kern_jail.c revision 185404
1/*-
2 * Copyright (c) 1999 Poul-Henning Kamp. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
6 * are met:
7 * 1. Redistributions of source code must retain the above copyright
8 *    notice, this list of conditions and the following disclaimer.
9 * 2. Redistributions in binary form must reproduce the above copyright
10 *    notice, this list of conditions and the following disclaimer in the
11 *    documentation and/or other materials provided with the distribution.
12 *
13 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
14 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
15 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
16 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
17 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
18 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
19 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
20 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
21 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
22 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
23 * SUCH DAMAGE.
24 */
25
26#include <sys/cdefs.h>
27__FBSDID("$FreeBSD: head/sys/kern/kern_jail.c 185404 2008-11-28 19:23:46Z bz $");
28
29#include "opt_mac.h"
30
31#include <sys/param.h>
32#include <sys/types.h>
33#include <sys/kernel.h>
34#include <sys/systm.h>
35#include <sys/errno.h>
36#include <sys/sysproto.h>
37#include <sys/malloc.h>
38#include <sys/priv.h>
39#include <sys/proc.h>
40#include <sys/taskqueue.h>
41#include <sys/fcntl.h>
42#include <sys/jail.h>
43#include <sys/lock.h>
44#include <sys/mutex.h>
45#include <sys/sx.h>
46#include <sys/namei.h>
47#include <sys/mount.h>
48#include <sys/queue.h>
49#include <sys/socket.h>
50#include <sys/syscallsubr.h>
51#include <sys/sysctl.h>
52#include <sys/vnode.h>
53#include <sys/vimage.h>
54#include <sys/osd.h>
55#include <net/if.h>
56#include <netinet/in.h>
57
58#include <security/mac/mac_framework.h>
59
60MALLOC_DEFINE(M_PRISON, "prison", "Prison structures");
61
62SYSCTL_NODE(_security, OID_AUTO, jail, CTLFLAG_RW, 0,
63    "Jail rules");
64
65int	jail_set_hostname_allowed = 1;
66SYSCTL_INT(_security_jail, OID_AUTO, set_hostname_allowed, CTLFLAG_RW,
67    &jail_set_hostname_allowed, 0,
68    "Processes in jail can set their hostnames");
69
70int	jail_socket_unixiproute_only = 1;
71SYSCTL_INT(_security_jail, OID_AUTO, socket_unixiproute_only, CTLFLAG_RW,
72    &jail_socket_unixiproute_only, 0,
73    "Processes in jail are limited to creating UNIX/IPv4/route sockets only");
74
75int	jail_sysvipc_allowed = 0;
76SYSCTL_INT(_security_jail, OID_AUTO, sysvipc_allowed, CTLFLAG_RW,
77    &jail_sysvipc_allowed, 0,
78    "Processes in jail can use System V IPC primitives");
79
80static int jail_enforce_statfs = 2;
81SYSCTL_INT(_security_jail, OID_AUTO, enforce_statfs, CTLFLAG_RW,
82    &jail_enforce_statfs, 0,
83    "Processes in jail cannot see all mounted file systems");
84
85int	jail_allow_raw_sockets = 0;
86SYSCTL_INT(_security_jail, OID_AUTO, allow_raw_sockets, CTLFLAG_RW,
87    &jail_allow_raw_sockets, 0,
88    "Prison root can create raw sockets");
89
90int	jail_chflags_allowed = 0;
91SYSCTL_INT(_security_jail, OID_AUTO, chflags_allowed, CTLFLAG_RW,
92    &jail_chflags_allowed, 0,
93    "Processes in jail can alter system file flags");
94
95int	jail_mount_allowed = 0;
96SYSCTL_INT(_security_jail, OID_AUTO, mount_allowed, CTLFLAG_RW,
97    &jail_mount_allowed, 0,
98    "Processes in jail can mount/unmount jail-friendly file systems");
99
100/* allprison, lastprid, and prisoncount are protected by allprison_lock. */
101struct	prisonlist allprison;
102struct	sx allprison_lock;
103int	lastprid = 0;
104int	prisoncount = 0;
105
106static void		 init_prison(void *);
107static void		 prison_complete(void *context, int pending);
108static int		 sysctl_jail_list(SYSCTL_HANDLER_ARGS);
109
110static void
111init_prison(void *data __unused)
112{
113
114	sx_init(&allprison_lock, "allprison");
115	LIST_INIT(&allprison);
116}
117
118SYSINIT(prison, SI_SUB_INTRINSIC, SI_ORDER_ANY, init_prison, NULL);
119
120/*
121 * struct jail_args {
122 *	struct jail *jail;
123 * };
124 */
125int
126jail(struct thread *td, struct jail_args *uap)
127{
128	struct nameidata nd;
129	struct prison *pr, *tpr;
130	struct jail j;
131	struct jail_attach_args jaa;
132	int vfslocked, error, tryprid;
133
134	error = copyin(uap->jail, &j, sizeof(j));
135	if (error)
136		return (error);
137	if (j.version != 0)
138		return (EINVAL);
139
140	pr = malloc(sizeof(*pr), M_PRISON, M_WAITOK | M_ZERO);
141	mtx_init(&pr->pr_mtx, "jail mutex", NULL, MTX_DEF);
142	pr->pr_ref = 1;
143	error = copyinstr(j.path, &pr->pr_path, sizeof(pr->pr_path), 0);
144	if (error)
145		goto e_killmtx;
146	NDINIT(&nd, LOOKUP, MPSAFE | FOLLOW | LOCKLEAF, UIO_SYSSPACE,
147	    pr->pr_path, td);
148	error = namei(&nd);
149	if (error)
150		goto e_killmtx;
151	vfslocked = NDHASGIANT(&nd);
152	pr->pr_root = nd.ni_vp;
153	VOP_UNLOCK(nd.ni_vp, 0);
154	NDFREE(&nd, NDF_ONLY_PNBUF);
155	VFS_UNLOCK_GIANT(vfslocked);
156	error = copyinstr(j.hostname, &pr->pr_host, sizeof(pr->pr_host), 0);
157	if (error)
158		goto e_dropvnref;
159	pr->pr_ip = j.ip_number;
160	pr->pr_linux = NULL;
161	pr->pr_securelevel = securelevel;
162	bzero(&pr->pr_osd, sizeof(pr->pr_osd));
163
164	/* Determine next pr_id and add prison to allprison list. */
165	sx_xlock(&allprison_lock);
166	tryprid = lastprid + 1;
167	if (tryprid == JAIL_MAX)
168		tryprid = 1;
169next:
170	LIST_FOREACH(tpr, &allprison, pr_list) {
171		if (tpr->pr_id == tryprid) {
172			tryprid++;
173			if (tryprid == JAIL_MAX) {
174				sx_xunlock(&allprison_lock);
175				error = EAGAIN;
176				goto e_dropvnref;
177			}
178			goto next;
179		}
180	}
181	pr->pr_id = jaa.jid = lastprid = tryprid;
182	LIST_INSERT_HEAD(&allprison, pr, pr_list);
183	prisoncount++;
184	sx_xunlock(&allprison_lock);
185
186	error = jail_attach(td, &jaa);
187	if (error)
188		goto e_dropprref;
189	mtx_lock(&pr->pr_mtx);
190	pr->pr_ref--;
191	mtx_unlock(&pr->pr_mtx);
192	td->td_retval[0] = jaa.jid;
193	return (0);
194e_dropprref:
195	sx_xlock(&allprison_lock);
196	LIST_REMOVE(pr, pr_list);
197	prisoncount--;
198	sx_xunlock(&allprison_lock);
199e_dropvnref:
200	vfslocked = VFS_LOCK_GIANT(pr->pr_root->v_mount);
201	vrele(pr->pr_root);
202	VFS_UNLOCK_GIANT(vfslocked);
203e_killmtx:
204	mtx_destroy(&pr->pr_mtx);
205	free(pr, M_PRISON);
206	return (error);
207}
208
209/*
210 * struct jail_attach_args {
211 *	int jid;
212 * };
213 */
214int
215jail_attach(struct thread *td, struct jail_attach_args *uap)
216{
217	struct proc *p;
218	struct ucred *newcred, *oldcred;
219	struct prison *pr;
220	int vfslocked, error;
221
222	/*
223	 * XXX: Note that there is a slight race here if two threads
224	 * in the same privileged process attempt to attach to two
225	 * different jails at the same time.  It is important for
226	 * user processes not to do this, or they might end up with
227	 * a process root from one prison, but attached to the jail
228	 * of another.
229	 */
230	error = priv_check(td, PRIV_JAIL_ATTACH);
231	if (error)
232		return (error);
233
234	p = td->td_proc;
235	sx_slock(&allprison_lock);
236	pr = prison_find(uap->jid);
237	if (pr == NULL) {
238		sx_sunlock(&allprison_lock);
239		return (EINVAL);
240	}
241	pr->pr_ref++;
242	mtx_unlock(&pr->pr_mtx);
243	sx_sunlock(&allprison_lock);
244
245	vfslocked = VFS_LOCK_GIANT(pr->pr_root->v_mount);
246	vn_lock(pr->pr_root, LK_EXCLUSIVE | LK_RETRY);
247	if ((error = change_dir(pr->pr_root, td)) != 0)
248		goto e_unlock;
249#ifdef MAC
250	if ((error = mac_vnode_check_chroot(td->td_ucred, pr->pr_root)))
251		goto e_unlock;
252#endif
253	VOP_UNLOCK(pr->pr_root, 0);
254	change_root(pr->pr_root, td);
255	VFS_UNLOCK_GIANT(vfslocked);
256
257	newcred = crget();
258	PROC_LOCK(p);
259	oldcred = p->p_ucred;
260	setsugid(p);
261	crcopy(newcred, oldcred);
262	newcred->cr_prison = pr;
263	p->p_ucred = newcred;
264	PROC_UNLOCK(p);
265	crfree(oldcred);
266	return (0);
267e_unlock:
268	VOP_UNLOCK(pr->pr_root, 0);
269	VFS_UNLOCK_GIANT(vfslocked);
270	mtx_lock(&pr->pr_mtx);
271	pr->pr_ref--;
272	mtx_unlock(&pr->pr_mtx);
273	return (error);
274}
275
276/*
277 * Returns a locked prison instance, or NULL on failure.
278 */
279struct prison *
280prison_find(int prid)
281{
282	struct prison *pr;
283
284	sx_assert(&allprison_lock, SX_LOCKED);
285	LIST_FOREACH(pr, &allprison, pr_list) {
286		if (pr->pr_id == prid) {
287			mtx_lock(&pr->pr_mtx);
288			if (pr->pr_ref == 0) {
289				mtx_unlock(&pr->pr_mtx);
290				break;
291			}
292			return (pr);
293		}
294	}
295	return (NULL);
296}
297
298void
299prison_free_locked(struct prison *pr)
300{
301
302	mtx_assert(&pr->pr_mtx, MA_OWNED);
303	pr->pr_ref--;
304	if (pr->pr_ref == 0) {
305		mtx_unlock(&pr->pr_mtx);
306		TASK_INIT(&pr->pr_task, 0, prison_complete, pr);
307		taskqueue_enqueue(taskqueue_thread, &pr->pr_task);
308		return;
309	}
310	mtx_unlock(&pr->pr_mtx);
311}
312
313void
314prison_free(struct prison *pr)
315{
316
317	mtx_lock(&pr->pr_mtx);
318	prison_free_locked(pr);
319}
320
321static void
322prison_complete(void *context, int pending)
323{
324	struct prison *pr;
325	int vfslocked;
326
327	pr = (struct prison *)context;
328
329	sx_xlock(&allprison_lock);
330	LIST_REMOVE(pr, pr_list);
331	prisoncount--;
332	sx_xunlock(&allprison_lock);
333
334	/* Free all OSD associated to this jail. */
335	osd_jail_exit(pr);
336
337	vfslocked = VFS_LOCK_GIANT(pr->pr_root->v_mount);
338	vrele(pr->pr_root);
339	VFS_UNLOCK_GIANT(vfslocked);
340
341	mtx_destroy(&pr->pr_mtx);
342	if (pr->pr_linux != NULL)
343		free(pr->pr_linux, M_PRISON);
344	free(pr, M_PRISON);
345}
346
347void
348prison_hold_locked(struct prison *pr)
349{
350
351	mtx_assert(&pr->pr_mtx, MA_OWNED);
352	KASSERT(pr->pr_ref > 0,
353	    ("Trying to hold dead prison (id=%d).", pr->pr_id));
354	pr->pr_ref++;
355}
356
357void
358prison_hold(struct prison *pr)
359{
360
361	mtx_lock(&pr->pr_mtx);
362	prison_hold_locked(pr);
363	mtx_unlock(&pr->pr_mtx);
364}
365
366u_int32_t
367prison_getip(struct ucred *cred)
368{
369
370	return (cred->cr_prison->pr_ip);
371}
372
373int
374prison_ip(struct ucred *cred, int flag, u_int32_t *ip)
375{
376	u_int32_t tmp;
377
378	if (!jailed(cred))
379		return (0);
380	if (flag)
381		tmp = *ip;
382	else
383		tmp = ntohl(*ip);
384	if (tmp == INADDR_ANY) {
385		if (flag)
386			*ip = cred->cr_prison->pr_ip;
387		else
388			*ip = htonl(cred->cr_prison->pr_ip);
389		return (0);
390	}
391	if (tmp == INADDR_LOOPBACK) {
392		if (flag)
393			*ip = cred->cr_prison->pr_ip;
394		else
395			*ip = htonl(cred->cr_prison->pr_ip);
396		return (0);
397	}
398	if (cred->cr_prison->pr_ip != tmp)
399		return (1);
400	return (0);
401}
402
403void
404prison_remote_ip(struct ucred *cred, int flag, u_int32_t *ip)
405{
406	u_int32_t tmp;
407
408	if (!jailed(cred))
409		return;
410	if (flag)
411		tmp = *ip;
412	else
413		tmp = ntohl(*ip);
414	if (tmp == INADDR_LOOPBACK) {
415		if (flag)
416			*ip = cred->cr_prison->pr_ip;
417		else
418			*ip = htonl(cred->cr_prison->pr_ip);
419		return;
420	}
421	return;
422}
423
424int
425prison_if(struct ucred *cred, struct sockaddr *sa)
426{
427	struct sockaddr_in *sai;
428	int ok;
429
430	sai = (struct sockaddr_in *)sa;
431	if ((sai->sin_family != AF_INET) && jail_socket_unixiproute_only)
432		ok = 1;
433	else if (sai->sin_family != AF_INET)
434		ok = 0;
435	else if (cred->cr_prison->pr_ip != ntohl(sai->sin_addr.s_addr))
436		ok = 1;
437	else
438		ok = 0;
439	return (ok);
440}
441
442/*
443 * Return 0 if jails permit p1 to frob p2, otherwise ESRCH.
444 */
445int
446prison_check(struct ucred *cred1, struct ucred *cred2)
447{
448
449	if (jailed(cred1)) {
450		if (!jailed(cred2))
451			return (ESRCH);
452		if (cred2->cr_prison != cred1->cr_prison)
453			return (ESRCH);
454	}
455
456	return (0);
457}
458
459/*
460 * Return 1 if the passed credential is in a jail, otherwise 0.
461 */
462int
463jailed(struct ucred *cred)
464{
465
466	return (cred->cr_prison != NULL);
467}
468
469/*
470 * Return the correct hostname for the passed credential.
471 */
472void
473getcredhostname(struct ucred *cred, char *buf, size_t size)
474{
475	INIT_VPROCG(cred->cr_vimage->v_procg);
476
477	if (jailed(cred)) {
478		mtx_lock(&cred->cr_prison->pr_mtx);
479		strlcpy(buf, cred->cr_prison->pr_host, size);
480		mtx_unlock(&cred->cr_prison->pr_mtx);
481	} else {
482		mtx_lock(&hostname_mtx);
483		strlcpy(buf, V_hostname, size);
484		mtx_unlock(&hostname_mtx);
485	}
486}
487
488/*
489 * Determine whether the subject represented by cred can "see"
490 * status of a mount point.
491 * Returns: 0 for permitted, ENOENT otherwise.
492 * XXX: This function should be called cr_canseemount() and should be
493 *      placed in kern_prot.c.
494 */
495int
496prison_canseemount(struct ucred *cred, struct mount *mp)
497{
498	struct prison *pr;
499	struct statfs *sp;
500	size_t len;
501
502	if (!jailed(cred) || jail_enforce_statfs == 0)
503		return (0);
504	pr = cred->cr_prison;
505	if (pr->pr_root->v_mount == mp)
506		return (0);
507	if (jail_enforce_statfs == 2)
508		return (ENOENT);
509	/*
510	 * If jail's chroot directory is set to "/" we should be able to see
511	 * all mount-points from inside a jail.
512	 * This is ugly check, but this is the only situation when jail's
513	 * directory ends with '/'.
514	 */
515	if (strcmp(pr->pr_path, "/") == 0)
516		return (0);
517	len = strlen(pr->pr_path);
518	sp = &mp->mnt_stat;
519	if (strncmp(pr->pr_path, sp->f_mntonname, len) != 0)
520		return (ENOENT);
521	/*
522	 * Be sure that we don't have situation where jail's root directory
523	 * is "/some/path" and mount point is "/some/pathpath".
524	 */
525	if (sp->f_mntonname[len] != '\0' && sp->f_mntonname[len] != '/')
526		return (ENOENT);
527	return (0);
528}
529
530void
531prison_enforce_statfs(struct ucred *cred, struct mount *mp, struct statfs *sp)
532{
533	char jpath[MAXPATHLEN];
534	struct prison *pr;
535	size_t len;
536
537	if (!jailed(cred) || jail_enforce_statfs == 0)
538		return;
539	pr = cred->cr_prison;
540	if (prison_canseemount(cred, mp) != 0) {
541		bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
542		strlcpy(sp->f_mntonname, "[restricted]",
543		    sizeof(sp->f_mntonname));
544		return;
545	}
546	if (pr->pr_root->v_mount == mp) {
547		/*
548		 * Clear current buffer data, so we are sure nothing from
549		 * the valid path left there.
550		 */
551		bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
552		*sp->f_mntonname = '/';
553		return;
554	}
555	/*
556	 * If jail's chroot directory is set to "/" we should be able to see
557	 * all mount-points from inside a jail.
558	 */
559	if (strcmp(pr->pr_path, "/") == 0)
560		return;
561	len = strlen(pr->pr_path);
562	strlcpy(jpath, sp->f_mntonname + len, sizeof(jpath));
563	/*
564	 * Clear current buffer data, so we are sure nothing from
565	 * the valid path left there.
566	 */
567	bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
568	if (*jpath == '\0') {
569		/* Should never happen. */
570		*sp->f_mntonname = '/';
571	} else {
572		strlcpy(sp->f_mntonname, jpath, sizeof(sp->f_mntonname));
573	}
574}
575
576/*
577 * Check with permission for a specific privilege is granted within jail.  We
578 * have a specific list of accepted privileges; the rest are denied.
579 */
580int
581prison_priv_check(struct ucred *cred, int priv)
582{
583
584	if (!jailed(cred))
585		return (0);
586
587	switch (priv) {
588
589		/*
590		 * Allow ktrace privileges for root in jail.
591		 */
592	case PRIV_KTRACE:
593
594#if 0
595		/*
596		 * Allow jailed processes to configure audit identity and
597		 * submit audit records (login, etc).  In the future we may
598		 * want to further refine the relationship between audit and
599		 * jail.
600		 */
601	case PRIV_AUDIT_GETAUDIT:
602	case PRIV_AUDIT_SETAUDIT:
603	case PRIV_AUDIT_SUBMIT:
604#endif
605
606		/*
607		 * Allow jailed processes to manipulate process UNIX
608		 * credentials in any way they see fit.
609		 */
610	case PRIV_CRED_SETUID:
611	case PRIV_CRED_SETEUID:
612	case PRIV_CRED_SETGID:
613	case PRIV_CRED_SETEGID:
614	case PRIV_CRED_SETGROUPS:
615	case PRIV_CRED_SETREUID:
616	case PRIV_CRED_SETREGID:
617	case PRIV_CRED_SETRESUID:
618	case PRIV_CRED_SETRESGID:
619
620		/*
621		 * Jail implements visibility constraints already, so allow
622		 * jailed root to override uid/gid-based constraints.
623		 */
624	case PRIV_SEEOTHERGIDS:
625	case PRIV_SEEOTHERUIDS:
626
627		/*
628		 * Jail implements inter-process debugging limits already, so
629		 * allow jailed root various debugging privileges.
630		 */
631	case PRIV_DEBUG_DIFFCRED:
632	case PRIV_DEBUG_SUGID:
633	case PRIV_DEBUG_UNPRIV:
634
635		/*
636		 * Allow jail to set various resource limits and login
637		 * properties, and for now, exceed process resource limits.
638		 */
639	case PRIV_PROC_LIMIT:
640	case PRIV_PROC_SETLOGIN:
641	case PRIV_PROC_SETRLIMIT:
642
643		/*
644		 * System V and POSIX IPC privileges are granted in jail.
645		 */
646	case PRIV_IPC_READ:
647	case PRIV_IPC_WRITE:
648	case PRIV_IPC_ADMIN:
649	case PRIV_IPC_MSGSIZE:
650	case PRIV_MQ_ADMIN:
651
652		/*
653		 * Jail implements its own inter-process limits, so allow
654		 * root processes in jail to change scheduling on other
655		 * processes in the same jail.  Likewise for signalling.
656		 */
657	case PRIV_SCHED_DIFFCRED:
658	case PRIV_SIGNAL_DIFFCRED:
659	case PRIV_SIGNAL_SUGID:
660
661		/*
662		 * Allow jailed processes to write to sysctls marked as jail
663		 * writable.
664		 */
665	case PRIV_SYSCTL_WRITEJAIL:
666
667		/*
668		 * Allow root in jail to manage a variety of quota
669		 * properties.  These should likely be conditional on a
670		 * configuration option.
671		 */
672	case PRIV_VFS_GETQUOTA:
673	case PRIV_VFS_SETQUOTA:
674
675		/*
676		 * Since Jail relies on chroot() to implement file system
677		 * protections, grant many VFS privileges to root in jail.
678		 * Be careful to exclude mount-related and NFS-related
679		 * privileges.
680		 */
681	case PRIV_VFS_READ:
682	case PRIV_VFS_WRITE:
683	case PRIV_VFS_ADMIN:
684	case PRIV_VFS_EXEC:
685	case PRIV_VFS_LOOKUP:
686	case PRIV_VFS_BLOCKRESERVE:	/* XXXRW: Slightly surprising. */
687	case PRIV_VFS_CHFLAGS_DEV:
688	case PRIV_VFS_CHOWN:
689	case PRIV_VFS_CHROOT:
690	case PRIV_VFS_RETAINSUGID:
691	case PRIV_VFS_FCHROOT:
692	case PRIV_VFS_LINK:
693	case PRIV_VFS_SETGID:
694	case PRIV_VFS_STAT:
695	case PRIV_VFS_STICKYFILE:
696		return (0);
697
698		/*
699		 * Depending on the global setting, allow privilege of
700		 * setting system flags.
701		 */
702	case PRIV_VFS_SYSFLAGS:
703		if (jail_chflags_allowed)
704			return (0);
705		else
706			return (EPERM);
707
708		/*
709		 * Depending on the global setting, allow privilege of
710		 * mounting/unmounting file systems.
711		 */
712	case PRIV_VFS_MOUNT:
713	case PRIV_VFS_UNMOUNT:
714	case PRIV_VFS_MOUNT_NONUSER:
715	case PRIV_VFS_MOUNT_OWNER:
716		if (jail_mount_allowed)
717			return (0);
718		else
719			return (EPERM);
720
721		/*
722		 * Allow jailed root to bind reserved ports and reuse in-use
723		 * ports.
724		 */
725	case PRIV_NETINET_RESERVEDPORT:
726	case PRIV_NETINET_REUSEPORT:
727		return (0);
728
729		/*
730		 * Allow jailed root to set certian IPv4/6 (option) headers.
731		 */
732	case PRIV_NETINET_SETHDROPTS:
733		return (0);
734
735		/*
736		 * Conditionally allow creating raw sockets in jail.
737		 */
738	case PRIV_NETINET_RAW:
739		if (jail_allow_raw_sockets)
740			return (0);
741		else
742			return (EPERM);
743
744		/*
745		 * Since jail implements its own visibility limits on netstat
746		 * sysctls, allow getcred.  This allows identd to work in
747		 * jail.
748		 */
749	case PRIV_NETINET_GETCRED:
750		return (0);
751
752	default:
753		/*
754		 * In all remaining cases, deny the privilege request.  This
755		 * includes almost all network privileges, many system
756		 * configuration privileges.
757		 */
758		return (EPERM);
759	}
760}
761
762static int
763sysctl_jail_list(SYSCTL_HANDLER_ARGS)
764{
765	struct xprison *xp, *sxp;
766	struct prison *pr;
767	int count, error;
768
769	if (jailed(req->td->td_ucred))
770		return (0);
771
772	sx_slock(&allprison_lock);
773	if ((count = prisoncount) == 0) {
774		sx_sunlock(&allprison_lock);
775		return (0);
776	}
777
778	sxp = xp = malloc(sizeof(*xp) * count, M_TEMP, M_WAITOK | M_ZERO);
779
780	LIST_FOREACH(pr, &allprison, pr_list) {
781		xp->pr_version = XPRISON_VERSION;
782		xp->pr_id = pr->pr_id;
783		xp->pr_ip = pr->pr_ip;
784		strlcpy(xp->pr_path, pr->pr_path, sizeof(xp->pr_path));
785		mtx_lock(&pr->pr_mtx);
786		strlcpy(xp->pr_host, pr->pr_host, sizeof(xp->pr_host));
787		mtx_unlock(&pr->pr_mtx);
788		xp++;
789	}
790	sx_sunlock(&allprison_lock);
791
792	error = SYSCTL_OUT(req, sxp, sizeof(*sxp) * count);
793	free(sxp, M_TEMP);
794	return (error);
795}
796
797SYSCTL_OID(_security_jail, OID_AUTO, list, CTLTYPE_STRUCT | CTLFLAG_RD,
798    NULL, 0, sysctl_jail_list, "S", "List of active jails");
799
800static int
801sysctl_jail_jailed(SYSCTL_HANDLER_ARGS)
802{
803	int error, injail;
804
805	injail = jailed(req->td->td_ucred);
806	error = SYSCTL_OUT(req, &injail, sizeof(injail));
807
808	return (error);
809}
810SYSCTL_PROC(_security_jail, OID_AUTO, jailed, CTLTYPE_INT | CTLFLAG_RD,
811    NULL, 0, sysctl_jail_jailed, "I", "Process in jail?");
812