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 1.33 2005/02/08 22:56:19 hsu 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>
104 #include <machine/limits.h>
106 #include <vm/vm_zone.h>
109 #include <net/if_types.h>
110 #include <net/route.h>
112 #include <netinet/in.h>
113 #include <netinet/in_pcb.h>
114 #include <netinet/in_var.h>
115 #include <netinet/ip_var.h>
117 #include <netinet/ip6.h>
118 #include <netinet6/ip6_var.h>
122 #include <netinet6/ipsec.h>
123 #include <netproto/key/key.h>
127 #if defined(IPSEC) || defined(IPSEC_ESP)
128 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!"
131 #include <netproto/ipsec/ipsec.h>
132 #include <netproto/ipsec/key.h>
134 #endif /* FAST_IPSEC */
136 struct in_addr zeroin_addr;
139 * These configure the range of local port addresses assigned to
140 * "unspecified" outgoing connections/packets/whatever.
142 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */
143 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */
145 int ipport_firstauto = IPPORT_RESERVED; /* 1024 */
146 int ipport_lastauto = IPPORT_USERRESERVED; /* 5000 */
148 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */
149 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */
152 RANGECHK(int var, int min, int max)
161 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS)
165 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req);
167 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1);
168 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1);
170 RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX);
171 RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX);
173 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX);
174 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX);
179 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports");
181 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW,
182 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", "");
183 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW,
184 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", "");
185 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW,
186 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", "");
187 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW,
188 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", "");
189 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW,
190 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", "");
191 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW,
192 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", "");
195 * in_pcb.c: manage the Protocol Control Blocks.
197 * NOTE: It is assumed that most of these functions will be called at
198 * splnet(). XXX - There are, unfortunately, a few exceptions to this
199 * rule that should be fixed.
201 * NOTE: The caller should initialize the cpu field to the cpu running the
202 * protocol stack associated with this inpcbinfo.
206 in_pcbinfo_init(struct inpcbinfo *pcbinfo)
208 LIST_INIT(&pcbinfo->pcblisthead);
213 * Allocate a PCB and associate it with the socket.
216 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo)
223 inp = zalloc(pcbinfo->ipi_zone);
226 bzero(inp, sizeof *inp);
227 inp->inp_gencnt = ++pcbinfo->ipi_gencnt;
228 inp->inp_pcbinfo = inp->inp_cpcbinfo = pcbinfo;
229 inp->inp_socket = so;
231 error = ipsec_init_policy(so, &inp->inp_sp);
233 zfree(pcbinfo->ipi_zone, inp);
238 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only)
239 inp->inp_flags |= IN6P_IPV6_V6ONLY;
240 if (ip6_auto_flowlabel)
241 inp->inp_flags |= IN6P_AUTOFLOWLABEL;
244 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list);
245 pcbinfo->ipi_count++;
250 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
252 struct socket *so = inp->inp_socket;
253 struct proc *p = td->td_proc;
254 unsigned short *lastport;
255 struct sockaddr_in *sin;
256 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
258 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
259 int error, prison = 0;
263 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */
264 return (EADDRNOTAVAIL);
265 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY)
266 return (EINVAL); /* already bound */
267 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT)))
268 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */
270 sin = (struct sockaddr_in *)nam;
271 if (nam->sa_len != sizeof *sin)
275 * We should check the family, but old programs
276 * incorrectly fail to initialize it.
278 if (sin->sin_family != AF_INET)
279 return (EAFNOSUPPORT);
281 if (sin->sin_addr.s_addr != INADDR_ANY &&
282 prison_ip(td, 0, &sin->sin_addr.s_addr))
284 lport = sin->sin_port;
285 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
287 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
288 * allow complete duplication of binding if
289 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
290 * and a multicast address is bound on both
291 * new and duplicated sockets.
293 if (so->so_options & SO_REUSEADDR)
294 reuseport = SO_REUSEADDR | SO_REUSEPORT;
295 } else if (sin->sin_addr.s_addr != INADDR_ANY) {
296 sin->sin_port = 0; /* yech... */
297 bzero(&sin->sin_zero, sizeof sin->sin_zero);
298 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL)
299 return (EADDRNOTAVAIL);
305 if (ntohs(lport) < IPPORT_RESERVED &&
306 p && suser_cred(p->p_ucred, PRISON_ROOT))
308 if (p && p->p_ucred->cr_prison)
310 if (so->so_cred->cr_uid != 0 &&
311 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
312 t = in_pcblookup_local(inp->inp_pcbinfo,
313 sin->sin_addr, lport,
314 prison ? 0 : INPLOOKUP_WILDCARD);
316 (!in_nullhost(sin->sin_addr) ||
317 !in_nullhost(t->inp_laddr) ||
318 (t->inp_socket->so_options &
319 SO_REUSEPORT) == 0) &&
320 (so->so_cred->cr_uid !=
321 t->inp_socket->so_cred->cr_uid)) {
323 if (!in_nullhost(sin->sin_addr) ||
324 !in_nullhost(t->inp_laddr) ||
326 INP_SOCKAF(t->inp_socket))
331 if (prison && prison_ip(td, 0, &sin->sin_addr.s_addr))
332 return (EADDRNOTAVAIL);
333 t = in_pcblookup_local(pcbinfo, sin->sin_addr,
334 lport, prison ? 0 : wild);
335 if (t && !(reuseport & t->inp_socket->so_options)) {
337 if (!in_nullhost(sin->sin_addr) ||
338 !in_nullhost(t->inp_laddr) ||
339 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))
344 inp->inp_laddr = sin->sin_addr;
350 if (inp->inp_laddr.s_addr != INADDR_ANY &&
351 prison_ip(td, 0, &inp->inp_laddr.s_addr )) {
352 inp->inp_laddr.s_addr = INADDR_ANY;
355 inp->inp_flags |= INP_ANONPORT;
357 if (inp->inp_flags & INP_HIGHPORT) {
358 first = ipport_hifirstauto; /* sysctl */
359 last = ipport_hilastauto;
360 lastport = &pcbinfo->lasthi;
361 } else if (inp->inp_flags & INP_LOWPORT) {
363 (error = suser_cred(p->p_ucred, PRISON_ROOT))) {
364 inp->inp_laddr.s_addr = INADDR_ANY;
367 first = ipport_lowfirstauto; /* 1023 */
368 last = ipport_lowlastauto; /* 600 */
369 lastport = &pcbinfo->lastlow;
371 first = ipport_firstauto; /* sysctl */
372 last = ipport_lastauto;
373 lastport = &pcbinfo->lastport;
376 * Simple check to ensure all ports are not used up causing
379 * We split the two cases (up and down) so that the direction
380 * is not being tested on each round of the loop.
386 count = first - last;
389 if (count-- < 0) { /* completely used? */
390 inp->inp_laddr.s_addr = INADDR_ANY;
391 return (EADDRNOTAVAIL);
394 if (*lastport > first || *lastport < last)
396 lport = htons(*lastport);
397 } while (in_pcblookup_local(pcbinfo,
398 inp->inp_laddr, lport, wild));
403 count = last - first;
406 if (count-- < 0) { /* completely used? */
407 inp->inp_laddr.s_addr = INADDR_ANY;
408 return (EADDRNOTAVAIL);
411 if (*lastport < first || *lastport > last)
413 lport = htons(*lastport);
414 } while (in_pcblookup_local(pcbinfo,
415 inp->inp_laddr, lport, wild));
418 inp->inp_lport = lport;
419 if (prison_ip(td, 0, &inp->inp_laddr.s_addr)) {
420 inp->inp_laddr.s_addr = INADDR_ANY;
424 if (in_pcbinsporthash(inp) != 0) {
425 inp->inp_laddr.s_addr = INADDR_ANY;
433 * Transform old in_pcbconnect() into an inner subroutine for new
434 * in_pcbconnect(): Do some validity-checking on the remote
435 * address (in mbuf 'nam') and then determine local host address
436 * (i.e., which interface) to use to access that remote host.
438 * This preserves definition of in_pcbconnect(), while supporting a
439 * slightly different version for T/TCP. (This is more than
440 * a bit of a kludge, but cleaning up the internal interfaces would
441 * have forced minor changes in every protocol).
444 in_pcbladdr(inp, nam, plocal_sin)
446 struct sockaddr *nam;
447 struct sockaddr_in **plocal_sin;
449 struct in_ifaddr *ia;
450 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
452 if (nam->sa_len != sizeof *sin)
454 if (sin->sin_family != AF_INET)
455 return (EAFNOSUPPORT);
456 if (sin->sin_port == 0)
457 return (EADDRNOTAVAIL);
458 if (!TAILQ_EMPTY(&in_ifaddrhead)) {
459 ia = TAILQ_FIRST(&in_ifaddrhead);
461 * If the destination address is INADDR_ANY,
462 * use the primary local address.
463 * If the supplied address is INADDR_BROADCAST,
464 * and the primary interface supports broadcast,
465 * choose the broadcast address for that interface.
467 if (sin->sin_addr.s_addr == INADDR_ANY)
468 sin->sin_addr = IA_SIN(ia)->sin_addr;
469 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST &&
470 (ia->ia_ifp->if_flags & IFF_BROADCAST))
471 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr;
473 if (inp->inp_laddr.s_addr == INADDR_ANY) {
476 ia = (struct in_ifaddr *)NULL;
478 * If route is known or can be allocated now,
479 * our src addr is taken from the i/f, else punt.
480 * Note that we should check the address family of the cached
481 * destination, in case of sharing the cache with IPv6.
483 ro = &inp->inp_route;
485 (!(ro->ro_rt->rt_flags & RTF_UP) ||
486 ro->ro_dst.sa_family != AF_INET ||
487 satosin(&ro->ro_dst)->sin_addr.s_addr !=
488 sin->sin_addr.s_addr ||
489 inp->inp_socket->so_options & SO_DONTROUTE)) {
491 ro->ro_rt = (struct rtentry *)NULL;
493 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/
494 (ro->ro_rt == (struct rtentry *)NULL ||
495 ro->ro_rt->rt_ifp == (struct ifnet *)NULL)) {
496 /* No route yet, so try to acquire one */
497 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
498 ro->ro_dst.sa_family = AF_INET;
499 ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
500 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
505 * If we found a route, use the address
506 * corresponding to the outgoing interface
507 * unless it is the loopback (in case a route
508 * to our address on another net goes to loopback).
510 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
511 ia = ifatoia(ro->ro_rt->rt_ifa);
513 u_short fport = sin->sin_port;
516 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin)));
518 ia = ifatoia(ifa_ifwithnet(sintosa(sin)));
519 sin->sin_port = fport;
521 ia = TAILQ_FIRST(&in_ifaddrhead);
523 return (EADDRNOTAVAIL);
526 * If the destination address is multicast and an outgoing
527 * interface has been set as a multicast option, use the
528 * address of that interface as our source address.
530 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) &&
531 inp->inp_moptions != NULL) {
532 struct ip_moptions *imo;
535 imo = inp->inp_moptions;
536 if (imo->imo_multicast_ifp != NULL) {
537 ifp = imo->imo_multicast_ifp;
538 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link)
539 if (ia->ia_ifp == ifp)
542 return (EADDRNOTAVAIL);
546 * Don't do pcblookup call here; return interface in plocal_sin
547 * and exit to caller, that will do the lookup.
549 *plocal_sin = &ia->ia_addr;
557 * Connect from a socket to a specified address.
558 * Both address and port must be specified in argument sin.
559 * If don't have a local address for this socket yet,
563 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
565 struct sockaddr_in *if_sin;
566 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
567 struct sockaddr_in sa;
568 struct ucred *cr = td->td_proc ? td->td_proc->p_ucred : NULL;
571 if (cr && cr->cr_prison != NULL && in_nullhost(inp->inp_laddr)) {
572 bzero(&sa, sizeof sa);
573 sa.sin_addr.s_addr = htonl(cr->cr_prison->pr_ip);
574 sa.sin_len = sizeof sa;
575 sa.sin_family = AF_INET;
576 error = in_pcbbind(inp, (struct sockaddr *)&sa, td);
581 /* Call inner routine to assign local interface address. */
582 if ((error = in_pcbladdr(inp, nam, &if_sin)) != 0)
585 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port,
586 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr,
587 inp->inp_lport, FALSE, NULL) != NULL) {
590 if (inp->inp_laddr.s_addr == INADDR_ANY) {
591 if (inp->inp_lport == 0) {
592 error = in_pcbbind(inp, (struct sockaddr *)NULL, td);
596 inp->inp_laddr = if_sin->sin_addr;
598 inp->inp_faddr = sin->sin_addr;
599 inp->inp_fport = sin->sin_port;
600 in_pcbinsconnhash(inp);
605 in_pcbdisconnect(inp)
609 inp->inp_faddr.s_addr = INADDR_ANY;
611 in_pcbremconnhash(inp);
612 if (inp->inp_socket->so_state & SS_NOFDREF)
620 struct socket *so = inp->inp_socket;
621 struct inpcbinfo *ipi = inp->inp_pcbinfo;
624 ipsec4_delete_pcbpolicy(inp);
626 inp->inp_gencnt = ++ipi->ipi_gencnt;
630 if (inp->inp_options)
631 m_free(inp->inp_options);
632 if (inp->inp_route.ro_rt)
633 rtfree(inp->inp_route.ro_rt);
634 ip_freemoptions(inp->inp_moptions);
636 zfree(ipi->ipi_zone, inp);
640 * The calling convention of in_setsockaddr() and in_setpeeraddr() was
641 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
642 * in struct pr_usrreqs, so that protocols can just reference then directly
643 * without the need for a wrapper function. The socket must have a valid
644 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
645 * except through a kernel programming error, so it is acceptable to panic
646 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
647 * because there actually /is/ a programming error somewhere... XXX)
650 in_setsockaddr(so, nam)
652 struct sockaddr **nam;
656 struct sockaddr_in *sin;
659 * Do the malloc first in case it blocks.
661 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
663 sin->sin_family = AF_INET;
664 sin->sin_len = sizeof *sin;
673 sin->sin_port = inp->inp_lport;
674 sin->sin_addr = inp->inp_laddr;
677 *nam = (struct sockaddr *)sin;
682 in_setpeeraddr(so, nam)
684 struct sockaddr **nam;
688 struct sockaddr_in *sin;
691 * Do the malloc first in case it blocks.
693 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
695 sin->sin_family = AF_INET;
696 sin->sin_len = sizeof *sin;
705 sin->sin_port = inp->inp_fport;
706 sin->sin_addr = inp->inp_faddr;
709 *nam = (struct sockaddr *)sin;
714 in_pcbnotifyall(head, faddr, errno, notify)
715 struct inpcbhead *head;
716 struct in_addr faddr;
717 void (*notify) (struct inpcb *, int);
719 struct inpcb *inp, *ninp;
723 * note: if INP_PLACEMARKER is set we must ignore the rest of
724 * the structure and skip it.
727 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) {
728 if (inp->inp_flags & INP_PLACEMARKER)
731 if (!(inp->inp_vflag & INP_IPV4))
734 if (inp->inp_faddr.s_addr != faddr.s_addr ||
735 inp->inp_socket == NULL)
737 (*notify)(inp, errno); /* can remove inp from list! */
743 in_pcbpurgeif0(head, ifp)
748 struct ip_moptions *imo;
751 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) {
752 if (inp->inp_flags & INP_PLACEMARKER)
754 imo = inp->inp_moptions;
755 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) {
757 * Unselect the outgoing interface if it is being
760 if (imo->imo_multicast_ifp == ifp)
761 imo->imo_multicast_ifp = NULL;
764 * Drop multicast group membership if we joined
765 * through the interface being detached.
767 for (i = 0, gap = 0; i < imo->imo_num_memberships;
769 if (imo->imo_membership[i]->inm_ifp == ifp) {
770 in_delmulti(imo->imo_membership[i]);
773 imo->imo_membership[i - gap] =
774 imo->imo_membership[i];
776 imo->imo_num_memberships -= gap;
782 * Check for alternatives when higher level complains
783 * about service problems. For now, invalidate cached
784 * routing information. If the route was created dynamically
785 * (by a redirect), time to try a default gateway again.
792 struct rt_addrinfo info;
794 if ((rt = inp->inp_route.ro_rt)) {
795 bzero(&info, sizeof info);
796 info.rti_flags = rt->rt_flags;
797 info.rti_info[RTAX_DST] = rt_key(rt);
798 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
799 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
800 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
801 if (rt->rt_flags & RTF_DYNAMIC)
802 rtrequest1(RTM_DELETE, &info, NULL);
803 inp->inp_route.ro_rt = NULL;
806 * A new route can be allocated
807 * the next time output is attempted.
813 * After a routing change, flush old routing
814 * and allocate a (hopefully) better one.
817 in_rtchange(inp, errno)
821 if (inp->inp_route.ro_rt) {
822 rtfree(inp->inp_route.ro_rt);
823 inp->inp_route.ro_rt = 0;
825 * A new route can be allocated the next time
826 * output is attempted.
832 * Lookup a PCB based on the local address and port.
835 in_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay)
836 struct inpcbinfo *pcbinfo;
837 struct in_addr laddr;
842 int matchwild = 3, wildcard;
843 u_short lport = lport_arg;
845 struct inpcbporthead *porthash;
846 struct inpcbport *phd;
847 struct inpcb *match = NULL;
850 * Best fit PCB lookup.
852 * First see if this local port is in use by looking on the
855 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
856 pcbinfo->porthashmask)];
857 LIST_FOREACH(phd, porthash, phd_hash) {
858 if (phd->phd_port == lport)
863 * Port is in use by one or more PCBs. Look for best
866 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
869 if ((inp->inp_vflag & INP_IPV4) == 0)
872 if (inp->inp_faddr.s_addr != INADDR_ANY)
874 if (inp->inp_laddr.s_addr != INADDR_ANY) {
875 if (laddr.s_addr == INADDR_ANY)
877 else if (inp->inp_laddr.s_addr != laddr.s_addr)
880 if (laddr.s_addr != INADDR_ANY)
883 if (wildcard && !wild_okay)
885 if (wildcard < matchwild) {
887 matchwild = wildcard;
888 if (matchwild == 0) {
898 * Lookup PCB in hash list.
901 in_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, ifp)
902 struct inpcbinfo *pcbinfo;
903 struct in_addr faddr, laddr;
904 u_int fport_arg, lport_arg;
908 struct inpcbhead *head;
910 u_short fport = fport_arg, lport = lport_arg;
913 * First look for an exact match.
915 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport,
916 laddr.s_addr, lport, pcbinfo->hashmask)];
917 LIST_FOREACH(inp, head, inp_hash) {
919 if (!(inp->inp_vflag & INP_IPV4))
922 if (in_hosteq(inp->inp_faddr, faddr) &&
923 in_hosteq(inp->inp_laddr, laddr) &&
924 inp->inp_fport == fport && inp->inp_lport == lport) {
931 struct inpcb *local_wild = NULL;
933 struct inpcb *local_wild_mapped = NULL;
935 struct inpcontainer *ic;
936 struct inpcontainerhead *chead;
938 chead = &pcbinfo->wildcardhashbase[
939 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)];
940 LIST_FOREACH(ic, chead, ic_list) {
943 if (!(inp->inp_vflag & INP_IPV4))
946 if (inp->inp_lport == lport) {
947 if (ifp && ifp->if_type == IFT_FAITH &&
948 !(inp->inp_flags & INP_FAITH))
950 if (inp->inp_laddr.s_addr == laddr.s_addr)
952 if (inp->inp_laddr.s_addr == INADDR_ANY) {
954 if (INP_CHECK_SOCKAF(inp->inp_socket,
956 local_wild_mapped = inp;
964 if (local_wild == NULL)
965 return (local_wild_mapped);
977 * Insert PCB into connection hash table.
980 in_pcbinsconnhash(struct inpcb *inp)
982 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo;
983 struct inpcbhead *bucket;
984 u_int32_t hashkey_faddr, hashkey_laddr;
987 if (inp->inp_vflag & INP_IPV6) {
988 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */;
989 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */;
992 hashkey_faddr = inp->inp_faddr.s_addr;
993 hashkey_laddr = inp->inp_laddr.s_addr;
998 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list"));
999 inp->inp_flags |= INP_CONNECTED;
1002 * Insert into the connection hash table.
1004 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr,
1005 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)];
1006 LIST_INSERT_HEAD(bucket, inp, inp_hash);
1010 * Remove PCB from connection hash table.
1013 in_pcbremconnhash(struct inpcb *inp)
1015 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected"));
1016 LIST_REMOVE(inp, inp_hash);
1017 inp->inp_flags &= ~INP_CONNECTED;
1021 * Insert PCB into port hash table.
1024 in_pcbinsporthash(struct inpcb *inp)
1026 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1027 struct inpcbporthead *pcbporthash;
1028 struct inpcbport *phd;
1031 * Insert into the port hash table.
1033 pcbporthash = &pcbinfo->porthashbase[
1034 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)];
1036 /* Go through port list and look for a head for this lport. */
1037 LIST_FOREACH(phd, pcbporthash, phd_hash)
1038 if (phd->phd_port == inp->inp_lport)
1041 /* If none exists, malloc one and tack it on. */
1043 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport),
1044 M_PCB, M_INTWAIT | M_NULLOK);
1046 return (ENOBUFS); /* XXX */
1047 phd->phd_port = inp->inp_lport;
1048 LIST_INIT(&phd->phd_pcblist);
1049 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash);
1053 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist);
1059 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1061 struct inpcontainer *ic;
1062 struct inpcontainerhead *bucket;
1064 bucket = &pcbinfo->wildcardhashbase[
1065 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1067 ic = malloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT);
1069 LIST_INSERT_HEAD(bucket, ic, ic_list);
1073 * Insert PCB into wildcard hash table.
1076 in_pcbinswildcardhash(struct inpcb *inp)
1078 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1080 KKASSERT(pcbinfo != NULL);
1082 in_pcbinswildcardhash_oncpu(inp, pcbinfo);
1083 inp->inp_flags |= INP_WILDCARD;
1087 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1089 struct inpcontainer *ic;
1090 struct inpcontainerhead *head;
1093 head = &pcbinfo->wildcardhashbase[
1094 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1096 LIST_FOREACH(ic, head, ic_list) {
1097 if (ic->ic_inp == inp)
1100 return; /* not found! */
1103 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */
1104 free(ic, M_TEMP); /* deallocate container */
1108 * Remove PCB from wildcard hash table.
1111 in_pcbremwildcardhash(struct inpcb *inp)
1113 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1115 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard"));
1116 in_pcbremwildcardhash_oncpu(inp, pcbinfo);
1117 inp->inp_flags &= ~INP_WILDCARD;
1121 * Remove PCB from various lists.
1127 if (inp->inp_lport) {
1128 struct inpcbport *phd = inp->inp_phd;
1130 LIST_REMOVE(inp, inp_portlist);
1131 if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
1132 LIST_REMOVE(phd, phd_hash);
1136 if (inp->inp_flags & INP_WILDCARD) {
1137 in_pcbremwildcardhash(inp);
1138 } else if (inp->inp_flags & INP_CONNECTED) {
1139 in_pcbremconnhash(inp);
1141 LIST_REMOVE(inp, inp_list);
1142 inp->inp_pcbinfo->ipi_count--;
1146 prison_xinpcb(struct thread *td, struct inpcb *inp)
1150 if (td->td_proc == NULL)
1152 cr = td->td_proc->p_ucred;
1153 if (cr->cr_prison == NULL)
1155 if (ntohl(inp->inp_laddr.s_addr) == cr->cr_prison->pr_ip)
1161 in_pcblist_global(SYSCTL_HANDLER_ARGS)
1163 struct inpcbinfo *pcbinfo = arg1;
1164 struct inpcb *inp, *marker;
1170 * The process of preparing the TCB list is too time-consuming and
1171 * resource-intensive to repeat twice on every request.
1173 if (req->oldptr == NULL) {
1174 n = pcbinfo->ipi_count;
1175 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb);
1179 if (req->newptr != NULL)
1183 * OK, now we're committed to doing something. Re-fetch ipi_count
1184 * after obtaining the generation count.
1186 gencnt = pcbinfo->ipi_gencnt;
1187 n = pcbinfo->ipi_count;
1189 marker = malloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO);
1190 marker->inp_flags |= INP_PLACEMARKER;
1191 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
1196 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) {
1197 LIST_REMOVE(marker, inp_list);
1198 LIST_INSERT_AFTER(inp, marker, inp_list);
1200 if (inp->inp_flags & INP_PLACEMARKER)
1202 if (inp->inp_gencnt > gencnt)
1204 if (prison_xinpcb(req->td, inp))
1206 xi.xi_len = sizeof xi;
1207 bcopy(inp, &xi.xi_inp, sizeof *inp);
1208 if (inp->inp_socket)
1209 sotoxsocket(inp->inp_socket, &xi.xi_socket);
1210 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0)
1214 LIST_REMOVE(marker, inp_list);
1215 if (error == 0 && i < n) {
1216 bzero(&xi, sizeof xi);
1217 xi.xi_len = sizeof xi;
1219 error = SYSCTL_OUT(req, &xi, sizeof xi);
1223 free(marker, M_TEMP);