2 * Copyright (c) 2004 Jeffrey M. Hsu. All rights reserved.
3 * Copyright (c) 2004 The DragonFly Project. All rights reserved.
5 * This code is derived from software contributed to The DragonFly Project
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of The DragonFly Project nor the names of its
17 * contributors may be used to endorse or promote products derived
18 * from this software without specific, prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * Copyright (c) 1982, 1986, 1991, 1993, 1995
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
66 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95
67 * $FreeBSD: src/sys/netinet/in_pcb.c,v 1.59.2.27 2004/01/02 04:06:42 ambrisko Exp $
70 #include "opt_ipsec.h"
71 #include "opt_inet6.h"
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/malloc.h>
77 #include <sys/domain.h>
78 #include <sys/protosw.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
84 #include <sys/kernel.h>
85 #include <sys/sysctl.h>
87 #include <sys/thread2.h>
88 #include <sys/socketvar2.h>
89 #include <sys/msgport2.h>
91 #include <machine/limits.h>
94 #include <net/if_types.h>
95 #include <net/route.h>
97 #include <netinet/in.h>
98 #include <netinet/in_pcb.h>
99 #include <netinet/in_var.h>
100 #include <netinet/ip_var.h>
102 #include <netinet/ip6.h>
103 #include <netinet6/ip6_var.h>
107 #include <netinet6/ipsec.h>
108 #include <netproto/key/key.h>
109 #include <netproto/ipsec/esp_var.h>
113 #if defined(IPSEC) || defined(IPSEC_ESP)
114 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!"
117 #include <netproto/ipsec/ipsec.h>
118 #include <netproto/ipsec/key.h>
120 #endif /* FAST_IPSEC */
122 struct in_addr zeroin_addr
;
125 * These configure the range of local port addresses assigned to
126 * "unspecified" outgoing connections/packets/whatever.
128 int ipport_lowfirstauto
= IPPORT_RESERVED
- 1; /* 1023 */
129 int ipport_lowlastauto
= IPPORT_RESERVEDSTART
; /* 600 */
131 int ipport_firstauto
= IPPORT_RESERVED
; /* 1024 */
132 int ipport_lastauto
= IPPORT_USERRESERVED
; /* 5000 */
134 int ipport_hifirstauto
= IPPORT_HIFIRSTAUTO
; /* 49152 */
135 int ipport_hilastauto
= IPPORT_HILASTAUTO
; /* 65535 */
137 #define RANGECHK(var, min, max) \
138 if ((var) < (min)) { (var) = (min); } \
139 else if ((var) > (max)) { (var) = (max); }
141 int udpencap_enable
= 1; /* enabled by default */
142 int udpencap_port
= 4500; /* triggers decapsulation */
145 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS
)
149 error
= sysctl_handle_int(oidp
, oidp
->oid_arg1
, oidp
->oid_arg2
, req
);
151 RANGECHK(ipport_lowfirstauto
, 1, IPPORT_RESERVED
- 1);
152 RANGECHK(ipport_lowlastauto
, 1, IPPORT_RESERVED
- 1);
154 RANGECHK(ipport_firstauto
, IPPORT_RESERVED
, USHRT_MAX
);
155 RANGECHK(ipport_lastauto
, IPPORT_RESERVED
, USHRT_MAX
);
157 RANGECHK(ipport_hifirstauto
, IPPORT_RESERVED
, USHRT_MAX
);
158 RANGECHK(ipport_hilastauto
, IPPORT_RESERVED
, USHRT_MAX
);
165 SYSCTL_NODE(_net_inet_ip
, IPPROTO_IP
, portrange
, CTLFLAG_RW
, 0, "IP Ports");
167 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, lowfirst
, CTLTYPE_INT
|CTLFLAG_RW
,
168 &ipport_lowfirstauto
, 0, &sysctl_net_ipport_check
, "I", "");
169 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, lowlast
, CTLTYPE_INT
|CTLFLAG_RW
,
170 &ipport_lowlastauto
, 0, &sysctl_net_ipport_check
, "I", "");
171 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, first
, CTLTYPE_INT
|CTLFLAG_RW
,
172 &ipport_firstauto
, 0, &sysctl_net_ipport_check
, "I", "");
173 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, last
, CTLTYPE_INT
|CTLFLAG_RW
,
174 &ipport_lastauto
, 0, &sysctl_net_ipport_check
, "I", "");
175 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, hifirst
, CTLTYPE_INT
|CTLFLAG_RW
,
176 &ipport_hifirstauto
, 0, &sysctl_net_ipport_check
, "I", "");
177 SYSCTL_PROC(_net_inet_ip_portrange
, OID_AUTO
, hilast
, CTLTYPE_INT
|CTLFLAG_RW
,
178 &ipport_hilastauto
, 0, &sysctl_net_ipport_check
, "I", "");
181 * in_pcb.c: manage the Protocol Control Blocks.
183 * NOTE: It is assumed that most of these functions will be called from
184 * a critical section. XXX - There are, unfortunately, a few exceptions
185 * to this rule that should be fixed.
187 * NOTE: The caller should initialize the cpu field to the cpu running the
188 * protocol stack associated with this inpcbinfo.
192 in_pcbinfo_init(struct inpcbinfo
*pcbinfo
)
194 LIST_INIT(&pcbinfo
->pcblisthead
);
196 pcbinfo
->portsave
= kmalloc(sizeof(*pcbinfo
->portsave
), M_PCB
,
200 struct baddynamicports baddynamicports
;
203 * Check if the specified port is invalid for dynamic allocation.
206 in_baddynamic(u_int16_t port
, u_int16_t proto
)
210 return (DP_ISSET(baddynamicports
.tcp
, port
));
213 /* Cannot preset this as it is a sysctl */
214 if (port
== udpencap_port
)
217 return (DP_ISSET(baddynamicports
.udp
, port
));
225 * Allocate a PCB and associate it with the socket.
228 in_pcballoc(struct socket
*so
, struct inpcbinfo
*pcbinfo
)
235 inp
= kmalloc(pcbinfo
->ipi_size
, M_PCB
, M_WAITOK
|M_ZERO
);
236 inp
->inp_gencnt
= ++pcbinfo
->ipi_gencnt
;
237 inp
->inp_pcbinfo
= inp
->inp_cpcbinfo
= pcbinfo
;
238 inp
->inp_socket
= so
;
240 error
= ipsec_init_policy(so
, &inp
->inp_sp
);
247 if (INP_SOCKAF(so
) == AF_INET6
&& ip6_v6only
)
248 inp
->inp_flags
|= IN6P_IPV6_V6ONLY
;
249 if (ip6_auto_flowlabel
)
250 inp
->inp_flags
|= IN6P_AUTOFLOWLABEL
;
254 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, inp
, inp_list
);
255 pcbinfo
->ipi_count
++;
260 * Unlink a pcb with the intention of moving it to another cpu with a
261 * different pcbinfo. While unlinked nothing should attempt to dereference
262 * inp_pcbinfo, NULL it out so we assert if it does.
265 in_pcbunlink(struct inpcb
*inp
, struct inpcbinfo
*pcbinfo
)
267 KKASSERT(inp
->inp_pcbinfo
== pcbinfo
);
269 LIST_REMOVE(inp
, inp_list
);
270 pcbinfo
->ipi_count
--;
271 inp
->inp_pcbinfo
= NULL
;
275 * Relink a pcb into a new pcbinfo.
278 in_pcblink(struct inpcb
*inp
, struct inpcbinfo
*pcbinfo
)
280 KKASSERT(inp
->inp_pcbinfo
== NULL
);
281 inp
->inp_pcbinfo
= pcbinfo
;
282 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, inp
, inp_list
);
283 pcbinfo
->ipi_count
++;
287 in_pcbbind(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
289 struct socket
*so
= inp
->inp_socket
;
290 unsigned short *lastport
;
291 struct sockaddr_in
*sin
;
292 struct sockaddr_in jsin
;
293 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
294 struct ucred
*cred
= NULL
;
296 int wild
= 0, reuseport
= (so
->so_options
& SO_REUSEPORT
);
299 if (TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
])) /* XXX broken! */
300 return (EADDRNOTAVAIL
);
301 if (inp
->inp_lport
!= 0 || inp
->inp_laddr
.s_addr
!= INADDR_ANY
)
302 return (EINVAL
); /* already bound */
304 if (!(so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)))
305 wild
= 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */
307 cred
= td
->td_proc
->p_ucred
;
310 * This has to be atomic. If the porthash is shared across multiple
311 * protocol threads (aka tcp) then the token will be non-NULL.
313 if (pcbinfo
->porttoken
)
314 lwkt_gettoken(pcbinfo
->porttoken
);
317 sin
= (struct sockaddr_in
*)nam
;
318 if (nam
->sa_len
!= sizeof *sin
) {
324 * We should check the family, but old programs
325 * incorrectly fail to initialize it.
327 if (sin
->sin_family
!= AF_INET
) {
328 error
= EAFNOSUPPORT
;
332 if (!prison_replace_wildcards(td
, nam
)) {
336 lport
= sin
->sin_port
;
337 if (IN_MULTICAST(ntohl(sin
->sin_addr
.s_addr
))) {
339 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
340 * allow complete duplication of binding if
341 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
342 * and a multicast address is bound on both
343 * new and duplicated sockets.
345 if (so
->so_options
& SO_REUSEADDR
)
346 reuseport
= SO_REUSEADDR
| SO_REUSEPORT
;
347 } else if (sin
->sin_addr
.s_addr
!= INADDR_ANY
) {
348 sin
->sin_port
= 0; /* yech... */
349 bzero(&sin
->sin_zero
, sizeof sin
->sin_zero
);
350 if (ifa_ifwithaddr((struct sockaddr
*)sin
) == NULL
) {
351 error
= EADDRNOTAVAIL
;
359 if (ntohs(lport
) < IPPORT_RESERVED
&&
361 priv_check_cred(cred
, PRIV_NETINET_RESERVEDPORT
, 0)) {
365 if (so
->so_cred
->cr_uid
!= 0 &&
366 !IN_MULTICAST(ntohl(sin
->sin_addr
.s_addr
))) {
367 t
= in_pcblookup_local(pcbinfo
,
373 (!in_nullhost(sin
->sin_addr
) ||
374 !in_nullhost(t
->inp_laddr
) ||
375 (t
->inp_socket
->so_options
&
376 SO_REUSEPORT
) == 0) &&
377 (so
->so_cred
->cr_uid
!=
378 t
->inp_socket
->so_cred
->cr_uid
)) {
380 if (!in_nullhost(sin
->sin_addr
) ||
381 !in_nullhost(t
->inp_laddr
) ||
383 INP_SOCKAF(t
->inp_socket
))
391 if (cred
&& !prison_replace_wildcards(td
, nam
)) {
392 error
= EADDRNOTAVAIL
;
395 t
= in_pcblookup_local(pcbinfo
, sin
->sin_addr
, lport
,
397 if (t
&& !(reuseport
& t
->inp_socket
->so_options
)) {
399 if (!in_nullhost(sin
->sin_addr
) ||
400 !in_nullhost(t
->inp_laddr
) ||
401 INP_SOCKAF(so
) == INP_SOCKAF(t
->inp_socket
))
409 inp
->inp_laddr
= sin
->sin_addr
;
415 jsin
.sin_family
= AF_INET
;
416 jsin
.sin_addr
.s_addr
= inp
->inp_laddr
.s_addr
;
417 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin
)) {
418 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
422 inp
->inp_laddr
.s_addr
= jsin
.sin_addr
.s_addr
;
424 inp
->inp_flags
|= INP_ANONPORT
;
426 if (inp
->inp_flags
& INP_HIGHPORT
) {
427 first
= ipport_hifirstauto
; /* sysctl */
428 last
= ipport_hilastauto
;
429 lastport
= &pcbinfo
->lasthi
;
430 } else if (inp
->inp_flags
& INP_LOWPORT
) {
432 (error
= priv_check_cred(cred
, PRIV_NETINET_RESERVEDPORT
, 0))) {
433 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
436 first
= ipport_lowfirstauto
; /* 1023 */
437 last
= ipport_lowlastauto
; /* 600 */
438 lastport
= &pcbinfo
->lastlow
;
440 first
= ipport_firstauto
; /* sysctl */
441 last
= ipport_lastauto
;
442 lastport
= &pcbinfo
->lastport
;
445 * Simple check to ensure all ports are not used up causing
448 * We split the two cases (up and down) so that the direction
449 * is not being tested on each round of the loop.
455 count
= first
- last
;
458 if (count
-- < 0) { /* completely used? */
459 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
460 error
= EADDRNOTAVAIL
;
464 if (*lastport
> first
|| *lastport
< last
)
466 lport
= htons(*lastport
);
467 } while (in_pcblookup_local(pcbinfo
, inp
->inp_laddr
,
473 count
= last
- first
;
476 if (count
-- < 0) { /* completely used? */
477 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
478 error
= EADDRNOTAVAIL
;
482 if (*lastport
< first
|| *lastport
> last
)
484 lport
= htons(*lastport
);
485 } while (in_pcblookup_local(pcbinfo
, inp
->inp_laddr
,
489 inp
->inp_lport
= lport
;
491 jsin
.sin_family
= AF_INET
;
492 jsin
.sin_addr
.s_addr
= inp
->inp_laddr
.s_addr
;
493 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin
)) {
494 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
499 inp
->inp_laddr
.s_addr
= jsin
.sin_addr
.s_addr
;
501 if (in_pcbinsporthash(inp
) != 0) {
502 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
509 if (pcbinfo
->porttoken
)
510 lwkt_reltoken(pcbinfo
->porttoken
);
514 static struct inpcb
*
515 in_pcblookup_addrport(struct inpcbinfo
*pcbinfo
, struct in_addr laddr
,
516 u_short lport
, struct in_addr faddr
, u_short fport
, struct ucred
*cred
)
519 struct inpcbporthead
*porthash
;
520 struct inpcbport
*phd
;
521 struct inpcb
*match
= NULL
;
524 * If the porthashbase is shared across several cpus we need
527 if (pcbinfo
->porttoken
)
528 lwkt_gettoken(pcbinfo
->porttoken
);
531 * Best fit PCB lookup.
533 * First see if this local port is in use by looking on the
536 porthash
= &pcbinfo
->porthashbase
[
537 INP_PCBPORTHASH(lport
, pcbinfo
->porthashmask
)];
538 LIST_FOREACH(phd
, porthash
, phd_hash
) {
539 if (phd
->phd_port
== lport
)
543 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
545 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
548 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
&&
549 inp
->inp_laddr
.s_addr
!= laddr
.s_addr
)
552 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
&&
553 inp
->inp_faddr
.s_addr
!= faddr
.s_addr
)
556 if (inp
->inp_fport
!= 0 && inp
->inp_fport
!= fport
)
561 inp
->inp_socket
->so_cred
->cr_prison
) {
567 if (pcbinfo
->porttoken
)
568 lwkt_reltoken(pcbinfo
->porttoken
);
573 in_pcbconn_bind(struct inpcb
*inp
, const struct sockaddr
*nam
,
576 struct proc
*p
= td
->td_proc
;
577 unsigned short *lastport
;
578 const struct sockaddr_in
*sin
= (const struct sockaddr_in
*)nam
;
579 struct sockaddr_in jsin
;
580 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
581 struct ucred
*cred
= NULL
;
584 int count
, error
, dup
= 0;
586 if (TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
])) /* XXX broken! */
587 return (EADDRNOTAVAIL
);
589 KKASSERT(inp
->inp_laddr
.s_addr
!= INADDR_ANY
);
590 if (inp
->inp_lport
!= 0)
591 return (EINVAL
); /* already bound */
597 * This has to be atomic. If the porthash is shared across multiple
598 * protocol threads (aka tcp) then the token will be non-NULL.
600 if (pcbinfo
->porttoken
)
601 lwkt_gettoken(pcbinfo
->porttoken
);
603 jsin
.sin_family
= AF_INET
;
604 jsin
.sin_addr
.s_addr
= inp
->inp_laddr
.s_addr
;
605 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin
)) {
606 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
610 inp
->inp_laddr
.s_addr
= jsin
.sin_addr
.s_addr
;
612 inp
->inp_flags
|= INP_ANONPORT
;
614 if (inp
->inp_flags
& INP_HIGHPORT
) {
615 first
= ipport_hifirstauto
; /* sysctl */
616 last
= ipport_hilastauto
;
617 lastport
= &pcbinfo
->lasthi
;
618 } else if (inp
->inp_flags
& INP_LOWPORT
) {
620 (error
= priv_check_cred(cred
, PRIV_NETINET_RESERVEDPORT
, 0))) {
621 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
624 first
= ipport_lowfirstauto
; /* 1023 */
625 last
= ipport_lowlastauto
; /* 600 */
626 lastport
= &pcbinfo
->lastlow
;
628 first
= ipport_firstauto
; /* sysctl */
629 last
= ipport_lastauto
;
630 lastport
= &pcbinfo
->lastport
;
635 * Simple check to ensure all ports are not used up causing
638 * We split the two cases (up and down) so that the direction
639 * is not being tested on each round of the loop.
645 count
= first
- last
;
648 if (count
-- < 0) { /* completely used? */
649 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
650 error
= EADDRNOTAVAIL
;
654 if (*lastport
> first
|| *lastport
< last
)
656 lport
= htons(*lastport
);
657 } while (in_pcblookup_addrport(pcbinfo
, inp
->inp_laddr
, lport
,
658 sin
->sin_addr
, sin
->sin_port
, cred
));
663 count
= last
- first
;
666 if (count
-- < 0) { /* completely used? */
667 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
668 error
= EADDRNOTAVAIL
;
672 if (*lastport
< first
|| *lastport
> last
)
674 lport
= htons(*lastport
);
675 } while (in_pcblookup_addrport(pcbinfo
, inp
->inp_laddr
, lport
,
676 sin
->sin_addr
, sin
->sin_port
, cred
));
679 /* This could happen on loopback interface */
680 if (sin
->sin_port
== lport
&&
681 sin
->sin_addr
.s_addr
== inp
->inp_laddr
.s_addr
) {
684 * Duplicate again; give up
686 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
687 error
= EADDRNOTAVAIL
;
693 inp
->inp_lport
= lport
;
695 jsin
.sin_family
= AF_INET
;
696 jsin
.sin_addr
.s_addr
= inp
->inp_laddr
.s_addr
;
697 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin
)) {
698 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
703 inp
->inp_laddr
.s_addr
= jsin
.sin_addr
.s_addr
;
705 if (in_pcbinsporthash(inp
) != 0) {
706 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
713 if (pcbinfo
->porttoken
)
714 lwkt_reltoken(pcbinfo
->porttoken
);
719 * Transform old in_pcbconnect() into an inner subroutine for new
720 * in_pcbconnect(): Do some validity-checking on the remote
721 * address (in mbuf 'nam') and then determine local host address
722 * (i.e., which interface) to use to access that remote host.
724 * This preserves definition of in_pcbconnect(), while supporting a
725 * slightly different version for T/TCP. (This is more than
726 * a bit of a kludge, but cleaning up the internal interfaces would
727 * have forced minor changes in every protocol).
730 in_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
731 struct sockaddr_in
**plocal_sin
, struct thread
*td
)
733 struct in_ifaddr
*ia
;
734 struct ucred
*cred
= NULL
;
735 struct sockaddr_in
*sin
= (struct sockaddr_in
*)nam
;
736 struct sockaddr
*jsin
;
737 int jailed
= 0, alloc_route
= 0;
739 if (nam
->sa_len
!= sizeof *sin
)
741 if (sin
->sin_family
!= AF_INET
)
742 return (EAFNOSUPPORT
);
743 if (sin
->sin_port
== 0)
744 return (EADDRNOTAVAIL
);
745 if (td
&& td
->td_proc
&& td
->td_proc
->p_ucred
)
746 cred
= td
->td_proc
->p_ucred
;
747 if (cred
&& cred
->cr_prison
)
749 if (!TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
])) {
750 ia
= TAILQ_FIRST(&in_ifaddrheads
[mycpuid
])->ia
;
752 * If the destination address is INADDR_ANY,
753 * use the primary local address.
754 * If the supplied address is INADDR_BROADCAST,
755 * and the primary interface supports broadcast,
756 * choose the broadcast address for that interface.
758 if (sin
->sin_addr
.s_addr
== INADDR_ANY
)
759 sin
->sin_addr
= IA_SIN(ia
)->sin_addr
;
760 else if (sin
->sin_addr
.s_addr
== (u_long
)INADDR_BROADCAST
&&
761 (ia
->ia_ifp
->if_flags
& IFF_BROADCAST
))
762 sin
->sin_addr
= satosin(&ia
->ia_broadaddr
)->sin_addr
;
764 if (inp
->inp_laddr
.s_addr
== INADDR_ANY
) {
769 * If route is known or can be allocated now,
770 * our src addr is taken from the i/f, else punt.
771 * Note that we should check the address family of the cached
772 * destination, in case of sharing the cache with IPv6.
774 ro
= &inp
->inp_route
;
776 (!(ro
->ro_rt
->rt_flags
& RTF_UP
) ||
777 ro
->ro_dst
.sa_family
!= AF_INET
||
778 satosin(&ro
->ro_dst
)->sin_addr
.s_addr
!=
779 sin
->sin_addr
.s_addr
||
780 inp
->inp_socket
->so_options
& SO_DONTROUTE
)) {
784 if (!(inp
->inp_socket
->so_options
& SO_DONTROUTE
) && /*XXX*/
785 (ro
->ro_rt
== NULL
||
786 ro
->ro_rt
->rt_ifp
== NULL
)) {
787 /* No route yet, so try to acquire one */
788 bzero(&ro
->ro_dst
, sizeof(struct sockaddr_in
));
789 ro
->ro_dst
.sa_family
= AF_INET
;
790 ro
->ro_dst
.sa_len
= sizeof(struct sockaddr_in
);
791 ((struct sockaddr_in
*) &ro
->ro_dst
)->sin_addr
=
797 * If we found a route, use the address
798 * corresponding to the outgoing interface
799 * unless it is the loopback (in case a route
800 * to our address on another net goes to loopback).
802 if (ro
->ro_rt
&& !(ro
->ro_rt
->rt_ifp
->if_flags
& IFF_LOOPBACK
)) {
804 if (jailed_ip(cred
->cr_prison
,
805 ro
->ro_rt
->rt_ifa
->ifa_addr
)) {
806 ia
= ifatoia(ro
->ro_rt
->rt_ifa
);
809 ia
= ifatoia(ro
->ro_rt
->rt_ifa
);
813 u_short fport
= sin
->sin_port
;
816 ia
= ifatoia(ifa_ifwithdstaddr(sintosa(sin
)));
817 if (ia
&& jailed
&& !jailed_ip(cred
->cr_prison
,
818 sintosa(&ia
->ia_addr
)))
821 ia
= ifatoia(ifa_ifwithnet(sintosa(sin
)));
822 if (ia
&& jailed
&& !jailed_ip(cred
->cr_prison
,
823 sintosa(&ia
->ia_addr
)))
825 sin
->sin_port
= fport
;
827 !TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
]))
828 ia
= TAILQ_FIRST(&in_ifaddrheads
[mycpuid
])->ia
;
829 if (ia
&& jailed
&& !jailed_ip(cred
->cr_prison
,
830 sintosa(&ia
->ia_addr
)))
833 if (!jailed
&& ia
== NULL
)
837 * If the destination address is multicast and an outgoing
838 * interface has been set as a multicast option, use the
839 * address of that interface as our source address.
841 if (!jailed
&& IN_MULTICAST(ntohl(sin
->sin_addr
.s_addr
)) &&
842 inp
->inp_moptions
!= NULL
) {
843 struct ip_moptions
*imo
;
846 imo
= inp
->inp_moptions
;
847 if (imo
->imo_multicast_ifp
!= NULL
) {
848 struct in_ifaddr_container
*iac
;
850 ifp
= imo
->imo_multicast_ifp
;
853 &in_ifaddrheads
[mycpuid
], ia_link
) {
854 if (iac
->ia
->ia_ifp
== ifp
) {
864 * Don't do pcblookup call here; return interface in plocal_sin
865 * and exit to caller, that will do the lookup.
867 if (ia
== NULL
&& jailed
) {
868 if ((jsin
= prison_get_nonlocal(cred
->cr_prison
, AF_INET
, NULL
)) != NULL
||
869 (jsin
= prison_get_local(cred
->cr_prison
, AF_INET
, NULL
)) != NULL
) {
870 *plocal_sin
= satosin(jsin
);
876 *plocal_sin
= &ia
->ia_addr
;
882 struct route
*ro
= &inp
->inp_route
;
884 if (ro
->ro_rt
!= NULL
)
886 bzero(ro
, sizeof(*ro
));
888 return (EADDRNOTAVAIL
);
893 * Connect from a socket to a specified address.
894 * Both address and port must be specified in argument sin.
895 * If don't have a local address for this socket yet,
899 in_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
901 struct sockaddr_in
*if_sin
;
902 struct sockaddr_in
*sin
= (struct sockaddr_in
*)nam
;
905 /* Call inner routine to assign local interface address. */
906 if ((error
= in_pcbladdr(inp
, nam
, &if_sin
, td
)) != 0)
909 if (in_pcblookup_hash(inp
->inp_cpcbinfo
, sin
->sin_addr
, sin
->sin_port
,
910 inp
->inp_laddr
.s_addr
?
911 inp
->inp_laddr
: if_sin
->sin_addr
,
912 inp
->inp_lport
, FALSE
, NULL
) != NULL
) {
915 if (inp
->inp_laddr
.s_addr
== INADDR_ANY
) {
916 if (inp
->inp_lport
== 0) {
917 error
= in_pcbbind(inp
, NULL
, td
);
921 inp
->inp_laddr
= if_sin
->sin_addr
;
923 inp
->inp_faddr
= sin
->sin_addr
;
924 inp
->inp_fport
= sin
->sin_port
;
925 in_pcbinsconnhash(inp
);
930 in_pcbdisconnect(struct inpcb
*inp
)
933 inp
->inp_faddr
.s_addr
= INADDR_ANY
;
935 in_pcbremconnhash(inp
);
936 if (inp
->inp_socket
->so_state
& SS_NOFDREF
)
941 in_pcbdetach(struct inpcb
*inp
)
943 struct socket
*so
= inp
->inp_socket
;
944 struct inpcbinfo
*ipi
= inp
->inp_pcbinfo
;
947 ipsec4_delete_pcbpolicy(inp
);
949 inp
->inp_gencnt
= ++ipi
->ipi_gencnt
;
950 KKASSERT((so
->so_state
& SS_ASSERTINPROG
) == 0);
953 sofree(so
); /* remove pcb ref */
954 if (inp
->inp_options
)
955 m_free(inp
->inp_options
);
956 if (inp
->inp_route
.ro_rt
)
957 rtfree(inp
->inp_route
.ro_rt
);
958 ip_freemoptions(inp
->inp_moptions
);
964 * The calling convention of in_setsockaddr() and in_setpeeraddr() was
965 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
966 * in struct pr_usrreqs, so that protocols can just reference then directly
967 * without the need for a wrapper function. The socket must have a valid
968 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
969 * except through a kernel programming error, so it is acceptable to panic
970 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
971 * because there actually /is/ a programming error somewhere... XXX)
974 in_setsockaddr(struct socket
*so
, struct sockaddr
**nam
)
977 struct sockaddr_in
*sin
;
980 * Do the malloc first in case it blocks.
982 sin
= kmalloc(sizeof *sin
, M_SONAME
, M_WAITOK
| M_ZERO
);
983 sin
->sin_family
= AF_INET
;
984 sin
->sin_len
= sizeof *sin
;
990 kfree(sin
, M_SONAME
);
993 sin
->sin_port
= inp
->inp_lport
;
994 sin
->sin_addr
= inp
->inp_laddr
;
997 *nam
= (struct sockaddr
*)sin
;
1002 in_setsockaddr_dispatch(netmsg_t msg
)
1006 error
= in_setsockaddr(msg
->base
.nm_so
, msg
->peeraddr
.nm_nam
);
1007 lwkt_replymsg(&msg
->lmsg
, error
);
1011 in_setpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
1014 struct sockaddr_in
*sin
;
1017 * Do the malloc first in case it blocks.
1019 sin
= kmalloc(sizeof *sin
, M_SONAME
, M_WAITOK
| M_ZERO
);
1020 sin
->sin_family
= AF_INET
;
1021 sin
->sin_len
= sizeof *sin
;
1027 kfree(sin
, M_SONAME
);
1028 return (ECONNRESET
);
1030 sin
->sin_port
= inp
->inp_fport
;
1031 sin
->sin_addr
= inp
->inp_faddr
;
1034 *nam
= (struct sockaddr
*)sin
;
1039 in_setpeeraddr_dispatch(netmsg_t msg
)
1043 error
= in_setpeeraddr(msg
->base
.nm_so
, msg
->peeraddr
.nm_nam
);
1044 lwkt_replymsg(&msg
->lmsg
, error
);
1048 in_pcbnotifyall(struct inpcbhead
*head
, struct in_addr faddr
, int err
,
1049 void (*notify
)(struct inpcb
*, int))
1051 struct inpcb
*inp
, *ninp
;
1054 * note: if INP_PLACEMARKER is set we must ignore the rest of
1055 * the structure and skip it.
1058 LIST_FOREACH_MUTABLE(inp
, head
, inp_list
, ninp
) {
1059 if (inp
->inp_flags
& INP_PLACEMARKER
)
1062 if (!(inp
->inp_vflag
& INP_IPV4
))
1065 if (inp
->inp_faddr
.s_addr
!= faddr
.s_addr
||
1066 inp
->inp_socket
== NULL
)
1068 (*notify
)(inp
, err
); /* can remove inp from list! */
1074 in_pcbpurgeif0(struct inpcb
*head
, struct ifnet
*ifp
)
1077 struct ip_moptions
*imo
;
1080 for (inp
= head
; inp
!= NULL
; inp
= LIST_NEXT(inp
, inp_list
)) {
1081 if (inp
->inp_flags
& INP_PLACEMARKER
)
1083 imo
= inp
->inp_moptions
;
1084 if ((inp
->inp_vflag
& INP_IPV4
) && imo
!= NULL
) {
1086 * Unselect the outgoing interface if it is being
1089 if (imo
->imo_multicast_ifp
== ifp
)
1090 imo
->imo_multicast_ifp
= NULL
;
1093 * Drop multicast group membership if we joined
1094 * through the interface being detached.
1096 for (i
= 0, gap
= 0; i
< imo
->imo_num_memberships
;
1098 if (imo
->imo_membership
[i
]->inm_ifp
== ifp
) {
1099 in_delmulti(imo
->imo_membership
[i
]);
1101 } else if (gap
!= 0)
1102 imo
->imo_membership
[i
- gap
] =
1103 imo
->imo_membership
[i
];
1105 imo
->imo_num_memberships
-= gap
;
1111 * Check for alternatives when higher level complains
1112 * about service problems. For now, invalidate cached
1113 * routing information. If the route was created dynamically
1114 * (by a redirect), time to try a default gateway again.
1117 in_losing(struct inpcb
*inp
)
1120 struct rt_addrinfo rtinfo
;
1122 if ((rt
= inp
->inp_route
.ro_rt
)) {
1123 bzero(&rtinfo
, sizeof(struct rt_addrinfo
));
1124 rtinfo
.rti_info
[RTAX_DST
] = rt_key(rt
);
1125 rtinfo
.rti_info
[RTAX_GATEWAY
] = rt
->rt_gateway
;
1126 rtinfo
.rti_info
[RTAX_NETMASK
] = rt_mask(rt
);
1127 rtinfo
.rti_flags
= rt
->rt_flags
;
1128 rt_missmsg(RTM_LOSING
, &rtinfo
, rt
->rt_flags
, 0);
1129 if (rt
->rt_flags
& RTF_DYNAMIC
)
1130 rtrequest1_global(RTM_DELETE
, &rtinfo
, NULL
, NULL
);
1131 inp
->inp_route
.ro_rt
= NULL
;
1134 * A new route can be allocated
1135 * the next time output is attempted.
1141 * After a routing change, flush old routing
1142 * and allocate a (hopefully) better one.
1145 in_rtchange(struct inpcb
*inp
, int err
)
1147 if (inp
->inp_route
.ro_rt
) {
1148 rtfree(inp
->inp_route
.ro_rt
);
1149 inp
->inp_route
.ro_rt
= NULL
;
1151 * A new route can be allocated the next time
1152 * output is attempted.
1158 * Lookup a PCB based on the local address and port.
1161 in_pcblookup_local(struct inpcbinfo
*pcbinfo
, struct in_addr laddr
,
1162 u_int lport_arg
, int wild_okay
, struct ucred
*cred
)
1165 int matchwild
= 3, wildcard
;
1166 u_short lport
= lport_arg
;
1167 struct inpcbporthead
*porthash
;
1168 struct inpcbport
*phd
;
1169 struct inpcb
*match
= NULL
;
1172 * If the porthashbase is shared across several cpus we need
1175 if (pcbinfo
->porttoken
)
1176 lwkt_gettoken(pcbinfo
->porttoken
);
1179 * Best fit PCB lookup.
1181 * First see if this local port is in use by looking on the
1184 porthash
= &pcbinfo
->porthashbase
[
1185 INP_PCBPORTHASH(lport
, pcbinfo
->porthashmask
)];
1186 LIST_FOREACH(phd
, porthash
, phd_hash
) {
1187 if (phd
->phd_port
== lport
)
1192 * Port is in use by one or more PCBs. Look for best
1195 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
1198 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
1201 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
)
1203 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
) {
1204 if (laddr
.s_addr
== INADDR_ANY
)
1206 else if (inp
->inp_laddr
.s_addr
!= laddr
.s_addr
)
1209 if (laddr
.s_addr
!= INADDR_ANY
)
1212 if (wildcard
&& !wild_okay
)
1214 if (wildcard
< matchwild
&&
1217 inp
->inp_socket
->so_cred
->cr_prison
)) {
1219 matchwild
= wildcard
;
1220 if (matchwild
== 0) {
1226 if (pcbinfo
->porttoken
)
1227 lwkt_reltoken(pcbinfo
->porttoken
);
1232 * Lookup PCB in hash list.
1235 in_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in_addr faddr
,
1236 u_int fport_arg
, struct in_addr laddr
, u_int lport_arg
,
1237 boolean_t wildcard
, struct ifnet
*ifp
)
1239 struct inpcbhead
*head
;
1240 struct inpcb
*inp
, *jinp
=NULL
;
1241 u_short fport
= fport_arg
, lport
= lport_arg
;
1244 * First look for an exact match.
1246 head
= &pcbinfo
->hashbase
[INP_PCBCONNHASH(faddr
.s_addr
, fport
,
1247 laddr
.s_addr
, lport
, pcbinfo
->hashmask
)];
1248 LIST_FOREACH(inp
, head
, inp_hash
) {
1250 if (!(inp
->inp_vflag
& INP_IPV4
))
1253 if (in_hosteq(inp
->inp_faddr
, faddr
) &&
1254 in_hosteq(inp
->inp_laddr
, laddr
) &&
1255 inp
->inp_fport
== fport
&& inp
->inp_lport
== lport
) {
1257 if (inp
->inp_socket
== NULL
||
1258 inp
->inp_socket
->so_cred
->cr_prison
== NULL
) {
1269 struct inpcb
*local_wild
= NULL
;
1270 struct inpcb
*jinp_wild
= NULL
;
1272 struct inpcb
*local_wild_mapped
= NULL
;
1274 struct inpcontainer
*ic
;
1275 struct inpcontainerhead
*chead
;
1276 struct sockaddr_in jsin
;
1280 * Order of socket selection:
1281 * 1. non-jailed, non-wild.
1282 * 2. non-jailed, wild.
1283 * 3. jailed, non-wild.
1286 jsin
.sin_family
= AF_INET
;
1287 chead
= &pcbinfo
->wildcardhashbase
[
1288 INP_PCBWILDCARDHASH(lport
, pcbinfo
->wildcardhashmask
)];
1289 LIST_FOREACH(ic
, chead
, ic_list
) {
1291 jsin
.sin_addr
.s_addr
= laddr
.s_addr
;
1293 if (!(inp
->inp_vflag
& INP_IPV4
))
1296 if (inp
->inp_socket
!= NULL
)
1297 cred
= inp
->inp_socket
->so_cred
;
1300 if (cred
!= NULL
&& jailed(cred
)) {
1304 if (!jailed_ip(cred
->cr_prison
,
1305 (struct sockaddr
*)&jsin
))
1308 if (inp
->inp_lport
== lport
) {
1309 if (ifp
&& ifp
->if_type
== IFT_FAITH
&&
1310 !(inp
->inp_flags
& INP_FAITH
))
1312 if (inp
->inp_laddr
.s_addr
== laddr
.s_addr
) {
1313 if (cred
!= NULL
&& jailed(cred
))
1318 if (inp
->inp_laddr
.s_addr
== INADDR_ANY
) {
1320 if (INP_CHECK_SOCKAF(inp
->inp_socket
,
1322 local_wild_mapped
= inp
;
1333 if (local_wild
!= NULL
)
1334 return (local_wild
);
1336 if (local_wild_mapped
!= NULL
)
1337 return (local_wild_mapped
);
1351 * Insert PCB into connection hash table.
1354 in_pcbinsconnhash(struct inpcb
*inp
)
1356 struct inpcbinfo
*pcbinfo
= inp
->inp_cpcbinfo
;
1357 struct inpcbhead
*bucket
;
1358 u_int32_t hashkey_faddr
, hashkey_laddr
;
1361 if (inp
->inp_vflag
& INP_IPV6
) {
1362 hashkey_faddr
= inp
->in6p_faddr
.s6_addr32
[3] /* XXX JH */;
1363 hashkey_laddr
= inp
->in6p_laddr
.s6_addr32
[3] /* XXX JH */;
1366 hashkey_faddr
= inp
->inp_faddr
.s_addr
;
1367 hashkey_laddr
= inp
->inp_laddr
.s_addr
;
1372 KASSERT(!(inp
->inp_flags
& INP_WILDCARD
),
1373 ("already on wildcardhash\n"));
1374 KASSERT(!(inp
->inp_flags
& INP_CONNECTED
),
1375 ("already on connhash\n"));
1376 inp
->inp_flags
|= INP_CONNECTED
;
1379 * Insert into the connection hash table.
1381 bucket
= &pcbinfo
->hashbase
[INP_PCBCONNHASH(hashkey_faddr
,
1382 inp
->inp_fport
, hashkey_laddr
, inp
->inp_lport
, pcbinfo
->hashmask
)];
1383 LIST_INSERT_HEAD(bucket
, inp
, inp_hash
);
1387 * Remove PCB from connection hash table.
1390 in_pcbremconnhash(struct inpcb
*inp
)
1392 KASSERT(inp
->inp_flags
& INP_CONNECTED
, ("inp not connected"));
1393 LIST_REMOVE(inp
, inp_hash
);
1394 inp
->inp_flags
&= ~INP_CONNECTED
;
1398 * Insert PCB into port hash table.
1401 in_pcbinsporthash(struct inpcb
*inp
)
1403 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
1404 struct inpcbporthead
*pcbporthash
;
1405 struct inpcbport
*phd
;
1408 * If the porthashbase is shared across several cpus we need
1411 if (pcbinfo
->porttoken
)
1412 lwkt_gettoken(pcbinfo
->porttoken
);
1415 * Insert into the port hash table.
1417 pcbporthash
= &pcbinfo
->porthashbase
[
1418 INP_PCBPORTHASH(inp
->inp_lport
, pcbinfo
->porthashmask
)];
1420 /* Go through port list and look for a head for this lport. */
1421 LIST_FOREACH(phd
, pcbporthash
, phd_hash
) {
1422 if (phd
->phd_port
== inp
->inp_lport
)
1426 /* If none exists, malloc one and tack it on. */
1428 KKASSERT(pcbinfo
->portsave
!= NULL
);
1429 phd
= pcbinfo
->portsave
;
1430 pcbinfo
->portsave
= NULL
;
1431 phd
->phd_port
= inp
->inp_lport
;
1432 LIST_INIT(&phd
->phd_pcblist
);
1433 LIST_INSERT_HEAD(pcbporthash
, phd
, phd_hash
);
1437 LIST_INSERT_HEAD(&phd
->phd_pcblist
, inp
, inp_portlist
);
1439 if (pcbinfo
->porttoken
)
1440 lwkt_reltoken(pcbinfo
->porttoken
);
1441 if (pcbinfo
->portsave
== NULL
) {
1442 pcbinfo
->portsave
= kmalloc(sizeof(*pcbinfo
->portsave
),
1443 M_PCB
, M_INTWAIT
| M_ZERO
);
1449 in_pcbinswildcardhash_oncpu(struct inpcb
*inp
, struct inpcbinfo
*pcbinfo
)
1451 struct inpcontainer
*ic
;
1452 struct inpcontainerhead
*bucket
;
1454 bucket
= &pcbinfo
->wildcardhashbase
[
1455 INP_PCBWILDCARDHASH(inp
->inp_lport
, pcbinfo
->wildcardhashmask
)];
1457 ic
= kmalloc(sizeof(struct inpcontainer
), M_TEMP
, M_INTWAIT
);
1459 LIST_INSERT_HEAD(bucket
, ic
, ic_list
);
1463 * Insert PCB into wildcard hash table.
1466 in_pcbinswildcardhash(struct inpcb
*inp
)
1468 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
1470 KASSERT(!(inp
->inp_flags
& INP_CONNECTED
),
1471 ("already on connhash\n"));
1472 KASSERT(!(inp
->inp_flags
& INP_WILDCARD
),
1473 ("already on wildcardhash\n"));
1474 inp
->inp_flags
|= INP_WILDCARD
;
1476 in_pcbinswildcardhash_oncpu(inp
, pcbinfo
);
1480 in_pcbremwildcardhash_oncpu(struct inpcb
*inp
, struct inpcbinfo
*pcbinfo
)
1482 struct inpcontainer
*ic
;
1483 struct inpcontainerhead
*head
;
1486 head
= &pcbinfo
->wildcardhashbase
[
1487 INP_PCBWILDCARDHASH(inp
->inp_lport
, pcbinfo
->wildcardhashmask
)];
1489 LIST_FOREACH(ic
, head
, ic_list
) {
1490 if (ic
->ic_inp
== inp
)
1493 return; /* not found! */
1496 LIST_REMOVE(ic
, ic_list
); /* remove container from bucket chain */
1497 kfree(ic
, M_TEMP
); /* deallocate container */
1501 * Remove PCB from wildcard hash table.
1504 in_pcbremwildcardhash(struct inpcb
*inp
)
1506 struct inpcbinfo
*pcbinfo
= inp
->inp_pcbinfo
;
1508 KASSERT(inp
->inp_flags
& INP_WILDCARD
, ("inp not wildcard"));
1509 in_pcbremwildcardhash_oncpu(inp
, pcbinfo
);
1510 inp
->inp_flags
&= ~INP_WILDCARD
;
1514 * Remove PCB from various lists.
1517 in_pcbremlists(struct inpcb
*inp
)
1519 struct inpcbinfo
*pcbinfo
;
1521 if (inp
->inp_lport
) {
1522 struct inpcbport
*phd
;
1524 pcbinfo
= inp
->inp_pcbinfo
;
1525 if (pcbinfo
->porttoken
)
1526 lwkt_gettoken(pcbinfo
->porttoken
);
1529 LIST_REMOVE(inp
, inp_portlist
);
1530 if (LIST_FIRST(&phd
->phd_pcblist
) == NULL
) {
1531 LIST_REMOVE(phd
, phd_hash
);
1534 if (pcbinfo
->porttoken
)
1535 lwkt_reltoken(pcbinfo
->porttoken
);
1537 if (inp
->inp_flags
& INP_WILDCARD
) {
1538 in_pcbremwildcardhash(inp
);
1539 } else if (inp
->inp_flags
& INP_CONNECTED
) {
1540 in_pcbremconnhash(inp
);
1542 LIST_REMOVE(inp
, inp_list
);
1543 inp
->inp_pcbinfo
->ipi_count
--;
1547 prison_xinpcb(struct thread
*td
, struct inpcb
*inp
)
1551 if (td
->td_proc
== NULL
)
1553 cr
= td
->td_proc
->p_ucred
;
1554 if (cr
->cr_prison
== NULL
)
1556 if (inp
->inp_socket
&& inp
->inp_socket
->so_cred
&&
1557 inp
->inp_socket
->so_cred
->cr_prison
&&
1558 cr
->cr_prison
== inp
->inp_socket
->so_cred
->cr_prison
)
1564 in_pcblist_global(SYSCTL_HANDLER_ARGS
)
1566 struct inpcbinfo
*pcbinfo
= arg1
;
1567 struct inpcb
*inp
, *marker
;
1572 * The process of preparing the TCB list is too time-consuming and
1573 * resource-intensive to repeat twice on every request.
1575 if (req
->oldptr
== NULL
) {
1576 n
= pcbinfo
->ipi_count
;
1577 req
->oldidx
= (n
+ n
/8 + 10) * sizeof(struct xinpcb
);
1581 if (req
->newptr
!= NULL
)
1585 * OK, now we're committed to doing something. Re-fetch ipi_count
1586 * after obtaining the generation count.
1588 n
= pcbinfo
->ipi_count
;
1590 marker
= kmalloc(sizeof(struct inpcb
), M_TEMP
, M_WAITOK
|M_ZERO
);
1591 marker
->inp_flags
|= INP_PLACEMARKER
;
1592 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, marker
, inp_list
);
1597 while ((inp
= LIST_NEXT(marker
, inp_list
)) != NULL
&& i
< n
) {
1598 LIST_REMOVE(marker
, inp_list
);
1599 LIST_INSERT_AFTER(inp
, marker
, inp_list
);
1601 if (inp
->inp_flags
& INP_PLACEMARKER
)
1603 if (prison_xinpcb(req
->td
, inp
))
1605 bzero(&xi
, sizeof xi
);
1606 xi
.xi_len
= sizeof xi
;
1607 bcopy(inp
, &xi
.xi_inp
, sizeof *inp
);
1608 if (inp
->inp_socket
)
1609 sotoxsocket(inp
->inp_socket
, &xi
.xi_socket
);
1610 if ((error
= SYSCTL_OUT(req
, &xi
, sizeof xi
)) != 0)
1614 LIST_REMOVE(marker
, inp_list
);
1615 if (error
== 0 && i
< n
) {
1616 bzero(&xi
, sizeof xi
);
1617 xi
.xi_len
= sizeof xi
;
1619 error
= SYSCTL_OUT(req
, &xi
, sizeof xi
);
1623 kfree(marker
, M_TEMP
);
1628 in_pcblist_global_nomarker(SYSCTL_HANDLER_ARGS
, struct xinpcb
**xi0
, int *nxi0
)
1630 struct inpcbinfo
*pcbinfo
= arg1
;
1639 * The process of preparing the PCB list is too time-consuming and
1640 * resource-intensive to repeat twice on every request.
1642 if (req
->oldptr
== NULL
) {
1643 int n
= pcbinfo
->ipi_count
;
1645 req
->oldidx
= (n
+ n
/8 + 10) * sizeof(struct xinpcb
);
1649 if (req
->newptr
!= NULL
)
1652 if (pcbinfo
->ipi_count
== 0)
1656 xi
= kmalloc(pcbinfo
->ipi_count
* sizeof(*xi
), M_TEMP
,
1657 M_WAITOK
| M_ZERO
| M_NULLOK
);
1661 LIST_FOREACH(inp
, &pcbinfo
->pcblisthead
, inp_list
) {
1662 struct xinpcb
*xi_ptr
= &xi
[nxi
];
1664 if (prison_xinpcb(req
->td
, inp
))
1667 xi_ptr
->xi_len
= sizeof(*xi_ptr
);
1668 bcopy(inp
, &xi_ptr
->xi_inp
, sizeof(*inp
));
1669 if (inp
->inp_socket
)
1670 sotoxsocket(inp
->inp_socket
, &xi_ptr
->xi_socket
);
1686 in_savefaddr(struct socket
*so
, const struct sockaddr
*faddr
)
1688 struct sockaddr_in
*sin
;
1690 KASSERT(faddr
->sa_family
== AF_INET
,
1691 ("not AF_INET faddr %d\n", faddr
->sa_family
));
1693 sin
= kmalloc(sizeof(*sin
), M_SONAME
, M_WAITOK
| M_ZERO
);
1694 sin
->sin_family
= AF_INET
;
1695 sin
->sin_len
= sizeof(*sin
);
1696 sin
->sin_port
= ((const struct sockaddr_in
*)faddr
)->sin_port
;
1697 sin
->sin_addr
= ((const struct sockaddr_in
*)faddr
)->sin_addr
;
1699 so
->so_faddr
= (struct sockaddr
*)sin
;