Add <sys/_null.h> for the definition of NULL:
[dragonfly.git] / sys / netinet / in_pcb.c
blobafaa40f5c444a0cb0343f63a8f38466b38936d7a
1 /*
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
6 * by Jeffrey M. Hsu.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
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
31 * SUCH DAMAGE.
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
40 * are met:
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
64 * SUCH DAMAGE.
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 $
68 * $DragonFly: src/sys/netinet/in_pcb.c,v 1.45 2008/03/26 14:44:59 sephe Exp $
71 #include "opt_ipsec.h"
72 #include "opt_inet6.h"
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/malloc.h>
77 #include <sys/mbuf.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/socket.h>
81 #include <sys/socketvar.h>
82 #include <sys/proc.h>
83 #include <sys/jail.h>
84 #include <sys/kernel.h>
85 #include <sys/sysctl.h>
86 #include <sys/thread2.h>
88 #include <machine/limits.h>
90 #include <vm/vm_zone.h>
92 #include <net/if.h>
93 #include <net/if_types.h>
94 #include <net/route.h>
96 #include <netinet/in.h>
97 #include <netinet/in_pcb.h>
98 #include <netinet/in_var.h>
99 #include <netinet/ip_var.h>
100 #ifdef INET6
101 #include <netinet/ip6.h>
102 #include <netinet6/ip6_var.h>
103 #endif /* INET6 */
105 #ifdef IPSEC
106 #include <netinet6/ipsec.h>
107 #include <netproto/key/key.h>
108 #endif
110 #ifdef FAST_IPSEC
111 #if defined(IPSEC) || defined(IPSEC_ESP)
112 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!"
113 #endif
115 #include <netproto/ipsec/ipsec.h>
116 #include <netproto/ipsec/key.h>
117 #define IPSEC
118 #endif /* FAST_IPSEC */
120 struct in_addr zeroin_addr;
123 * These configure the range of local port addresses assigned to
124 * "unspecified" outgoing connections/packets/whatever.
126 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */
127 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */
129 int ipport_firstauto = IPPORT_RESERVED; /* 1024 */
130 int ipport_lastauto = IPPORT_USERRESERVED; /* 5000 */
132 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */
133 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */
135 static __inline void
136 RANGECHK(int var, int min, int max)
138 if (var < min)
139 var = min;
140 else if (var > max)
141 var = max;
144 static int
145 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS)
147 int error;
149 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req);
150 if (!error) {
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);
160 return (error);
163 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports");
165 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW,
166 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", "");
167 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW,
168 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", "");
169 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW,
170 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", "");
171 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW,
172 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", "");
173 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW,
174 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", "");
175 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW,
176 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", "");
179 * in_pcb.c: manage the Protocol Control Blocks.
181 * NOTE: It is assumed that most of these functions will be called from
182 * a critical section. XXX - There are, unfortunately, a few exceptions
183 * to this rule that should be fixed.
185 * NOTE: The caller should initialize the cpu field to the cpu running the
186 * protocol stack associated with this inpcbinfo.
189 void
190 in_pcbinfo_init(struct inpcbinfo *pcbinfo)
192 LIST_INIT(&pcbinfo->pcblisthead);
193 pcbinfo->cpu = -1;
197 * Allocate a PCB and associate it with the socket.
200 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo)
202 struct inpcb *inp;
203 #ifdef IPSEC
204 int error;
205 #endif
207 inp = zalloc(pcbinfo->ipi_zone);
208 if (inp == NULL)
209 return (ENOBUFS);
210 bzero(inp, sizeof *inp);
211 inp->inp_gencnt = ++pcbinfo->ipi_gencnt;
212 inp->inp_pcbinfo = inp->inp_cpcbinfo = pcbinfo;
213 inp->inp_socket = so;
214 #ifdef IPSEC
215 error = ipsec_init_policy(so, &inp->inp_sp);
216 if (error != 0) {
217 zfree(pcbinfo->ipi_zone, inp);
218 return (error);
220 #endif
221 #ifdef INET6
222 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only)
223 inp->inp_flags |= IN6P_IPV6_V6ONLY;
224 if (ip6_auto_flowlabel)
225 inp->inp_flags |= IN6P_AUTOFLOWLABEL;
226 #endif
227 so->so_pcb = inp;
228 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list);
229 pcbinfo->ipi_count++;
230 return (0);
234 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
236 struct socket *so = inp->inp_socket;
237 struct proc *p = td->td_proc;
238 unsigned short *lastport;
239 struct sockaddr_in *sin;
240 struct sockaddr_in jsin;
241 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
242 struct ucred *cred = NULL;
243 u_short lport = 0;
244 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
245 int error;
247 KKASSERT(p);
249 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */
250 return (EADDRNOTAVAIL);
251 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY)
252 return (EINVAL); /* already bound */
253 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT)))
254 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */
255 if (p)
256 cred = p->p_ucred;
257 if (nam != NULL) {
258 sin = (struct sockaddr_in *)nam;
259 if (nam->sa_len != sizeof *sin)
260 return (EINVAL);
261 #ifdef notdef
263 * We should check the family, but old programs
264 * incorrectly fail to initialize it.
266 if (sin->sin_family != AF_INET)
267 return (EAFNOSUPPORT);
268 #endif
269 if (!prison_replace_wildcards(td, nam))
270 return (EINVAL);
271 lport = sin->sin_port;
272 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
274 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
275 * allow complete duplication of binding if
276 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
277 * and a multicast address is bound on both
278 * new and duplicated sockets.
280 if (so->so_options & SO_REUSEADDR)
281 reuseport = SO_REUSEADDR | SO_REUSEPORT;
282 } else if (sin->sin_addr.s_addr != INADDR_ANY) {
283 sin->sin_port = 0; /* yech... */
284 bzero(&sin->sin_zero, sizeof sin->sin_zero);
285 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL)
286 return (EADDRNOTAVAIL);
288 if (lport != 0) {
289 struct inpcb *t;
291 /* GROSS */
292 if (ntohs(lport) < IPPORT_RESERVED &&
293 cred && suser_cred(cred, PRISON_ROOT))
294 return (EACCES);
295 if (so->so_cred->cr_uid != 0 &&
296 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) {
297 t = in_pcblookup_local(inp->inp_pcbinfo,
298 sin->sin_addr, lport,
299 INPLOOKUP_WILDCARD, cred);
300 if (t &&
301 (!in_nullhost(sin->sin_addr) ||
302 !in_nullhost(t->inp_laddr) ||
303 (t->inp_socket->so_options &
304 SO_REUSEPORT) == 0) &&
305 (so->so_cred->cr_uid !=
306 t->inp_socket->so_cred->cr_uid)) {
307 #ifdef INET6
308 if (!in_nullhost(sin->sin_addr) ||
309 !in_nullhost(t->inp_laddr) ||
310 INP_SOCKAF(so) ==
311 INP_SOCKAF(t->inp_socket))
312 #endif
313 return (EADDRINUSE);
316 if (cred && !prison_replace_wildcards(td, nam))
317 return (EADDRNOTAVAIL);
318 t = in_pcblookup_local(pcbinfo, sin->sin_addr, lport,
319 wild, cred);
320 if (t && !(reuseport & t->inp_socket->so_options)) {
321 #ifdef INET6
322 if (!in_nullhost(sin->sin_addr) ||
323 !in_nullhost(t->inp_laddr) ||
324 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))
325 #endif
326 return (EADDRINUSE);
329 inp->inp_laddr = sin->sin_addr;
331 if (lport == 0) {
332 ushort first, last;
333 int count;
335 jsin.sin_family = AF_INET;
336 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr;
337 if (!prison_replace_wildcards(td, (struct sockaddr *)&jsin)) {
338 inp->inp_laddr.s_addr = INADDR_ANY;
339 return (EINVAL);
341 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr;
343 inp->inp_flags |= INP_ANONPORT;
345 if (inp->inp_flags & INP_HIGHPORT) {
346 first = ipport_hifirstauto; /* sysctl */
347 last = ipport_hilastauto;
348 lastport = &pcbinfo->lasthi;
349 } else if (inp->inp_flags & INP_LOWPORT) {
350 if (cred &&
351 (error = suser_cred(cred, PRISON_ROOT))) {
352 inp->inp_laddr.s_addr = INADDR_ANY;
353 return (error);
355 first = ipport_lowfirstauto; /* 1023 */
356 last = ipport_lowlastauto; /* 600 */
357 lastport = &pcbinfo->lastlow;
358 } else {
359 first = ipport_firstauto; /* sysctl */
360 last = ipport_lastauto;
361 lastport = &pcbinfo->lastport;
364 * Simple check to ensure all ports are not used up causing
365 * a deadlock here.
367 * We split the two cases (up and down) so that the direction
368 * is not being tested on each round of the loop.
370 if (first > last) {
372 * counting down
374 count = first - last;
376 do {
377 if (count-- < 0) { /* completely used? */
378 inp->inp_laddr.s_addr = INADDR_ANY;
379 return (EADDRNOTAVAIL);
381 --*lastport;
382 if (*lastport > first || *lastport < last)
383 *lastport = first;
384 lport = htons(*lastport);
385 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr,
386 lport, wild, cred));
387 } else {
389 * counting up
391 count = last - first;
393 do {
394 if (count-- < 0) { /* completely used? */
395 inp->inp_laddr.s_addr = INADDR_ANY;
396 return (EADDRNOTAVAIL);
398 ++*lastport;
399 if (*lastport < first || *lastport > last)
400 *lastport = first;
401 lport = htons(*lastport);
402 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr,
403 lport, wild, cred));
406 inp->inp_lport = lport;
408 jsin.sin_family = AF_INET;
409 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr;
410 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin)) {
411 inp->inp_laddr.s_addr = INADDR_ANY;
412 inp->inp_lport = 0;
413 return (EINVAL);
415 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr;
417 if (in_pcbinsporthash(inp) != 0) {
418 inp->inp_laddr.s_addr = INADDR_ANY;
419 inp->inp_lport = 0;
420 return (EAGAIN);
422 return (0);
426 * Transform old in_pcbconnect() into an inner subroutine for new
427 * in_pcbconnect(): Do some validity-checking on the remote
428 * address (in mbuf 'nam') and then determine local host address
429 * (i.e., which interface) to use to access that remote host.
431 * This preserves definition of in_pcbconnect(), while supporting a
432 * slightly different version for T/TCP. (This is more than
433 * a bit of a kludge, but cleaning up the internal interfaces would
434 * have forced minor changes in every protocol).
437 in_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
438 struct sockaddr_in **plocal_sin, struct thread *td)
440 struct in_ifaddr *ia;
441 struct ucred *cred = NULL;
442 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
443 struct sockaddr *jsin;
444 int jailed = 0, alloc_route = 0;
446 if (nam->sa_len != sizeof *sin)
447 return (EINVAL);
448 if (sin->sin_family != AF_INET)
449 return (EAFNOSUPPORT);
450 if (sin->sin_port == 0)
451 return (EADDRNOTAVAIL);
452 if (td && td->td_proc && td->td_proc->p_ucred)
453 cred = td->td_proc->p_ucred;
454 if (cred && cred->cr_prison)
455 jailed = 1;
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) {
472 struct route *ro;
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;
482 if (ro->ro_rt &&
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)) {
488 RTFREE(ro->ro_rt);
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 =
499 sin->sin_addr;
500 rtalloc(ro);
501 alloc_route = 1;
504 * If we found a route, use the address
505 * corresponding to the outgoing interface
506 * unless it is the loopback (in case a route
507 * to our address on another net goes to loopback).
509 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) {
510 if (jailed) {
511 if (jailed_ip(cred->cr_prison,
512 ro->ro_rt->rt_ifa->ifa_addr)) {
513 ia = ifatoia(ro->ro_rt->rt_ifa);
515 } else {
516 ia = ifatoia(ro->ro_rt->rt_ifa);
519 if (ia == NULL) {
520 u_short fport = sin->sin_port;
522 sin->sin_port = 0;
523 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin)));
524 if (ia && jailed && !jailed_ip(cred->cr_prison,
525 sintosa(&ia->ia_addr)))
526 ia = NULL;
527 if (ia == NULL)
528 ia = ifatoia(ifa_ifwithnet(sintosa(sin)));
529 if (ia && jailed && !jailed_ip(cred->cr_prison,
530 sintosa(&ia->ia_addr)))
531 ia = NULL;
532 sin->sin_port = fport;
533 if (ia == NULL)
534 ia = TAILQ_FIRST(&in_ifaddrhead);
535 if (ia && jailed && !jailed_ip(cred->cr_prison,
536 sintosa(&ia->ia_addr)))
537 ia = NULL;
539 if (!jailed && ia == NULL)
540 goto fail;
543 * If the destination address is multicast and an outgoing
544 * interface has been set as a multicast option, use the
545 * address of that interface as our source address.
547 if (!jailed && IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) &&
548 inp->inp_moptions != NULL) {
549 struct ip_moptions *imo;
550 struct ifnet *ifp;
552 imo = inp->inp_moptions;
553 if (imo->imo_multicast_ifp != NULL) {
554 ifp = imo->imo_multicast_ifp;
555 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link)
556 if (ia->ia_ifp == ifp)
557 break;
558 if (ia == NULL)
559 goto fail;
563 * Don't do pcblookup call here; return interface in plocal_sin
564 * and exit to caller, that will do the lookup.
566 if (ia == NULL && jailed) {
567 if ((jsin = prison_get_nonlocal(cred->cr_prison, AF_INET, NULL)) != NULL ||
568 (jsin = prison_get_local(cred->cr_prison, AF_INET, NULL)) != NULL) {
569 *plocal_sin = satosin(jsin);
570 } else {
571 /* IPv6 only Jail */
572 goto fail;
574 } else {
575 *plocal_sin = &ia->ia_addr;
578 return (0);
579 fail:
580 if (alloc_route) {
581 struct route *ro = &inp->inp_route;
583 if (ro->ro_rt != NULL)
584 RTFREE(ro->ro_rt);
585 bzero(ro, sizeof(*ro));
587 return (EADDRNOTAVAIL);
591 * Outer subroutine:
592 * Connect from a socket to a specified address.
593 * Both address and port must be specified in argument sin.
594 * If don't have a local address for this socket yet,
595 * then pick one.
598 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
600 struct sockaddr_in *if_sin;
601 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
602 int error;
604 /* Call inner routine to assign local interface address. */
605 if ((error = in_pcbladdr(inp, nam, &if_sin, td)) != 0)
606 return (error);
608 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port,
609 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr,
610 inp->inp_lport, FALSE, NULL) != NULL) {
611 return (EADDRINUSE);
613 if (inp->inp_laddr.s_addr == INADDR_ANY) {
614 if (inp->inp_lport == 0) {
615 error = in_pcbbind(inp, (struct sockaddr *)NULL, td);
616 if (error)
617 return (error);
619 inp->inp_laddr = if_sin->sin_addr;
621 inp->inp_faddr = sin->sin_addr;
622 inp->inp_fport = sin->sin_port;
623 in_pcbinsconnhash(inp);
624 return (0);
627 void
628 in_pcbdisconnect(struct inpcb *inp)
631 inp->inp_faddr.s_addr = INADDR_ANY;
632 inp->inp_fport = 0;
633 in_pcbremconnhash(inp);
634 if (inp->inp_socket->so_state & SS_NOFDREF)
635 in_pcbdetach(inp);
638 void
639 in_pcbdetach(struct inpcb *inp)
641 struct socket *so = inp->inp_socket;
642 struct inpcbinfo *ipi = inp->inp_pcbinfo;
644 #ifdef IPSEC
645 ipsec4_delete_pcbpolicy(inp);
646 #endif /*IPSEC*/
647 inp->inp_gencnt = ++ipi->ipi_gencnt;
648 in_pcbremlists(inp);
649 so->so_pcb = 0;
650 sofree(so);
651 if (inp->inp_options)
652 m_free(inp->inp_options);
653 if (inp->inp_route.ro_rt)
654 rtfree(inp->inp_route.ro_rt);
655 ip_freemoptions(inp->inp_moptions);
656 inp->inp_vflag = 0;
657 zfree(ipi->ipi_zone, inp);
661 * The calling convention of in_setsockaddr() and in_setpeeraddr() was
662 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
663 * in struct pr_usrreqs, so that protocols can just reference then directly
664 * without the need for a wrapper function. The socket must have a valid
665 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
666 * except through a kernel programming error, so it is acceptable to panic
667 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
668 * because there actually /is/ a programming error somewhere... XXX)
671 in_setsockaddr(struct socket *so, struct sockaddr **nam)
673 struct inpcb *inp;
674 struct sockaddr_in *sin;
677 * Do the malloc first in case it blocks.
679 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
680 M_WAITOK | M_ZERO);
681 sin->sin_family = AF_INET;
682 sin->sin_len = sizeof *sin;
684 crit_enter();
685 inp = so->so_pcb;
686 if (!inp) {
687 crit_exit();
688 kfree(sin, M_SONAME);
689 return (ECONNRESET);
691 sin->sin_port = inp->inp_lport;
692 sin->sin_addr = inp->inp_laddr;
693 crit_exit();
695 *nam = (struct sockaddr *)sin;
696 return (0);
700 in_setpeeraddr(struct socket *so, struct sockaddr **nam)
702 struct inpcb *inp;
703 struct sockaddr_in *sin;
706 * Do the malloc first in case it blocks.
708 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
709 M_WAITOK | M_ZERO);
710 sin->sin_family = AF_INET;
711 sin->sin_len = sizeof *sin;
713 crit_enter();
714 inp = so->so_pcb;
715 if (!inp) {
716 crit_exit();
717 kfree(sin, M_SONAME);
718 return (ECONNRESET);
720 sin->sin_port = inp->inp_fport;
721 sin->sin_addr = inp->inp_faddr;
722 crit_exit();
724 *nam = (struct sockaddr *)sin;
725 return (0);
728 void
729 in_pcbnotifyall(struct inpcbhead *head, struct in_addr faddr, int err,
730 void (*notify)(struct inpcb *, int))
732 struct inpcb *inp, *ninp;
735 * note: if INP_PLACEMARKER is set we must ignore the rest of
736 * the structure and skip it.
738 crit_enter();
739 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) {
740 if (inp->inp_flags & INP_PLACEMARKER)
741 continue;
742 #ifdef INET6
743 if (!(inp->inp_vflag & INP_IPV4))
744 continue;
745 #endif
746 if (inp->inp_faddr.s_addr != faddr.s_addr ||
747 inp->inp_socket == NULL)
748 continue;
749 (*notify)(inp, err); /* can remove inp from list! */
751 crit_exit();
754 void
755 in_pcbpurgeif0(struct inpcb *head, struct ifnet *ifp)
757 struct inpcb *inp;
758 struct ip_moptions *imo;
759 int i, gap;
761 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) {
762 if (inp->inp_flags & INP_PLACEMARKER)
763 continue;
764 imo = inp->inp_moptions;
765 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) {
767 * Unselect the outgoing interface if it is being
768 * detached.
770 if (imo->imo_multicast_ifp == ifp)
771 imo->imo_multicast_ifp = NULL;
774 * Drop multicast group membership if we joined
775 * through the interface being detached.
777 for (i = 0, gap = 0; i < imo->imo_num_memberships;
778 i++) {
779 if (imo->imo_membership[i]->inm_ifp == ifp) {
780 in_delmulti(imo->imo_membership[i]);
781 gap++;
782 } else if (gap != 0)
783 imo->imo_membership[i - gap] =
784 imo->imo_membership[i];
786 imo->imo_num_memberships -= gap;
792 * Check for alternatives when higher level complains
793 * about service problems. For now, invalidate cached
794 * routing information. If the route was created dynamically
795 * (by a redirect), time to try a default gateway again.
797 void
798 in_losing(struct inpcb *inp)
800 struct rtentry *rt;
801 struct rt_addrinfo rtinfo;
803 if ((rt = inp->inp_route.ro_rt)) {
804 bzero(&rtinfo, sizeof(struct rt_addrinfo));
805 rtinfo.rti_info[RTAX_DST] = rt_key(rt);
806 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
807 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt);
808 rtinfo.rti_flags = rt->rt_flags;
809 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0);
810 if (rt->rt_flags & RTF_DYNAMIC)
811 rtrequest1_global(RTM_DELETE, &rtinfo, NULL, NULL);
812 inp->inp_route.ro_rt = NULL;
813 rtfree(rt);
815 * A new route can be allocated
816 * the next time output is attempted.
822 * After a routing change, flush old routing
823 * and allocate a (hopefully) better one.
825 void
826 in_rtchange(struct inpcb *inp, int err)
828 if (inp->inp_route.ro_rt) {
829 rtfree(inp->inp_route.ro_rt);
830 inp->inp_route.ro_rt = NULL;
832 * A new route can be allocated the next time
833 * output is attempted.
839 * Lookup a PCB based on the local address and port.
841 struct inpcb *
842 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr,
843 u_int lport_arg, int wild_okay, struct ucred *cred)
845 struct inpcb *inp;
846 int matchwild = 3, wildcard;
847 u_short lport = lport_arg;
849 struct inpcbporthead *porthash;
850 struct inpcbport *phd;
851 struct inpcb *match = NULL;
854 * Best fit PCB lookup.
856 * First see if this local port is in use by looking on the
857 * port hash list.
859 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
860 pcbinfo->porthashmask)];
861 LIST_FOREACH(phd, porthash, phd_hash) {
862 if (phd->phd_port == lport)
863 break;
865 if (phd != NULL) {
867 * Port is in use by one or more PCBs. Look for best
868 * fit.
870 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
871 wildcard = 0;
872 #ifdef INET6
873 if ((inp->inp_vflag & INP_IPV4) == 0)
874 continue;
875 #endif
876 if (inp->inp_faddr.s_addr != INADDR_ANY)
877 wildcard++;
878 if (inp->inp_laddr.s_addr != INADDR_ANY) {
879 if (laddr.s_addr == INADDR_ANY)
880 wildcard++;
881 else if (inp->inp_laddr.s_addr != laddr.s_addr)
882 continue;
883 } else {
884 if (laddr.s_addr != INADDR_ANY)
885 wildcard++;
887 if (wildcard && !wild_okay)
888 continue;
889 if (wildcard < matchwild &&
890 (cred == NULL ||
891 cred->cr_prison ==
892 inp->inp_socket->so_cred->cr_prison)) {
893 match = inp;
894 matchwild = wildcard;
895 if (matchwild == 0) {
896 break;
901 return (match);
905 * Lookup PCB in hash list.
907 struct inpcb *
908 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg,
909 struct in_addr laddr, u_int lport_arg, boolean_t wildcard,
910 struct ifnet *ifp)
912 struct inpcbhead *head;
913 struct inpcb *inp, *jinp=NULL;
914 u_short fport = fport_arg, lport = lport_arg;
917 * First look for an exact match.
919 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport,
920 laddr.s_addr, lport, pcbinfo->hashmask)];
921 LIST_FOREACH(inp, head, inp_hash) {
922 #ifdef INET6
923 if (!(inp->inp_vflag & INP_IPV4))
924 continue;
925 #endif
926 if (in_hosteq(inp->inp_faddr, faddr) &&
927 in_hosteq(inp->inp_laddr, laddr) &&
928 inp->inp_fport == fport && inp->inp_lport == lport) {
929 /* found */
930 if (inp->inp_socket == NULL ||
931 inp->inp_socket->so_cred->cr_prison == NULL) {
932 return (inp);
933 } else {
934 if (jinp == NULL)
935 jinp = inp;
939 if (jinp != NULL)
940 return (jinp);
941 if (wildcard) {
942 struct inpcb *local_wild = NULL;
943 struct inpcb *jinp_wild = NULL;
944 #ifdef INET6
945 struct inpcb *local_wild_mapped = NULL;
946 #endif
947 struct inpcontainer *ic;
948 struct inpcontainerhead *chead;
949 struct sockaddr_in jsin;
950 struct ucred *cred;
953 * Order of socket selection:
954 * 1. non-jailed, non-wild.
955 * 2. non-jailed, wild.
956 * 3. jailed, non-wild.
957 * 4. jailed, wild.
959 jsin.sin_family = AF_INET;
960 chead = &pcbinfo->wildcardhashbase[
961 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)];
962 LIST_FOREACH(ic, chead, ic_list) {
963 inp = ic->ic_inp;
964 jsin.sin_addr.s_addr = laddr.s_addr;
965 #ifdef INET6
966 if (!(inp->inp_vflag & INP_IPV4))
967 continue;
968 #endif
969 if (inp->inp_socket != NULL)
970 cred = inp->inp_socket->so_cred;
971 else
972 cred = NULL;
973 if (cred != NULL && jailed(cred)) {
974 if (jinp != NULL)
975 continue;
976 else
977 if (!jailed_ip(cred->cr_prison,
978 (struct sockaddr *)&jsin))
979 continue;
981 if (inp->inp_lport == lport) {
982 if (ifp && ifp->if_type == IFT_FAITH &&
983 !(inp->inp_flags & INP_FAITH))
984 continue;
985 if (inp->inp_laddr.s_addr == laddr.s_addr) {
986 if (cred != NULL && jailed(cred))
987 jinp = inp;
988 else
989 return (inp);
991 if (inp->inp_laddr.s_addr == INADDR_ANY) {
992 #ifdef INET6
993 if (INP_CHECK_SOCKAF(inp->inp_socket,
994 AF_INET6))
995 local_wild_mapped = inp;
996 else
997 #endif
998 if (cred != NULL &&
999 jailed(cred))
1000 jinp_wild = inp;
1001 else
1002 local_wild = inp;
1006 if (local_wild != NULL)
1007 return (local_wild);
1008 #ifdef INET6
1009 if (local_wild_mapped != NULL)
1010 return (local_wild_mapped);
1011 #endif
1012 if (jinp != NULL)
1013 return (jinp);
1014 return (jinp_wild);
1018 * Not found.
1020 return (NULL);
1024 * Insert PCB into connection hash table.
1026 void
1027 in_pcbinsconnhash(struct inpcb *inp)
1029 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo;
1030 struct inpcbhead *bucket;
1031 u_int32_t hashkey_faddr, hashkey_laddr;
1033 #ifdef INET6
1034 if (inp->inp_vflag & INP_IPV6) {
1035 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */;
1036 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */;
1037 } else {
1038 #endif
1039 hashkey_faddr = inp->inp_faddr.s_addr;
1040 hashkey_laddr = inp->inp_laddr.s_addr;
1041 #ifdef INET6
1043 #endif
1045 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list"));
1046 inp->inp_flags |= INP_CONNECTED;
1049 * Insert into the connection hash table.
1051 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr,
1052 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)];
1053 LIST_INSERT_HEAD(bucket, inp, inp_hash);
1057 * Remove PCB from connection hash table.
1059 void
1060 in_pcbremconnhash(struct inpcb *inp)
1062 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected"));
1063 LIST_REMOVE(inp, inp_hash);
1064 inp->inp_flags &= ~INP_CONNECTED;
1068 * Insert PCB into port hash table.
1071 in_pcbinsporthash(struct inpcb *inp)
1073 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1074 struct inpcbporthead *pcbporthash;
1075 struct inpcbport *phd;
1078 * Insert into the port hash table.
1080 pcbporthash = &pcbinfo->porthashbase[
1081 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)];
1083 /* Go through port list and look for a head for this lport. */
1084 LIST_FOREACH(phd, pcbporthash, phd_hash)
1085 if (phd->phd_port == inp->inp_lport)
1086 break;
1088 /* If none exists, malloc one and tack it on. */
1089 if (phd == NULL) {
1090 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport),
1091 M_PCB, M_INTWAIT | M_NULLOK);
1092 if (phd == NULL)
1093 return (ENOBUFS); /* XXX */
1094 phd->phd_port = inp->inp_lport;
1095 LIST_INIT(&phd->phd_pcblist);
1096 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash);
1099 inp->inp_phd = phd;
1100 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist);
1102 return (0);
1105 void
1106 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1108 struct inpcontainer *ic;
1109 struct inpcontainerhead *bucket;
1111 bucket = &pcbinfo->wildcardhashbase[
1112 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1114 ic = kmalloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT);
1115 ic->ic_inp = inp;
1116 LIST_INSERT_HEAD(bucket, ic, ic_list);
1120 * Insert PCB into wildcard hash table.
1122 void
1123 in_pcbinswildcardhash(struct inpcb *inp)
1125 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1127 KKASSERT(pcbinfo != NULL);
1129 in_pcbinswildcardhash_oncpu(inp, pcbinfo);
1130 inp->inp_flags |= INP_WILDCARD;
1133 void
1134 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo)
1136 struct inpcontainer *ic;
1137 struct inpcontainerhead *head;
1139 /* find bucket */
1140 head = &pcbinfo->wildcardhashbase[
1141 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)];
1143 LIST_FOREACH(ic, head, ic_list) {
1144 if (ic->ic_inp == inp)
1145 goto found;
1147 return; /* not found! */
1149 found:
1150 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */
1151 kfree(ic, M_TEMP); /* deallocate container */
1155 * Remove PCB from wildcard hash table.
1157 void
1158 in_pcbremwildcardhash(struct inpcb *inp)
1160 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
1162 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard"));
1163 in_pcbremwildcardhash_oncpu(inp, pcbinfo);
1164 inp->inp_flags &= ~INP_WILDCARD;
1168 * Remove PCB from various lists.
1170 void
1171 in_pcbremlists(struct inpcb *inp)
1173 if (inp->inp_lport) {
1174 struct inpcbport *phd = inp->inp_phd;
1176 LIST_REMOVE(inp, inp_portlist);
1177 if (LIST_FIRST(&phd->phd_pcblist) == NULL) {
1178 LIST_REMOVE(phd, phd_hash);
1179 kfree(phd, M_PCB);
1182 if (inp->inp_flags & INP_WILDCARD) {
1183 in_pcbremwildcardhash(inp);
1184 } else if (inp->inp_flags & INP_CONNECTED) {
1185 in_pcbremconnhash(inp);
1187 LIST_REMOVE(inp, inp_list);
1188 inp->inp_pcbinfo->ipi_count--;
1192 prison_xinpcb(struct thread *td, struct inpcb *inp)
1194 struct ucred *cr;
1196 if (td->td_proc == NULL)
1197 return (0);
1198 cr = td->td_proc->p_ucred;
1199 if (cr->cr_prison == NULL)
1200 return (0);
1201 if (inp->inp_socket && inp->inp_socket->so_cred &&
1202 inp->inp_socket->so_cred->cr_prison &&
1203 cr->cr_prison == inp->inp_socket->so_cred->cr_prison)
1204 return (0);
1205 return (1);
1209 in_pcblist_global(SYSCTL_HANDLER_ARGS)
1211 struct inpcbinfo *pcbinfo = arg1;
1212 struct inpcb *inp, *marker;
1213 struct xinpcb xi;
1214 int error, i, n;
1215 inp_gen_t gencnt;
1218 * The process of preparing the TCB list is too time-consuming and
1219 * resource-intensive to repeat twice on every request.
1221 if (req->oldptr == NULL) {
1222 n = pcbinfo->ipi_count;
1223 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb);
1224 return 0;
1227 if (req->newptr != NULL)
1228 return EPERM;
1231 * OK, now we're committed to doing something. Re-fetch ipi_count
1232 * after obtaining the generation count.
1234 gencnt = pcbinfo->ipi_gencnt;
1235 n = pcbinfo->ipi_count;
1237 marker = kmalloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO);
1238 marker->inp_flags |= INP_PLACEMARKER;
1239 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
1241 i = 0;
1242 error = 0;
1244 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) {
1245 LIST_REMOVE(marker, inp_list);
1246 LIST_INSERT_AFTER(inp, marker, inp_list);
1248 if (inp->inp_flags & INP_PLACEMARKER)
1249 continue;
1250 if (inp->inp_gencnt > gencnt)
1251 continue;
1252 if (prison_xinpcb(req->td, inp))
1253 continue;
1254 bzero(&xi, sizeof xi);
1255 xi.xi_len = sizeof xi;
1256 bcopy(inp, &xi.xi_inp, sizeof *inp);
1257 if (inp->inp_socket)
1258 sotoxsocket(inp->inp_socket, &xi.xi_socket);
1259 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0)
1260 break;
1261 ++i;
1263 LIST_REMOVE(marker, inp_list);
1264 if (error == 0 && i < n) {
1265 bzero(&xi, sizeof xi);
1266 xi.xi_len = sizeof xi;
1267 while (i < n) {
1268 error = SYSCTL_OUT(req, &xi, sizeof xi);
1269 ++i;
1272 kfree(marker, M_TEMP);
1273 return(error);