, &src_storage
, ifscope
, &error
);
483 struct rtentry
*rt
= inp
->in6p_route
.ro_rt
;
485 * If in6_selectsrc() returns a route, it should be one
486 * which points to the same ifp as outif. Just in case
487 * it isn't, use the one from the route for consistency.
488 * Otherwise if there is no route, leave outif alone as
489 * it could still be useful to the caller.
491 if (rt
!= NULL
&& rt
->rt_ifp
!= *outif
) {
492 ifnet_reference(rt
->rt_ifp
); /* for caller */
494 ifnet_release(*outif
);
500 if (outif
!= NULL
&& (*outif
) != NULL
&&
501 inp_restricted_send(inp
, *outif
)) {
502 soevent(inp
->inp_socket
,
503 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
504 error
= EHOSTUNREACH
;
507 error
= EADDRNOTAVAIL
;
511 *plocal_addr6
= *addr6
;
513 * Don't do pcblookup call here; return interface in
514 * plocal_addr6 and exit to caller, that will do the lookup.
521 * Connect from a socket to a specified address.
522 * Both address and port must be specified in argument sin.
523 * If don't have a local address for this socket yet,
527 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct proc
*p
)
529 struct in6_addr addr6
;
530 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)(void *)nam
;
533 struct ifnet
*outif
= NULL
;
534 struct socket
*so
= inp
->inp_socket
;
538 so
->so_state_change_cnt
++;
541 if (so
->so_proto
->pr_protocol
== IPPROTO_UDP
&&
542 sin6
->sin6_port
== htons(53) && !(so
->so_flags1
& SOF1_DNS_COUNTED
)) {
543 so
->so_flags1
|= SOF1_DNS_COUNTED
;
544 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet_dgram_dns
);
548 * Call inner routine, to assign local interface address.
549 * in6_pcbladdr() may automatically fill in sin6_scope_id.
551 * in6_pcbladdr() might return an ifp with its reference held
552 * even in the error case, so make sure that it's released
553 * whenever it's non-NULL.
555 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, &outif
)) != 0) {
556 if (outif
!= NULL
&& inp_restricted_send(inp
, outif
))
558 (SO_FILT_HINT_LOCKED
| SO_FILT_HINT_IFDENIED
));
561 socket_unlock(so
, 0);
562 pcb
= in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
563 sin6
->sin6_port
, IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
) ?
564 &addr6
: &inp
->in6p_laddr
, inp
->inp_lport
, 0, NULL
);
567 in_pcb_checkstate(pcb
, WNT_RELEASE
, pcb
== inp
? 1 : 0);
571 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
572 if (inp
->inp_lport
== 0) {
573 error
= in6_pcbbind(inp
, NULL
, p
);
577 inp
->in6p_laddr
= addr6
;
578 inp
->in6p_last_outifp
= outif
; /* no reference needed */
579 inp
->in6p_flags
|= INP_IN6ADDR_ANY
;
581 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
582 /* lock inversion issue, mostly with udp multicast packets */
583 socket_unlock(so
, 0);
584 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
587 inp
->in6p_faddr
= sin6
->sin6_addr
;
588 inp
->inp_fport
= sin6
->sin6_port
;
589 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
590 nstat_pcb_invalidate_cache(inp
);
592 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
596 ifnet_release(outif
);
602 in6_pcbdisconnect(struct inpcb
*inp
)
604 struct socket
*so
= inp
->inp_socket
;
608 so
->so_state_change_cnt
++;
611 if (!lck_rw_try_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
)) {
612 /* lock inversion issue, mostly with udp multicast packets */
613 socket_unlock(so
, 0);
614 lck_rw_lock_exclusive(inp
->inp_pcbinfo
->ipi_lock
);
617 if (nstat_collect
&& SOCK_PROTO(so
) == IPPROTO_UDP
)
618 nstat_pcb_cache(inp
);
619 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof (inp
->in6p_faddr
));
621 /* clear flowinfo - RFC 6437 */
622 inp
->inp_flow
&= ~IPV6_FLOWLABEL_MASK
;
624 lck_rw_done(inp
->inp_pcbinfo
->ipi_lock
);
626 * A multipath subflow socket would have its SS_NOFDREF set by default,
627 * so check for SOF_MP_SUBFLOW socket flag before detaching the PCB;
628 * when the socket is closed for real, SOF_MP_SUBFLOW would be cleared.
630 if (!(so
->so_flags
& SOF_MP_SUBFLOW
) && (so
->so_state
& SS_NOFDREF
))
635 in6_pcbdetach(struct inpcb
*inp
)
637 struct socket
*so
= inp
->inp_socket
;
639 if (so
->so_pcb
== NULL
) {
640 /* PCB has been disposed */
641 panic("%s: inp=%p so=%p proto=%d so_pcb is null!\n", __func__
,
642 inp
, so
, SOCK_PROTO(so
));
647 if (inp
->in6p_sp
!= NULL
) {
648 (void) ipsec6_delete_pcbpolicy(inp
);
652 if (inp
->inp_stat
!= NULL
&& SOCK_PROTO(so
) == IPPROTO_UDP
) {
653 if (inp
->inp_stat
->rxpackets
== 0 && inp
->inp_stat
->txpackets
== 0) {
654 INC_ATOMIC_INT64_LIM(net_api_stats
.nas_socket_inet6_dgram_no_data
);
659 * Let NetworkStatistics know this PCB is going away
660 * before we detach it.
663 (SOCK_PROTO(so
) == IPPROTO_TCP
|| SOCK_PROTO(so
) == IPPROTO_UDP
))
664 nstat_pcb_detach(inp
);
665 /* mark socket state as dead */
666 if (in_pcb_checkstate(inp
, WNT_STOPUSING
, 1) != WNT_STOPUSING
) {
667 panic("%s: so=%p proto=%d couldn't set to STOPUSING\n",
668 __func__
, so
, SOCK_PROTO(so
));
672 if (!(so
->so_flags
& SOF_PCBCLEARING
)) {
673 struct ip_moptions
*imo
;
674 struct ip6_moptions
*im6o
;
677 if (inp
->in6p_options
!= NULL
) {
678 m_freem(inp
->in6p_options
);
679 inp
->in6p_options
= NULL
;
681 ip6_freepcbopts(inp
->in6p_outputopts
);
682 ROUTE_RELEASE(&inp
->in6p_route
);
683 /* free IPv4 related resources in case of mapped addr */
684 if (inp
->inp_options
!= NULL
) {
685 (void) m_free(inp
->inp_options
);
686 inp
->inp_options
= NULL
;
688 im6o
= inp
->in6p_moptions
;
689 inp
->in6p_moptions
= NULL
;
691 imo
= inp
->inp_moptions
;
692 inp
->inp_moptions
= NULL
;
694 sofreelastref(so
, 0);
695 inp
->inp_state
= INPCB_STATE_DEAD
;
696 /* makes sure we're not called twice from so_close */
697 so
->so_flags
|= SOF_PCBCLEARING
;
699 inpcb_gc_sched(inp
->inp_pcbinfo
, INPCB_TIMER_FAST
);
702 * See inp_join_group() for why we need to unlock
704 if (im6o
!= NULL
|| imo
!= NULL
) {
705 socket_unlock(so
, 0);
716 in6_sockaddr(in_port_t port
, struct in6_addr
*addr_p
)
718 struct sockaddr_in6
*sin6
;
720 MALLOC(sin6
, struct sockaddr_in6
*, sizeof (*sin6
), M_SONAME
, M_WAITOK
);
723 bzero(sin6
, sizeof (*sin6
));
724 sin6
->sin6_family
= AF_INET6
;
725 sin6
->sin6_len
= sizeof (*sin6
);
726 sin6
->sin6_port
= port
;
727 sin6
->sin6_addr
= *addr_p
;
729 /* would be good to use sa6_recoverscope(), except for locking */
730 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
731 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
733 sin6
->sin6_scope_id
= 0; /* XXX */
734 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
735 sin6
->sin6_addr
.s6_addr16
[1] = 0;
737 return ((struct sockaddr
*)sin6
);
741 in6_sockaddr_s(in_port_t port
, struct in6_addr
*addr_p
,
742 struct sockaddr_in6
*sin6
)
744 bzero(sin6
, sizeof (*sin6
));
745 sin6
->sin6_family
= AF_INET6
;
746 sin6
->sin6_len
= sizeof (*sin6
);
747 sin6
->sin6_port
= port
;
748 sin6
->sin6_addr
= *addr_p
;
750 /* would be good to use sa6_recoverscope(), except for locking */
751 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
752 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
754 sin6
->sin6_scope_id
= 0; /* XXX */
755 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
756 sin6
->sin6_addr
.s6_addr16
[1] = 0;
760 * The calling convention of in6_getsockaddr() and in6_getpeeraddr() was
761 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
762 * in struct pr_usrreqs, so that protocols can just reference then directly
763 * without the need for a wrapper function.
766 in6_getsockaddr(struct socket
*so
, struct sockaddr
**nam
)
769 struct in6_addr addr
;
772 if ((inp
= sotoinpcb(so
)) == NULL
)
775 port
= inp
->inp_lport
;
776 addr
= inp
->in6p_laddr
;
778 *nam
= in6_sockaddr(port
, &addr
);
785 in6_getsockaddr_s(struct socket
*so
, struct sockaddr_in6
*ss
)
788 struct in6_addr addr
;
792 bzero(ss
, sizeof (*ss
));
794 if ((inp
= sotoinpcb(so
)) == NULL
)
797 port
= inp
->inp_lport
;
798 addr
= inp
->in6p_laddr
;
800 in6_sockaddr_s(port
, &addr
, ss
);
805 in6_getpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
808 struct in6_addr addr
;
811 if ((inp
= sotoinpcb(so
)) == NULL
)
814 port
= inp
->inp_fport
;
815 addr
= inp
->in6p_faddr
;
817 *nam
= in6_sockaddr(port
, &addr
);
824 in6_mapped_sockaddr(struct socket
*so
, struct sockaddr
**nam
)
826 struct inpcb
*inp
= sotoinpcb(so
);
831 if (inp
->inp_vflag
& INP_IPV4
) {
832 error
= in_getsockaddr(so
, nam
);
834 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
836 /* scope issues will be handled in in6_getsockaddr(). */
837 error
= in6_getsockaddr(so
, nam
);
843 in6_mapped_peeraddr(struct socket
*so
, struct sockaddr
**nam
)
845 struct inpcb
*inp
= sotoinpcb(so
);
850 if (inp
->inp_vflag
& INP_IPV4
) {
851 error
= in_getpeeraddr(so
, nam
);
853 error
= in6_sin_2_v4mapsin6_in_sock(nam
);
855 /* scope issues will be handled in in6_getpeeraddr(). */
856 error
= in6_getpeeraddr(so
, nam
);
862 * Pass some notification to all connections of a protocol
863 * associated with address dst. The local address and/or port numbers
864 * may be specified to limit the search. The "usual action" will be
865 * taken, depending on the ctlinput cmd. The caller must filter any
866 * cmds that are uninteresting (e.g., no error in the map).
867 * Call the protocol specific routine (if any) to report
868 * any errors for each matching socket.
871 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, u_int fport_arg
,
872 const struct sockaddr
*src
, u_int lport_arg
, int cmd
, void *cmdarg
,
873 void (*notify
)(struct inpcb
*, int))
875 struct inpcbhead
*head
= pcbinfo
->ipi_listhead
;
876 struct inpcb
*inp
, *ninp
;
877 struct sockaddr_in6 sa6_src
, *sa6_dst
;
878 u_short fport
= fport_arg
, lport
= lport_arg
;
882 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
)
885 sa6_dst
= (struct sockaddr_in6
*)(void *)dst
;
886 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
))
890 * note that src can be NULL when we get notify by local fragmentation.
892 sa6_src
= (src
== NULL
) ?
893 sa6_any
: *(struct sockaddr_in6
*)(uintptr_t)(size_t)src
;
894 flowinfo
= sa6_src
.sin6_flowinfo
;
897 * Redirects go to all references to the destination,
898 * and use in6_rtchange to invalidate the route cache.
899 * Dead host indications: also use in6_rtchange to invalidate
900 * the cache, and deliver the error to all the sockets.
901 * Otherwise, if we have knowledge of the local port and address,
902 * deliver only to that socket.
904 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
907 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof (sa6_src
.sin6_addr
));
909 if (cmd
!= PRC_HOSTDEAD
)
910 notify
= in6_rtchange
;
912 errno
= inet6ctlerrmap
[cmd
];
913 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
914 for (inp
= LIST_FIRST(head
); inp
!= NULL
; inp
= ninp
) {
915 ninp
= LIST_NEXT(inp
, inp_list
);
917 if (!(inp
->inp_vflag
& INP_IPV6
))
921 * If the error designates a new path MTU for a destination
922 * and the application (associated with this socket) wanted to
923 * know the value, notify. Note that we notify for all
924 * disconnected sockets if the corresponding application
925 * wanted. This is because some UDP applications keep sending
926 * sockets disconnected.
927 * XXX: should we avoid to notify the value to TCP sockets?
929 if (cmd
== PRC_MSGSIZE
)
930 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)(void *)dst
,
931 (u_int32_t
*)cmdarg
);
934 * Detect if we should notify the error. If no source and
935 * destination ports are specifed, but non-zero flowinfo and
936 * local address match, notify the error. This is the case
937 * when the error is delivered with an encrypted buffer
938 * by ESP. Otherwise, just compare addresses and ports
941 if (lport
== 0 && fport
== 0 && flowinfo
&&
942 inp
->inp_socket
!= NULL
&&
943 flowinfo
== (inp
->inp_flow
& IPV6_FLOWLABEL_MASK
) &&
944 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
))
946 else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
947 &sa6_dst
->sin6_addr
) || inp
->inp_socket
== NULL
||
948 (lport
&& inp
->inp_lport
!= lport
) ||
949 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
950 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
951 &sa6_src
.sin6_addr
)) || (fport
&& inp
->inp_fport
!= fport
))
956 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) ==
959 socket_lock(inp
->inp_socket
, 1);
960 (*notify
)(inp
, errno
);
961 (void) in_pcb_checkstate(inp
, WNT_RELEASE
, 1);
962 socket_unlock(inp
->inp_socket
, 1);
965 lck_rw_done(pcbinfo
->ipi_lock
);
969 * Lookup a PCB based on the local address and port.
972 in6_pcblookup_local(struct inpcbinfo
*pcbinfo
, struct in6_addr
*laddr
,
973 u_int lport_arg
, int wild_okay
)
976 int matchwild
= 3, wildcard
;
977 u_short lport
= lport_arg
;
978 struct inpcbporthead
*porthash
;
979 struct inpcb
*match
= NULL
;
980 struct inpcbport
*phd
;
983 struct inpcbhead
*head
;
985 * Look for an unconnected (wildcard foreign addr) PCB that
986 * matches the local address and port we're looking for.
988 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
989 pcbinfo
->ipi_hashmask
)];
990 LIST_FOREACH(inp
, head
, inp_hash
) {
991 if (!(inp
->inp_vflag
& INP_IPV6
))
993 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
994 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
995 inp
->inp_lport
== lport
) {
1008 * Best fit PCB lookup.
1010 * First see if this local port is in use by looking on the
1013 porthash
= &pcbinfo
->ipi_porthashbase
[INP_PCBPORTHASH(lport
,
1014 pcbinfo
->ipi_porthashmask
)];
1015 LIST_FOREACH(phd
, porthash
, phd_hash
) {
1016 if (phd
->phd_port
== lport
)
1021 * Port is in use by one or more PCBs. Look for best
1024 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
1026 if (!(inp
->inp_vflag
& INP_IPV6
))
1028 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
1030 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
1031 if (IN6_IS_ADDR_UNSPECIFIED(laddr
))
1033 else if (!IN6_ARE_ADDR_EQUAL(
1034 &inp
->in6p_laddr
, laddr
))
1037 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
))
1040 if (wildcard
< matchwild
) {
1042 matchwild
= wildcard
;
1043 if (matchwild
== 0) {
1053 * Check for alternatives when higher level complains
1054 * about service problems. For now, invalidate cached
1055 * routing information. If the route was created dynamically
1056 * (by a redirect), time to try a default gateway again.
1059 in6_losing(struct inpcb
*in6p
)
1063 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
1065 if (rt
->rt_flags
& RTF_DYNAMIC
) {
1067 * Prevent another thread from modifying rt_key,
1068 * rt_gateway via rt_setgate() after the rt_lock
1069 * is dropped by marking the route as defunct.
1071 rt
->rt_flags
|= RTF_CONDEMNED
;
1073 (void) rtrequest(RTM_DELETE
, rt_key(rt
),
1074 rt
->rt_gateway
, rt_mask(rt
), rt
->rt_flags
, NULL
);
1079 * A new route can be allocated
1080 * the next time output is attempted.
1083 ROUTE_RELEASE(&in6p
->in6p_route
);
1087 * After a routing change, flush old routing
1088 * and allocate a (hopefully) better one.
1091 in6_rtchange(struct inpcb
*inp
, int errno
)
1093 #pragma unused(errno)
1095 * A new route can be allocated the next time
1096 * output is attempted.
1098 ROUTE_RELEASE(&inp
->in6p_route
);
1102 * Check if PCB exists hash list. Also returns uid and gid of socket
1105 in6_pcblookup_hash_exists(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1106 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1107 uid_t
*uid
, gid_t
*gid
, struct ifnet
*ifp
)
1109 struct inpcbhead
*head
;
1111 u_short fport
= fport_arg
, lport
= lport_arg
;
1117 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1120 * First look for an exact match.
1122 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1123 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1124 LIST_FOREACH(inp
, head
, inp_hash
) {
1125 if (!(inp
->inp_vflag
& INP_IPV6
))
1128 if (inp_restricted_recv(inp
, ifp
))
1131 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1132 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1133 inp
->inp_fport
== fport
&&
1134 inp
->inp_lport
== lport
) {
1135 if ((found
= (inp
->inp_socket
!= NULL
))) {
1137 * Found. Check if pcb is still valid
1139 *uid
= kauth_cred_getuid(
1140 inp
->inp_socket
->so_cred
);
1141 *gid
= kauth_cred_getgid(
1142 inp
->inp_socket
->so_cred
);
1144 lck_rw_done(pcbinfo
->ipi_lock
);
1149 struct inpcb
*local_wild
= NULL
;
1151 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1152 pcbinfo
->ipi_hashmask
)];
1153 LIST_FOREACH(inp
, head
, inp_hash
) {
1154 if (!(inp
->inp_vflag
& INP_IPV6
))
1157 if (inp_restricted_recv(inp
, ifp
))
1160 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1161 inp
->inp_lport
== lport
) {
1162 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1164 found
= (inp
->inp_socket
!= NULL
);
1166 *uid
= kauth_cred_getuid(
1167 inp
->inp_socket
->so_cred
);
1168 *gid
= kauth_cred_getgid(
1169 inp
->inp_socket
->so_cred
);
1171 lck_rw_done(pcbinfo
->ipi_lock
);
1173 } else if (IN6_IS_ADDR_UNSPECIFIED(
1174 &inp
->in6p_laddr
)) {
1180 if ((found
= (local_wild
->inp_socket
!= NULL
))) {
1181 *uid
= kauth_cred_getuid(
1182 local_wild
->inp_socket
->so_cred
);
1183 *gid
= kauth_cred_getgid(
1184 local_wild
->inp_socket
->so_cred
);
1186 lck_rw_done(pcbinfo
->ipi_lock
);
1194 lck_rw_done(pcbinfo
->ipi_lock
);
1199 * Lookup PCB in hash list.
1202 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1203 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
, int wildcard
,
1206 struct inpcbhead
*head
;
1208 u_short fport
= fport_arg
, lport
= lport_arg
;
1210 lck_rw_lock_shared(pcbinfo
->ipi_lock
);
1213 * First look for an exact match.
1215 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(faddr
->s6_addr32
[3] /* XXX */,
1216 lport
, fport
, pcbinfo
->ipi_hashmask
)];
1217 LIST_FOREACH(inp
, head
, inp_hash
) {
1218 if (!(inp
->inp_vflag
& INP_IPV6
))
1221 if (inp_restricted_recv(inp
, ifp
))
1224 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1225 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1226 inp
->inp_fport
== fport
&&
1227 inp
->inp_lport
== lport
) {
1229 * Found. Check if pcb is still valid
1231 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) !=
1233 lck_rw_done(pcbinfo
->ipi_lock
);
1236 /* it's there but dead, say it isn't found */
1237 lck_rw_done(pcbinfo
->ipi_lock
);
1243 struct inpcb
*local_wild
= NULL
;
1245 head
= &pcbinfo
->ipi_hashbase
[INP_PCBHASH(INADDR_ANY
, lport
, 0,
1246 pcbinfo
->ipi_hashmask
)];
1247 LIST_FOREACH(inp
, head
, inp_hash
) {
1248 if (!(inp
->inp_vflag
& INP_IPV6
))
1251 if (inp_restricted_recv(inp
, ifp
))
1254 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1255 inp
->inp_lport
== lport
) {
1256 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1258 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
,
1259 0) != WNT_STOPUSING
) {
1260 lck_rw_done(pcbinfo
->ipi_lock
);
1263 /* dead; say it isn't found */
1264 lck_rw_done(pcbinfo
->ipi_lock
);
1267 } else if (IN6_IS_ADDR_UNSPECIFIED(
1268 &inp
->in6p_laddr
)) {
1273 if (local_wild
&& in_pcb_checkstate(local_wild
,
1274 WNT_ACQUIRE
, 0) != WNT_STOPUSING
) {
1275 lck_rw_done(pcbinfo
->ipi_lock
);
1276 return (local_wild
);
1278 lck_rw_done(pcbinfo
->ipi_lock
);
1286 lck_rw_done(pcbinfo
->ipi_lock
);
1291 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
1295 ip
= mtod(m
, struct ip6_hdr
*);
1296 bzero(sin6
, sizeof (*sin6
));
1297 sin6
->sin6_len
= sizeof (*sin6
);
1298 sin6
->sin6_family
= AF_INET6
;
1299 sin6
->sin6_addr
= ip
->ip6_src
;
1300 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
)) {
1301 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1302 if ((m
->m_pkthdr
.pkt_flags
& (PKTF_LOOP
|PKTF_IFAINFO
)) ==
1303 (PKTF_LOOP
|PKTF_IFAINFO
))
1304 sin6
->sin6_scope_id
= m
->m_pkthdr
.src_ifindex
;
1305 else if (m
->m_pkthdr
.rcvif
!= NULL
)
1306 sin6
->sin6_scope_id
= m
->m_pkthdr
.rcvif
->if_index
;
1311 * The following routines implement this scheme:
1313 * Callers of ip6_output() that intend to cache the route in the inpcb pass
1314 * a local copy of the struct route to ip6_output(). Using a local copy of
1315 * the cached route significantly simplifies things as IP no longer has to
1316 * worry about having exclusive access to the passed in struct route, since
1317 * it's defined in the caller's stack; in essence, this allows for a lock-
1318 * less operation when updating the struct route at the IP level and below,
1319 * whenever necessary. The scheme works as follows:
1321 * Prior to dropping the socket's lock and calling ip6_output(), the caller
1322 * copies the struct route from the inpcb into its stack, and adds a reference
1323 * to the cached route entry, if there was any. The socket's lock is then
1324 * dropped and ip6_output() is called with a pointer to the copy of struct
1325 * route defined on the stack (not to the one in the inpcb.)
1327 * Upon returning from ip6_output(), the caller then acquires the socket's
1328 * lock and synchronizes the cache; if there is no route cached in the inpcb,
1329 * it copies the local copy of struct route (which may or may not contain any
1330 * route) back into the cache; otherwise, if the inpcb has a route cached in
1331 * it, the one in the local copy will be freed, if there's any. Trashing the
1332 * cached route in the inpcb can be avoided because ip6_output() is single-
1333 * threaded per-PCB (i.e. multiple transmits on a PCB are always serialized
1334 * by the socket/transport layer.)
1337 in6p_route_copyout(struct inpcb
*inp
, struct route_in6
*dst
)
1339 struct route_in6
*src
= &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_copyout((struct route
*)dst
, (struct route
*)src
, sizeof (*dst
));
1351 in6p_route_copyin(struct inpcb
*inp
, struct route_in6
*src
)
1353 struct route_in6
*dst
= &inp
->in6p_route
;
1355 socket_lock_assert_owned(inp
->inp_socket
);
1357 /* Minor sanity check */
1358 if (src
->ro_rt
!= NULL
&& rt_key(src
->ro_rt
)->sa_family
!= AF_INET6
)
1359 panic("%s: wrong or corrupted route: %p", __func__
, src
);
1361 route_copyin((struct route
*)src
, (struct route
*)dst
, sizeof (*src
));