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) 2004 Jeffrey M. Hsu. All rights reserved.
37 * License terms: all terms for the DragonFly license above plus the following:
39 * 4. All advertising materials mentioning features or use of this software
40 * must display the following acknowledgement:
42 * This product includes software developed by Jeffrey M. Hsu
43 * for the DragonFly Project.
45 * This requirement may be waived with permission from Jeffrey Hsu.
46 * This requirement will sunset and may be removed on July 8 2005,
47 * after which the standard DragonFly license (as shown above) will
52 * Copyright (c) 1982, 1986, 1991, 1993, 1995
53 * The Regents of the University of California. All rights reserved.
55 * Redistribution and use in source and binary forms, with or without
56 * modification, are permitted provided that the following conditions
58 * 1. Redistributions of source code must retain the above copyright
59 * notice, this list of conditions and the following disclaimer.
60 * 2. Redistributions in binary form must reproduce the above copyright
61 * notice, this list of conditions and the following disclaimer in the
62 * documentation and/or other materials provided with the distribution.
63 * 3. All advertising materials mentioning features or use of this software
64 * must display the following acknowledgement:
65 * This product includes software developed by the University of
66 * California, Berkeley and its contributors.
67 * 4. Neither the name of the University nor the names of its contributors
68 * may be used to endorse or promote products derived from this software
69 * without specific prior written permission.
71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
83 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95
84 * $FreeBSD: src/sys/netinet/in_pcb.c,v 1.59.2.27 2004/01/02 04:06:42 ambrisko Exp $
85 * $DragonFly: src/sys/netinet/in_pcb.c,v 1.42 2007/01/18 12:34:46 victor Exp $
88 #include "opt_ipsec.h"
89 #include "opt_inet6.h"
91 #include <sys/param.h>
92 #include <sys/systm.h>
93 #include <sys/malloc.h>
95 #include <sys/domain.h>
96 #include <sys/protosw.h>
97 #include <sys/socket.h>
98 #include <sys/socketvar.h>
100 #include <sys/jail.h>
101 #include <sys/kernel.h>
102 #include <sys/sysctl.h>
103 #include <sys/thread2.h>
105 #include <machine/limits.h>
107 #include <vm/vm_zone.h>
110 #include <net/if_types.h>
111 #include <net/route.h>
113 #include <netinet/in.h>
114 #include <netinet/in_pcb.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip_var.h>
118 #include <netinet/ip6.h>
119 #include <netinet6/ip6_var.h>
123 #include <netinet6/ipsec.h>
124 #include <netproto/key/key.h>
128 #if defined(IPSEC) || defined(IPSEC_ESP)
129 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!"
132 #include <netproto/ipsec/ipsec.h>
133 #include <netproto/ipsec/key.h>
135 #endif /* FAST_IPSEC */
137 struct in_addr zeroin_addr;
140 * These configure the range of local port addresses assigned to
141 * "unspecified" outgoing connections/packets/whatever.
143 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */
144 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */
146 int ipport_firstauto = IPPORT_RESERVED; /* 1024 */
147 int ipport_lastauto = IPPORT_USERRESERVED; /* 5000 */
149 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */
150 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */
153 RANGECHK(int var, int min, int max)
162 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS)
166 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req);
168 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1);
169 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1);
171 RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX);
172 RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX);
174 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX);
175 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX);
180 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports");
182 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW,
183 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", "");
184 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW,
185 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", "");
186 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW,
187 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", "");
188 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW,
189 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", "");
190 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW,
191 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", "");
192 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW,
193 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", "");
196 * in_pcb.c: manage the Protocol Control Blocks.
198 * NOTE: It is assumed that most of these functions will be called from
199 * a critical section. XXX - There are, unfortunately, a few exceptions
200 * to this rule that should be fixed.
202 * NOTE: The caller should initialize the cpu field to the cpu running the
203 * protocol stack associated with this inpcbinfo.
207 in_pcbinfo_init(struct inpcbinfo *pcbinfo)
209 LIST_INIT(&pcbinfo->pcblisthead);
214 * Allocate a PCB and associate it with the socket.
217 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo)
224 inp = zalloc(pcbinfo->ipi_zone);
227 bzero(inp, sizeof *inp);
228 inp->inp_gencnt = ++pcbinfo->ipi_gencnt;
229 inp->inp_pcbinfo = inp->inp_cpcbinfo = pcbinfo;
230 inp->inp_socket = so;
232 error = ipsec_init_policy(so, &inp->inp_sp);
234 zfree(pcbinfo->ipi_zone, inp);
239 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only)
240 inp->inp_flags |= IN6P_IPV6_V6ONLY;
241 if (ip6_auto_flowlabel)
242 inp->inp_flags |= IN6P_AUTOFLOWLABEL;
245 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list);
246 pcbinfo->ipi_count++;
251 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
253 struct socket *so = inp->inp_socket;
254 struct proc *p = td->td_proc;
255 unsigned short *lastport;
256 struct sockaddr_in *sin;
257 struct sockaddr_in jsin;
258 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
259 struct ucred *cred = NULL;
261 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
266 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */
267 return (EADDRNOTAVAIL);
268 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY)
269 return (EINVAL); /* already bound */
270 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT)))
271 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */
275 sin = (struct sockaddr_in *)nam;
276 if (nam->sa_len != sizeof *sin)
280 * We should check the family, but old programs
281 * incorrectly fail to initialize it.
283 if (sin->sin_family != AF_INET)
284 return (EAFNOSUPPORT);
286 if (!prison_replace_wildcards(td, nam))
288 lport = sin->sin_port;
289 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
291 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
292 * allow complete duplication of binding if
293 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
294 * and a multicast address is bound on both
295 * new and duplicated sockets.
297 if (so->so_options & SO_REUSEADDR)
298 reuseport = SO_REUSEADDR | SO_REUSEPORT;
299 } else if (sin->sin_addr.s_addr != INADDR_ANY) {
300 sin->sin_port = 0; /* yech... */
301 bzero(&sin->sin_zero, sizeof sin->sin_zero);
302 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL)
303 return (EADDRNOTAVAIL);
309 if (ntohs(lport) < IPPORT_RESERVED &&
310 cred && suser_cred(cred, PRISON_ROOT))
312 if (so->so_cred->cr_uid != 0 &&
313 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
314 t = in_pcblookup_local(inp->inp_pcbinfo,
315 sin->sin_addr, lport,
316 INPLOOKUP_WILDCARD, cred);
318 (!in_nullhost(sin->sin_addr) ||
319 !in_nullhost(t->inp_laddr) ||
320 (t->inp_socket->so_options &
321 SO_REUSEPORT) == 0) &&
322 (so->so_cred->cr_uid !=
323 t->inp_socket->so_cred->cr_uid)) {
325 if (!in_nullhost(sin->sin_addr) ||
326 !in_nullhost(t->inp_laddr) ||
328 INP_SOCKAF(t->inp_socket))
333 if (cred && !prison_replace_wildcards(td, nam))
334 return (EADDRNOTAVAIL);
335 t = in_pcblookup_local(pcbinfo, sin->sin_addr, lport,
337 if (t && !(reuseport & t->inp_socket->so_options)) {
339 if (!in_nullhost(sin->sin_addr) ||
340 !in_nullhost(t->inp_laddr) ||
341 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))
346 inp->inp_laddr = sin->sin_addr;
352 jsin.sin_family = AF_INET;
353 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr;
354 if (!prison_replace_wildcards(td, (struct sockaddr *)&jsin)) {
355 inp->inp_laddr.s_addr = INADDR_ANY;
358 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr;
360 inp->inp_flags |= INP_ANONPORT;
362 if (inp->inp_flags & INP_HIGHPORT) {
363 first = ipport_hifirstauto; /* sysctl */
364 last = ipport_hilastauto;
365 lastport = &pcbinfo->lasthi;
366 } else if (inp->inp_flags & INP_LOWPORT) {
368 (error = suser_cred(cred, PRISON_ROOT))) {
369 inp->inp_laddr.s_addr = INADDR_ANY;
372 first = ipport_lowfirstauto; /* 1023 */
373 last = ipport_lowlastauto; /* 600 */
374 lastport = &pcbinfo->lastlow;
376 first = ipport_firstauto; /* sysctl */
377 last = ipport_lastauto;
378 lastport = &pcbinfo->lastport;
381 * Simple check to ensure all ports are not used up causing
384 * We split the two cases (up and down) so that the direction
385 * is not being tested on each round of the loop.
391 count = first - last;
394 if (count-- < 0) { /* completely used? */
395 inp->inp_laddr.s_addr = INADDR_ANY;
396 return (EADDRNOTAVAIL);
399 if (*lastport > first || *lastport < last)
401 lport = htons(*lastport);
402 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr,
408 count = last - first;
411 if (count-- < 0) { /* completely used? */
412 inp->inp_laddr.s_addr = INADDR_ANY;
413 return (EADDRNOTAVAIL);
416 if (*lastport < first || *lastport > last)
418 lport = htons(*lastport);
419 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr,
423 inp->inp_lport = lport;
425 jsin.sin_family = AF_INET;
426 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr;
427 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin)) {
428 inp->inp_laddr.s_addr = INADDR_ANY;
432 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr;
434 if (in_pcbinsporthash(inp) != 0) {
435 inp->inp_laddr.s_addr = INADDR_ANY;
443 * Transform old in_pcbconnect() into an inner subroutine for new
444 * in_pcbconnect(): Do some validity-checking on the remote
445 * address (in mbuf 'nam') and then determine local host address
446 * (i.e., which interface) to use to access that remote host.
448 * This preserves definition of in_pcbconnect(), while supporting a
449 * slightly different version for T/TCP. (This is more than
450 * a bit of a kludge, but cleaning up the internal interfaces would
451 * have forced minor changes in every protocol).
454 in_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
455 struct sockaddr_in **plocal_sin, struct thread *td)
457 struct in_ifaddr *ia;
458 struct ucred *cred = NULL;
459 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
460 struct sockaddr *jsin;
463 if (nam->sa_len != sizeof *sin)
465 if (sin->sin_family != AF_INET)
466 return (EAFNOSUPPORT);
467 if (sin->sin_port == 0)
468 return (EADDRNOTAVAIL);
469 if (td && td->td_proc && td->td_proc->p_ucred)
470 cred = td->td_proc->p_ucred;
471 if (cred && cred->cr_prison)
473 if (!TAILQ_EMPTY(&in_ifaddrhead)) {
474 ia = TAILQ_FIRST(&in_ifaddrhead);
476 * If the destination address is INADDR_ANY,
477 * use the primary local address.
478 * If the supplied address is INADDR_BROADCAST,
479 * and the primary interface supports broadcast,
480 * choose the broadcast address for that interface.
482 if (sin->sin_addr.s_addr == INADDR_ANY)
483 sin->sin_addr = IA_SIN(ia)->sin_addr;
484 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST &&
485 (ia->ia_ifp->if_flags & IFF_BROADCAST))
486 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr;
488 if (inp->inp_laddr.s_addr == INADDR_ANY) {
491 ia = (struct in_ifaddr *)NULL;
493 * If route is known or can be allocated now,
494 * our src addr is taken from the i/f, else punt.
495 * Note that we should check the address family of the cached
496 * destination, in case of sharing the cache with IPv6.
498 ro = &inp->inp_route;
500 (!(ro->ro_rt->rt_flags & RTF_UP) ||
501 ro->ro_dst.sa_family != AF_INET ||
502 satosin(&ro->ro_dst)->sin_addr.s_addr !=
503 sin->sin_addr.s_addr ||
504 inp->inp_socket->so_options & SO_DONTROUTE)) {
506 ro->ro_rt = (struct rtentry *)NULL;
508 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/
509 (ro->ro_rt == (struct rtentry *)NULL ||
510 ro->ro_rt->rt_ifp == (struct ifnet *)NULL)) {
511 /* No route yet, so try to acquire one */
512 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
513 ro->ro_dst.sa_family = AF_INET;
514 ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
515 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
520 * If we found a route, use the address
521 * corresponding to the outgoing interface
522 * unless it is the loopback (in case a route
523 * to our address on another net goes to loopback).
525 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) {
527 if (jailed_ip(cred->cr_prison,
528 ro->ro_rt->rt_ifa->ifa_addr)) {
529 ia = ifatoia(ro->ro_rt->rt_ifa);
532 ia = ifatoia(ro->ro_rt->rt_ifa);
536 u_short fport = sin->sin_port;
539 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin)));
540 if (ia && jailed && !jailed_ip(cred->cr_prison,
541 sintosa(&ia->ia_addr)))
544 ia = ifatoia(ifa_ifwithnet(sintosa(sin)));
545 if (ia && jailed && !jailed_ip(cred->cr_prison,
546 sintosa(&ia->ia_addr)))
548 sin->sin_port = fport;
550 ia = TAILQ_FIRST(&in_ifaddrhead);
551 if (ia && jailed && !jailed_ip(cred->cr_prison,
552 sintosa(&ia->ia_addr)))
555 if (!jailed && ia == NULL)
556 return (EADDRNOTAVAIL);
559 * If the destination address is multicast and an outgoing
560 * interface has been set as a multicast option, use the
561 * address of that interface as our source address.
563 if (!jailed && IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) &&
564 inp->inp_moptions != NULL) {
565 struct ip_moptions *imo;
568 imo = inp->inp_moptions;
569 if (imo->imo_multicast_ifp != NULL) {
570 ifp = imo->imo_multicast_ifp;
571 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link)
572 if (ia->ia_ifp == ifp)
575 return (EADDRNOTAVAIL);
579 * Don't do pcblookup call here; return interface in plocal_sin
580 * and exit to caller, that will do the lookup.
582 if (ia == NULL && jailed) {
583 if ((jsin = prison_get_nonlocal(cred->cr_prison, AF_INET, NULL)) != NULL ||
584 (jsin = prison_get_local(cred->cr_prison, AF_INET, NULL)) != NULL)
585 *plocal_sin = satosin(jsin);
588 return (EADDRNOTAVAIL);
590 *plocal_sin = &ia->ia_addr;
598 * Connect from a socket to a specified address.
599 * Both address and port must be specified in argument sin.
600 * If don't have a local address for this socket yet,
604 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
606 struct sockaddr_in *if_sin;
607 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
610 /* Call inner routine to assign local interface address. */
611 if ((error = in_pcbladdr(inp, nam, &if_sin, td)) != 0)
614 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port,
615 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr,
616 inp->inp_lport, FALSE, NULL) != NULL) {
619 if (inp->inp_laddr.s_addr == INADDR_ANY) {
620 if (inp->inp_lport == 0) {
621 error = in_pcbbind(inp, (struct sockaddr *)NULL, td);
625 inp->inp_laddr = if_sin->sin_addr;
627 inp->inp_faddr = sin->sin_addr;
628 inp->inp_fport = sin->sin_port;
629 in_pcbinsconnhash(inp);
634 in_pcbdisconnect(struct inpcb *inp)
637 inp->inp_faddr.s_addr = INADDR_ANY;
639 in_pcbremconnhash(inp);
640 if (inp->inp_socket->so_state & SS_NOFDREF)
645 in_pcbdetach(struct inpcb *inp)
647 struct socket *so = inp->inp_socket;
648 struct inpcbinfo *ipi = inp->inp_pcbinfo;
651 ipsec4_delete_pcbpolicy(inp);
653 inp->inp_gencnt = ++ipi->ipi_gencnt;
657 if (inp->inp_options)
658 m_free(inp->inp_options);
659 if (inp->inp_route.ro_rt)
660 rtfree(inp->inp_route.ro_rt);
661 ip_freemoptions(inp->inp_moptions);
663 zfree(ipi->ipi_zone, inp);
667 * The calling convention of in_setsockaddr() and in_setpeeraddr() was
668 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
669 * in struct pr_usrreqs, so that protocols can just reference then directly
670 * without the need for a wrapper function. The socket must have a valid
671 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
672 * except through a kernel programming error, so it is acceptable to panic
673 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
674 * because there actually /is/ a programming error somewhere... XXX)
677 in_setsockaddr(struct socket *so, struct sockaddr **nam)
680 struct sockaddr_in *sin;
683 * Do the malloc first in case it blocks.
685 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
687 sin->sin_family = AF_INET;
688 sin->sin_len = sizeof *sin;
694 kfree(sin, M_SONAME);
697 sin->sin_port = inp->inp_lport;
698 sin->sin_addr = inp->inp_laddr;
701 *nam = (struct sockaddr *)sin;
706 in_setpeeraddr(struct socket *so, struct sockaddr **nam)
709 struct sockaddr_in *sin;
712 * Do the malloc first in case it blocks.
714 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
716 sin->sin_family = AF_INET;
717 sin->sin_len = sizeof *sin;
723 kfree(sin, M_SONAME);
726 sin->sin_port = inp->inp_fport;
727 sin->sin_addr = inp->inp_faddr;
730 *nam = (struct sockaddr *)sin;
735 in_pcbnotifyall(struct inpcbhead *head, struct in_addr faddr, int err,
736 void (*notify)(struct inpcb *, int))
738 struct inpcb *inp, *ninp;
741 * note: if INP_PLACEMARKER is set we must ignore the rest of
742 * the structure and skip it.
745 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) {
746 if (inp->inp_flags & INP_PLACEMARKER)
749 if (!(inp->inp_vflag & INP_IPV4))
752 if (inp->inp_faddr.s_addr != faddr.s_addr ||
753 inp->inp_socket == NULL)
755 (*notify)(inp, err); /* can remove inp from list! */
761 in_pcbpurgeif0(struct inpcb *head, struct ifnet *ifp)
764 struct ip_moptions *imo;
767 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) {
768 if (inp->inp_flags & INP_PLACEMARKER)
770 imo = inp->inp_moptions;
771 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) {
773 * Unselect the outgoing interface if it is being
776 if (imo->imo_multicast_ifp == ifp)
777 imo->imo_multicast_ifp = NULL;
780 * Drop multicast group membership if we joined
781 * through the interface being detached.
783 for (i = 0, gap = 0; i < imo->imo_num_memberships;
785 if (imo->imo_membership[i]->inm_ifp == ifp) {
786 in_delmulti(imo->imo_membership[i]);
789 imo->imo_membership[i - gap] =
790 imo->imo_membership[i];
792 imo->imo_num_memberships -= gap;
798 * Check for alternatives when higher level complains
799 * about service problems. For now, invalidate cached
800 * routing information. If the route was created dynamically
801 * (by a redirect), time to try a default gateway again.
804 in_losing(struct inpcb *inp)
807 struct rt_addrinfo rtinfo;
809 if ((rt = inp->inp_route.ro_rt)) {
810 bzero(&rtinfo, sizeof(struct rt_addrinfo));
811 rtinfo.rti_info[RTAX_DST] = rt_key(rt);
812 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
813 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt);
814 rtinfo.rti_flags = rt->rt_flags;
815 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0);
816 if (rt->rt_flags & RTF_DYNAMIC)
817 rtrequest1_global(RTM_DELETE, &rtinfo, NULL, NULL);
818 inp->inp_route.ro_rt = NULL;
821 * A new route can be allocated
822 * the next time output is attempted.
828 * After a routing change, flush old routing
829 * and allocate a (hopefully) better one.
832 in_rtchange(struct inpcb *inp, int err)
834 if (inp->inp_route.ro_rt) {
835 rtfree(inp->inp_route.ro_rt);
836 inp->inp_route.ro_rt = NULL;
838 * A new route can be allocated the next time
839 * output is attempted.
845 * Lookup a PCB based on the local address and port.
848 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr,
849 u_int lport_arg, int wild_okay, struct ucred *cred)
852 int matchwild = 3, wildcard;
853 u_short lport = lport_arg;
855 struct inpcbporthead *porthash;
856 struct inpcbport *phd;
857 struct inpcb *match = NULL;
860 * Best fit PCB lookup.
862 * First see if this local port is in use by looking on the
865 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
866 pcbinfo->porthashmask)];
867 LIST_FOREACH(phd, porthash, phd_hash) {
868 if (phd->phd_port == lport)
873 * Port is in use by one or more PCBs. Look for best
876 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
879 if ((inp->inp_vflag & INP_IPV4) == 0)
882 if (inp->inp_faddr.s_addr != INADDR_ANY)
884 if (inp->inp_laddr.s_addr != INADDR_ANY) {
885 if (laddr.s_addr == INADDR_ANY)
887 else if (inp->inp_laddr.s_addr != laddr.s_addr)
890 if (laddr.s_addr != INADDR_ANY)
893 if (wildcard && !wild_okay)
895 if (wildcard < matchwild &&
898 inp->inp_socket->so_cred->cr_prison)) {
900 matchwild = wildcard;
901 if (matchwild == 0) {
911 * Lookup PCB in hash list.
914 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg,
915 struct in_addr laddr, u_int lport_arg, boolean_t wildcard,
918 struct inpcbhead *head;
919 struct inpcb *inp, *jinp=NULL;
920 u_short fport = fport_arg, lport = lport_arg;
923 * First look for an exact match.
925 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport,
926 laddr.s_addr, lport, pcbinfo->hashmask)];
927 LIST_FOREACH(inp, head, inp_hash) {
929 if (!(inp->inp_vflag & INP_IPV4))
932 if (in_hosteq(inp->inp_faddr, faddr) &&
933 in_hosteq(inp->inp_laddr, laddr) &&
934 inp->inp_fport == fport && inp->inp_lport == lport) {
936 if (inp->inp_socket == NULL ||
937 inp->inp_socket->so_cred->cr_prison == NULL) {
948 struct inpcb *local_wild = NULL;
949 struct inpcb *jinp_wild = NULL;
951 struct inpcb *local_wild_mapped = NULL;
953 struct inpcontainer *ic;
954 struct inpcontainerhead *chead;
955 struct sockaddr_in jsin;
959 * Order of socket selection:
960 * 1. non-jailed, non-wild.
961 * 2. non-jailed, wild.
962 * 3. jailed, non-wild.
965 jsin.sin_family = AF_INET;
966 chead = &pcbinfo->wildcardhashbase[
967 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)];
968 LIST_FOREACH(ic, chead, ic_list) {
970 jsin.sin_addr.s_addr = laddr.s_addr;
972 if (!(inp->inp_vflag & INP_IPV4))
975 if (inp->inp_socket != NULL)
976 cred = inp->inp_socket->so_cred;
979 if (cred != NULL && jailed(cred)) {
983 if (!jailed_ip(cred->cr_prison,
984 (struct sockaddr *)&jsin))
987 if (inp->inp_lport == lport) {
988 if (ifp && ifp->if_type == IFT_FAITH &&
989 !(inp->inp_flags & INP_FAITH))
991 if (inp->inp_laddr.s_addr == laddr.s_addr) {
992 if (cred != NULL && jailed(cred))
997 if (inp->inp_laddr.s_addr == INADDR_ANY) {
999 if (INP_CHECK_SOCKAF(inp->inp_socket,
1001 local_wild_mapped = inp;
1012 if (local_wild != NULL)
1013 return (local_wild);
1015 if (local_wild_mapped != NULL)
1016 return (local_wild_mapped);
1030 * Insert PCB into connection hash table.
1033 in_pcbinsconnhash(struct inpcb *inp)
1035 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo;
1036 struct inpcbhead *bucket;
1037 u_int32_t hashkey_faddr, hashkey_laddr;
1040 if (inp->inp_vflag & INP_IPV6) {
1041 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */;
1042 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */;
1045 hashkey_faddr = inp->inp_faddr.s_addr;
1046 hashkey_laddr = inp->inp_laddr.s_addr;
1051 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list"));
1052 inp->inp_flags |= INP_CONNECTED;
1055 * Insert into the connection hash table.
1057 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr,
1058 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)];
1059 LIST_INSERT_HEAD(bucket, inp, inp_hash);
1063 * Remove PCB from connection hash table.
1066 in_pcbremconnhash(struct inpcb *inp)
1068 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected"));
1069 LIST_REMOVE(inp, inp_hash);
1070 inp->inp_flags &= ~INP_CONNECTED;
1074 * Insert PCB into port hash table.
1077 in_pcbinsporthash(struct inpcb *inp)
1079 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1080 struct inpcbporthead *pcbporthash;
1081 struct inpcbport *phd;
1084 * Insert into the port hash table.
1086 pcbporthash = &pcbinfo->porthashbase[
1087 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)];
1089 /* Go through port list and look for a head for this lport. */
1090 LIST_FOREACH(phd, pcbporthash, phd_hash)
1091 if (phd->phd_port == inp->inp_lport)
1094 /* If none exists, malloc one and tack it on. */
1096 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport),
1097 M_PCB, M_INTWAIT | M_NULLOK);
1099 return (ENOBUFS); /* XXX */
1100 phd->phd_port = inp->inp_lport;
1101 LIST_INIT(&phd->phd_pcblist);
1102 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash);
1106 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist);
1112 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1114 struct inpcontainer *ic;
1115 struct inpcontainerhead *bucket;
1117 bucket = &pcbinfo->wildcardhashbase[
1118 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1120 ic = kmalloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT);
1122 LIST_INSERT_HEAD(bucket, ic, ic_list);
1126 * Insert PCB into wildcard hash table.
1129 in_pcbinswildcardhash(struct inpcb *inp)
1131 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1133 KKASSERT(pcbinfo != NULL);
1135 in_pcbinswildcardhash_oncpu(inp, pcbinfo);
1136 inp->inp_flags |= INP_WILDCARD;
1140 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1142 struct inpcontainer *ic;
1143 struct inpcontainerhead *head;
1146 head = &pcbinfo->wildcardhashbase[
1147 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1149 LIST_FOREACH(ic, head, ic_list) {
1150 if (ic->ic_inp == inp)
1153 return; /* not found! */
1156 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */
1157 kfree(ic, M_TEMP); /* deallocate container */
1161 * Remove PCB from wildcard hash table.
1164 in_pcbremwildcardhash(struct inpcb *inp)
1166 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1168 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard"));
1169 in_pcbremwildcardhash_oncpu(inp, pcbinfo);
1170 inp->inp_flags &= ~INP_WILDCARD;
1174 * Remove PCB from various lists.
1177 in_pcbremlists(struct inpcb *inp)
1179 if (inp->inp_lport) {
1180 struct inpcbport *phd = inp->inp_phd;
1182 LIST_REMOVE(inp, inp_portlist);
1183 if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
1184 LIST_REMOVE(phd, phd_hash);
1188 if (inp->inp_flags & INP_WILDCARD) {
1189 in_pcbremwildcardhash(inp);
1190 } else if (inp->inp_flags & INP_CONNECTED) {
1191 in_pcbremconnhash(inp);
1193 LIST_REMOVE(inp, inp_list);
1194 inp->inp_pcbinfo->ipi_count--;
1198 prison_xinpcb(struct thread *td, struct inpcb *inp)
1202 if (td->td_proc == NULL)
1204 cr = td->td_proc->p_ucred;
1205 if (cr->cr_prison == NULL)
1207 if (inp->inp_socket && inp->inp_socket->so_cred &&
1208 inp->inp_socket->so_cred->cr_prison &&
1209 cr->cr_prison == inp->inp_socket->so_cred->cr_prison)
1215 in_pcblist_global(SYSCTL_HANDLER_ARGS)
1217 struct inpcbinfo *pcbinfo = arg1;
1218 struct inpcb *inp, *marker;
1224 * The process of preparing the TCB list is too time-consuming and
1225 * resource-intensive to repeat twice on every request.
1227 if (req->oldptr == NULL) {
1228 n = pcbinfo->ipi_count;
1229 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb);
1233 if (req->newptr != NULL)
1237 * OK, now we're committed to doing something. Re-fetch ipi_count
1238 * after obtaining the generation count.
1240 gencnt = pcbinfo->ipi_gencnt;
1241 n = pcbinfo->ipi_count;
1243 marker = kmalloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO);
1244 marker->inp_flags |= INP_PLACEMARKER;
1245 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
1250 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) {
1251 LIST_REMOVE(marker, inp_list);
1252 LIST_INSERT_AFTER(inp, marker, inp_list);
1254 if (inp->inp_flags & INP_PLACEMARKER)
1256 if (inp->inp_gencnt > gencnt)
1258 if (prison_xinpcb(req->td, inp))
1260 bzero(&xi, sizeof xi);
1261 xi.xi_len = sizeof xi;
1262 bcopy(inp, &xi.xi_inp, sizeof *inp);
1263 if (inp->inp_socket)
1264 sotoxsocket(inp->inp_socket, &xi.xi_socket);
1265 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0)
1269 LIST_REMOVE(marker, inp_list);
1270 if (error == 0 && i < n) {
1271 bzero(&xi, sizeof xi);
1272 xi.xi_len = sizeof xi;
1274 error = SYSCTL_OUT(req, &xi, sizeof xi);
1278 kfree(marker, M_TEMP);