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 1.39 2006/09/05 00:55:48 dillon 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 inpcbinfo *pcbinfo = inp->inp_pcbinfo;
259 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
260 int error, prison = 0;
264 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */
265 return (EADDRNOTAVAIL);
266 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY)
267 return (EINVAL); /* already bound */
268 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT)))
269 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */
271 sin = (struct sockaddr_in *)nam;
272 if (nam->sa_len != sizeof *sin)
276 * We should check the family, but old programs
277 * incorrectly fail to initialize it.
279 if (sin->sin_family != AF_INET)
280 return (EAFNOSUPPORT);
282 if (sin->sin_addr.s_addr != INADDR_ANY &&
283 prison_ip(td, 0, &sin->sin_addr.s_addr))
285 lport = sin->sin_port;
286 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
288 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
289 * allow complete duplication of binding if
290 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
291 * and a multicast address is bound on both
292 * new and duplicated sockets.
294 if (so->so_options & SO_REUSEADDR)
295 reuseport = SO_REUSEADDR | SO_REUSEPORT;
296 } else if (sin->sin_addr.s_addr != INADDR_ANY) {
297 sin->sin_port = 0; /* yech... */
298 bzero(&sin->sin_zero, sizeof sin->sin_zero);
299 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL)
300 return (EADDRNOTAVAIL);
306 if (ntohs(lport) < IPPORT_RESERVED &&
307 p && suser_cred(p->p_ucred, PRISON_ROOT))
309 if (p && p->p_ucred->cr_prison)
311 if (so->so_cred->cr_uid != 0 &&
312 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
313 t = in_pcblookup_local(inp->inp_pcbinfo,
314 sin->sin_addr, lport,
315 prison ? 0 : INPLOOKUP_WILDCARD);
317 (!in_nullhost(sin->sin_addr) ||
318 !in_nullhost(t->inp_laddr) ||
319 (t->inp_socket->so_options &
320 SO_REUSEPORT) == 0) &&
321 (so->so_cred->cr_uid !=
322 t->inp_socket->so_cred->cr_uid)) {
324 if (!in_nullhost(sin->sin_addr) ||
325 !in_nullhost(t->inp_laddr) ||
327 INP_SOCKAF(t->inp_socket))
332 if (prison && prison_ip(td, 0, &sin->sin_addr.s_addr))
333 return (EADDRNOTAVAIL);
334 t = in_pcblookup_local(pcbinfo, sin->sin_addr,
335 lport, prison ? 0 : wild);
336 if (t && !(reuseport & t->inp_socket->so_options)) {
338 if (!in_nullhost(sin->sin_addr) ||
339 !in_nullhost(t->inp_laddr) ||
340 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))
345 inp->inp_laddr = sin->sin_addr;
351 if (inp->inp_laddr.s_addr != INADDR_ANY &&
352 prison_ip(td, 0, &inp->inp_laddr.s_addr )) {
353 inp->inp_laddr.s_addr = INADDR_ANY;
356 inp->inp_flags |= INP_ANONPORT;
358 if (inp->inp_flags & INP_HIGHPORT) {
359 first = ipport_hifirstauto; /* sysctl */
360 last = ipport_hilastauto;
361 lastport = &pcbinfo->lasthi;
362 } else if (inp->inp_flags & INP_LOWPORT) {
364 (error = suser_cred(p->p_ucred, PRISON_ROOT))) {
365 inp->inp_laddr.s_addr = INADDR_ANY;
368 first = ipport_lowfirstauto; /* 1023 */
369 last = ipport_lowlastauto; /* 600 */
370 lastport = &pcbinfo->lastlow;
372 first = ipport_firstauto; /* sysctl */
373 last = ipport_lastauto;
374 lastport = &pcbinfo->lastport;
377 * Simple check to ensure all ports are not used up causing
380 * We split the two cases (up and down) so that the direction
381 * is not being tested on each round of the loop.
387 count = first - last;
390 if (count-- < 0) { /* completely used? */
391 inp->inp_laddr.s_addr = INADDR_ANY;
392 return (EADDRNOTAVAIL);
395 if (*lastport > first || *lastport < last)
397 lport = htons(*lastport);
398 } while (in_pcblookup_local(pcbinfo,
399 inp->inp_laddr, lport, wild));
404 count = last - first;
407 if (count-- < 0) { /* completely used? */
408 inp->inp_laddr.s_addr = INADDR_ANY;
409 return (EADDRNOTAVAIL);
412 if (*lastport < first || *lastport > last)
414 lport = htons(*lastport);
415 } while (in_pcblookup_local(pcbinfo,
416 inp->inp_laddr, lport, wild));
419 inp->inp_lport = lport;
420 if (prison_ip(td, 0, &inp->inp_laddr.s_addr)) {
421 inp->inp_laddr.s_addr = INADDR_ANY;
425 if (in_pcbinsporthash(inp) != 0) {
426 inp->inp_laddr.s_addr = INADDR_ANY;
434 * Transform old in_pcbconnect() into an inner subroutine for new
435 * in_pcbconnect(): Do some validity-checking on the remote
436 * address (in mbuf 'nam') and then determine local host address
437 * (i.e., which interface) to use to access that remote host.
439 * This preserves definition of in_pcbconnect(), while supporting a
440 * slightly different version for T/TCP. (This is more than
441 * a bit of a kludge, but cleaning up the internal interfaces would
442 * have forced minor changes in every protocol).
445 in_pcbladdr(struct inpcb *inp, struct sockaddr *nam, struct sockaddr_in **plocal_sin)
447 struct in_ifaddr *ia;
448 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
450 if (nam->sa_len != sizeof *sin)
452 if (sin->sin_family != AF_INET)
453 return (EAFNOSUPPORT);
454 if (sin->sin_port == 0)
455 return (EADDRNOTAVAIL);
456 if (!TAILQ_EMPTY(&in_ifaddrhead)) {
457 ia = TAILQ_FIRST(&in_ifaddrhead);
459 * If the destination address is INADDR_ANY,
460 * use the primary local address.
461 * If the supplied address is INADDR_BROADCAST,
462 * and the primary interface supports broadcast,
463 * choose the broadcast address for that interface.
465 if (sin->sin_addr.s_addr == INADDR_ANY)
466 sin->sin_addr = IA_SIN(ia)->sin_addr;
467 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST &&
468 (ia->ia_ifp->if_flags & IFF_BROADCAST))
469 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr;
471 if (inp->inp_laddr.s_addr == INADDR_ANY) {
474 ia = (struct in_ifaddr *)NULL;
476 * If route is known or can be allocated now,
477 * our src addr is taken from the i/f, else punt.
478 * Note that we should check the address family of the cached
479 * destination, in case of sharing the cache with IPv6.
481 ro = &inp->inp_route;
483 (!(ro->ro_rt->rt_flags & RTF_UP) ||
484 ro->ro_dst.sa_family != AF_INET ||
485 satosin(&ro->ro_dst)->sin_addr.s_addr !=
486 sin->sin_addr.s_addr ||
487 inp->inp_socket->so_options & SO_DONTROUTE)) {
489 ro->ro_rt = (struct rtentry *)NULL;
491 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/
492 (ro->ro_rt == (struct rtentry *)NULL ||
493 ro->ro_rt->rt_ifp == (struct ifnet *)NULL)) {
494 /* No route yet, so try to acquire one */
495 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
496 ro->ro_dst.sa_family = AF_INET;
497 ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
498 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
503 * If we found a route, use the address
504 * corresponding to the outgoing interface
505 * unless it is the loopback (in case a route
506 * to our address on another net goes to loopback).
508 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
509 ia = ifatoia(ro->ro_rt->rt_ifa);
511 u_short fport = sin->sin_port;
514 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin)));
516 ia = ifatoia(ifa_ifwithnet(sintosa(sin)));
517 sin->sin_port = fport;
519 ia = TAILQ_FIRST(&in_ifaddrhead);
521 return (EADDRNOTAVAIL);
524 * If the destination address is multicast and an outgoing
525 * interface has been set as a multicast option, use the
526 * address of that interface as our source address.
528 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) &&
529 inp->inp_moptions != NULL) {
530 struct ip_moptions *imo;
533 imo = inp->inp_moptions;
534 if (imo->imo_multicast_ifp != NULL) {
535 ifp = imo->imo_multicast_ifp;
536 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link)
537 if (ia->ia_ifp == ifp)
540 return (EADDRNOTAVAIL);
544 * Don't do pcblookup call here; return interface in plocal_sin
545 * and exit to caller, that will do the lookup.
547 *plocal_sin = &ia->ia_addr;
555 * Connect from a socket to a specified address.
556 * Both address and port must be specified in argument sin.
557 * If don't have a local address for this socket yet,
561 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
563 struct sockaddr_in *if_sin;
564 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
565 struct sockaddr_in sa;
566 struct ucred *cr = td->td_proc ? td->td_proc->p_ucred : NULL;
569 if (cr && cr->cr_prison != NULL && in_nullhost(inp->inp_laddr)) {
570 bzero(&sa, sizeof sa);
571 sa.sin_addr.s_addr = htonl(cr->cr_prison->pr_ip);
572 sa.sin_len = sizeof sa;
573 sa.sin_family = AF_INET;
574 error = in_pcbbind(inp, (struct sockaddr *)&sa, td);
579 /* Call inner routine to assign local interface address. */
580 if ((error = in_pcbladdr(inp, nam, &if_sin)) != 0)
583 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port,
584 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr,
585 inp->inp_lport, FALSE, NULL) != NULL) {
588 if (inp->inp_laddr.s_addr == INADDR_ANY) {
589 if (inp->inp_lport == 0) {
590 error = in_pcbbind(inp, (struct sockaddr *)NULL, td);
594 inp->inp_laddr = if_sin->sin_addr;
596 inp->inp_faddr = sin->sin_addr;
597 inp->inp_fport = sin->sin_port;
598 in_pcbinsconnhash(inp);
603 in_pcbdisconnect(struct inpcb *inp)
606 inp->inp_faddr.s_addr = INADDR_ANY;
608 in_pcbremconnhash(inp);
609 if (inp->inp_socket->so_state & SS_NOFDREF)
614 in_pcbdetach(struct inpcb *inp)
616 struct socket *so = inp->inp_socket;
617 struct inpcbinfo *ipi = inp->inp_pcbinfo;
620 ipsec4_delete_pcbpolicy(inp);
622 inp->inp_gencnt = ++ipi->ipi_gencnt;
626 if (inp->inp_options)
627 m_free(inp->inp_options);
628 if (inp->inp_route.ro_rt)
629 rtfree(inp->inp_route.ro_rt);
630 ip_freemoptions(inp->inp_moptions);
632 zfree(ipi->ipi_zone, inp);
636 * The calling convention of in_setsockaddr() and in_setpeeraddr() was
637 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
638 * in struct pr_usrreqs, so that protocols can just reference then directly
639 * without the need for a wrapper function. The socket must have a valid
640 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
641 * except through a kernel programming error, so it is acceptable to panic
642 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
643 * because there actually /is/ a programming error somewhere... XXX)
646 in_setsockaddr(struct socket *so, struct sockaddr **nam)
649 struct sockaddr_in *sin;
652 * Do the malloc first in case it blocks.
654 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
656 sin->sin_family = AF_INET;
657 sin->sin_len = sizeof *sin;
663 kfree(sin, M_SONAME);
666 sin->sin_port = inp->inp_lport;
667 sin->sin_addr = inp->inp_laddr;
670 *nam = (struct sockaddr *)sin;
675 in_setpeeraddr(struct socket *so, struct sockaddr **nam)
678 struct sockaddr_in *sin;
681 * Do the malloc first in case it blocks.
683 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
685 sin->sin_family = AF_INET;
686 sin->sin_len = sizeof *sin;
692 kfree(sin, M_SONAME);
695 sin->sin_port = inp->inp_fport;
696 sin->sin_addr = inp->inp_faddr;
699 *nam = (struct sockaddr *)sin;
704 in_pcbnotifyall(struct inpcbhead *head, struct in_addr faddr, int errno,
705 void (*notify)(struct inpcb *, int))
707 struct inpcb *inp, *ninp;
710 * note: if INP_PLACEMARKER is set we must ignore the rest of
711 * the structure and skip it.
714 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) {
715 if (inp->inp_flags & INP_PLACEMARKER)
718 if (!(inp->inp_vflag & INP_IPV4))
721 if (inp->inp_faddr.s_addr != faddr.s_addr ||
722 inp->inp_socket == NULL)
724 (*notify)(inp, errno); /* can remove inp from list! */
730 in_pcbpurgeif0(struct inpcb *head, struct ifnet *ifp)
733 struct ip_moptions *imo;
736 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) {
737 if (inp->inp_flags & INP_PLACEMARKER)
739 imo = inp->inp_moptions;
740 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) {
742 * Unselect the outgoing interface if it is being
745 if (imo->imo_multicast_ifp == ifp)
746 imo->imo_multicast_ifp = NULL;
749 * Drop multicast group membership if we joined
750 * through the interface being detached.
752 for (i = 0, gap = 0; i < imo->imo_num_memberships;
754 if (imo->imo_membership[i]->inm_ifp == ifp) {
755 in_delmulti(imo->imo_membership[i]);
758 imo->imo_membership[i - gap] =
759 imo->imo_membership[i];
761 imo->imo_num_memberships -= gap;
767 * Check for alternatives when higher level complains
768 * about service problems. For now, invalidate cached
769 * routing information. If the route was created dynamically
770 * (by a redirect), time to try a default gateway again.
773 in_losing(struct inpcb *inp)
776 struct rt_addrinfo rtinfo;
778 if ((rt = inp->inp_route.ro_rt)) {
779 bzero(&rtinfo, sizeof(struct rt_addrinfo));
780 rtinfo.rti_info[RTAX_DST] = rt_key(rt);
781 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
782 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt);
783 rtinfo.rti_flags = rt->rt_flags;
784 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0);
785 if (rt->rt_flags & RTF_DYNAMIC)
786 rtrequest1_global(RTM_DELETE, &rtinfo, NULL, NULL);
787 inp->inp_route.ro_rt = NULL;
790 * A new route can be allocated
791 * the next time output is attempted.
797 * After a routing change, flush old routing
798 * and allocate a (hopefully) better one.
801 in_rtchange(struct inpcb *inp, int errno)
803 if (inp->inp_route.ro_rt) {
804 rtfree(inp->inp_route.ro_rt);
805 inp->inp_route.ro_rt = NULL;
807 * A new route can be allocated the next time
808 * output is attempted.
814 * Lookup a PCB based on the local address and port.
817 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr, u_int lport_arg,
821 int matchwild = 3, wildcard;
822 u_short lport = lport_arg;
824 struct inpcbporthead *porthash;
825 struct inpcbport *phd;
826 struct inpcb *match = NULL;
829 * Best fit PCB lookup.
831 * First see if this local port is in use by looking on the
834 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
835 pcbinfo->porthashmask)];
836 LIST_FOREACH(phd, porthash, phd_hash) {
837 if (phd->phd_port == lport)
842 * Port is in use by one or more PCBs. Look for best
845 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
848 if ((inp->inp_vflag & INP_IPV4) == 0)
851 if (inp->inp_faddr.s_addr != INADDR_ANY)
853 if (inp->inp_laddr.s_addr != INADDR_ANY) {
854 if (laddr.s_addr == INADDR_ANY)
856 else if (inp->inp_laddr.s_addr != laddr.s_addr)
859 if (laddr.s_addr != INADDR_ANY)
862 if (wildcard && !wild_okay)
864 if (wildcard < matchwild) {
866 matchwild = wildcard;
867 if (matchwild == 0) {
877 * Lookup PCB in hash list.
880 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg,
881 struct in_addr laddr, u_int lport_arg, boolean_t wildcard,
884 struct inpcbhead *head;
886 u_short fport = fport_arg, lport = lport_arg;
889 * First look for an exact match.
891 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport,
892 laddr.s_addr, lport, pcbinfo->hashmask)];
893 LIST_FOREACH(inp, head, inp_hash) {
895 if (!(inp->inp_vflag & INP_IPV4))
898 if (in_hosteq(inp->inp_faddr, faddr) &&
899 in_hosteq(inp->inp_laddr, laddr) &&
900 inp->inp_fport == fport && inp->inp_lport == lport) {
907 struct inpcb *local_wild = NULL;
909 struct inpcb *local_wild_mapped = NULL;
911 struct inpcontainer *ic;
912 struct inpcontainerhead *chead;
914 chead = &pcbinfo->wildcardhashbase[
915 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)];
916 LIST_FOREACH(ic, chead, ic_list) {
919 if (!(inp->inp_vflag & INP_IPV4))
922 if (inp->inp_lport == lport) {
923 if (ifp && ifp->if_type == IFT_FAITH &&
924 !(inp->inp_flags & INP_FAITH))
926 if (inp->inp_laddr.s_addr == laddr.s_addr)
928 if (inp->inp_laddr.s_addr == INADDR_ANY) {
930 if (INP_CHECK_SOCKAF(inp->inp_socket,
932 local_wild_mapped = inp;
940 if (local_wild == NULL)
941 return (local_wild_mapped);
953 * Insert PCB into connection hash table.
956 in_pcbinsconnhash(struct inpcb *inp)
958 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo;
959 struct inpcbhead *bucket;
960 u_int32_t hashkey_faddr, hashkey_laddr;
963 if (inp->inp_vflag & INP_IPV6) {
964 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */;
965 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */;
968 hashkey_faddr = inp->inp_faddr.s_addr;
969 hashkey_laddr = inp->inp_laddr.s_addr;
974 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list"));
975 inp->inp_flags |= INP_CONNECTED;
978 * Insert into the connection hash table.
980 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr,
981 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)];
982 LIST_INSERT_HEAD(bucket, inp, inp_hash);
986 * Remove PCB from connection hash table.
989 in_pcbremconnhash(struct inpcb *inp)
991 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected"));
992 LIST_REMOVE(inp, inp_hash);
993 inp->inp_flags &= ~INP_CONNECTED;
997 * Insert PCB into port hash table.
1000 in_pcbinsporthash(struct inpcb *inp)
1002 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1003 struct inpcbporthead *pcbporthash;
1004 struct inpcbport *phd;
1007 * Insert into the port hash table.
1009 pcbporthash = &pcbinfo->porthashbase[
1010 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)];
1012 /* Go through port list and look for a head for this lport. */
1013 LIST_FOREACH(phd, pcbporthash, phd_hash)
1014 if (phd->phd_port == inp->inp_lport)
1017 /* If none exists, malloc one and tack it on. */
1019 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport),
1020 M_PCB, M_INTWAIT | M_NULLOK);
1022 return (ENOBUFS); /* XXX */
1023 phd->phd_port = inp->inp_lport;
1024 LIST_INIT(&phd->phd_pcblist);
1025 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash);
1029 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist);
1035 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1037 struct inpcontainer *ic;
1038 struct inpcontainerhead *bucket;
1040 bucket = &pcbinfo->wildcardhashbase[
1041 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1043 ic = kmalloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT);
1045 LIST_INSERT_HEAD(bucket, ic, ic_list);
1049 * Insert PCB into wildcard hash table.
1052 in_pcbinswildcardhash(struct inpcb *inp)
1054 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1056 KKASSERT(pcbinfo != NULL);
1058 in_pcbinswildcardhash_oncpu(inp, pcbinfo);
1059 inp->inp_flags |= INP_WILDCARD;
1063 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1065 struct inpcontainer *ic;
1066 struct inpcontainerhead *head;
1069 head = &pcbinfo->wildcardhashbase[
1070 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1072 LIST_FOREACH(ic, head, ic_list) {
1073 if (ic->ic_inp == inp)
1076 return; /* not found! */
1079 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */
1080 kfree(ic, M_TEMP); /* deallocate container */
1084 * Remove PCB from wildcard hash table.
1087 in_pcbremwildcardhash(struct inpcb *inp)
1089 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1091 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard"));
1092 in_pcbremwildcardhash_oncpu(inp, pcbinfo);
1093 inp->inp_flags &= ~INP_WILDCARD;
1097 * Remove PCB from various lists.
1100 in_pcbremlists(struct inpcb *inp)
1102 if (inp->inp_lport) {
1103 struct inpcbport *phd = inp->inp_phd;
1105 LIST_REMOVE(inp, inp_portlist);
1106 if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
1107 LIST_REMOVE(phd, phd_hash);
1111 if (inp->inp_flags & INP_WILDCARD) {
1112 in_pcbremwildcardhash(inp);
1113 } else if (inp->inp_flags & INP_CONNECTED) {
1114 in_pcbremconnhash(inp);
1116 LIST_REMOVE(inp, inp_list);
1117 inp->inp_pcbinfo->ipi_count--;
1121 prison_xinpcb(struct thread *td, struct inpcb *inp)
1125 if (td->td_proc == NULL)
1127 cr = td->td_proc->p_ucred;
1128 if (cr->cr_prison == NULL)
1130 if (ntohl(inp->inp_laddr.s_addr) == cr->cr_prison->pr_ip)
1136 in_pcblist_global(SYSCTL_HANDLER_ARGS)
1138 struct inpcbinfo *pcbinfo = arg1;
1139 struct inpcb *inp, *marker;
1145 * The process of preparing the TCB list is too time-consuming and
1146 * resource-intensive to repeat twice on every request.
1148 if (req->oldptr == NULL) {
1149 n = pcbinfo->ipi_count;
1150 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb);
1154 if (req->newptr != NULL)
1158 * OK, now we're committed to doing something. Re-fetch ipi_count
1159 * after obtaining the generation count.
1161 gencnt = pcbinfo->ipi_gencnt;
1162 n = pcbinfo->ipi_count;
1164 marker = kmalloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO);
1165 marker->inp_flags |= INP_PLACEMARKER;
1166 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
1171 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) {
1172 LIST_REMOVE(marker, inp_list);
1173 LIST_INSERT_AFTER(inp, marker, inp_list);
1175 if (inp->inp_flags & INP_PLACEMARKER)
1177 if (inp->inp_gencnt > gencnt)
1179 if (prison_xinpcb(req->td, inp))
1181 bzero(&xi, sizeof xi);
1182 xi.xi_len = sizeof xi;
1183 bcopy(inp, &xi.xi_inp, sizeof *inp);
1184 if (inp->inp_socket)
1185 sotoxsocket(inp->inp_socket, &xi.xi_socket);
1186 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0)
1190 LIST_REMOVE(marker, inp_list);
1191 if (error == 0 && i < n) {
1192 bzero(&xi, sizeof xi);
1193 xi.xi_len = sizeof xi;
1195 error = SYSCTL_OUT(req, &xi, sizeof xi);
1199 kfree(marker, M_TEMP);