1 /*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 1999 Poul-Henning Kamp.
5 * Copyright (c) 2009 James Gritton.
6 * All rights reserved.
7 *
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 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * $FreeBSD$
30 */
31
32 #ifndef _SYS_JAIL_H_
33 #define _SYS_JAIL_H_
34
35 #ifdef _KERNEL
36 struct jail_v0 {
37 u_int32_t version;
38 char *path;
39 char *hostname;
40 u_int32_t ip_number;
41 };
42 #endif
43
44 struct jail {
45 uint32_t version;
46 char *path;
47 char *hostname;
48 char *jailname;
49 uint32_t ip4s;
50 uint32_t ip6s;
51 struct in_addr *ip4;
52 struct in6_addr *ip6;
53 };
54 #define JAIL_API_VERSION 2
55
56 /*
57 * For all xprison structs, always keep the pr_version an int and
58 * the first variable so userspace can easily distinguish them.
59 */
60 #ifndef _KERNEL
61 struct xprison_v1 {
62 int pr_version;
63 int pr_id;
64 char pr_path[MAXPATHLEN];
65 char pr_host[MAXHOSTNAMELEN];
66 u_int32_t pr_ip;
67 };
68 #endif
69
70 struct xprison {
71 int pr_version;
72 int pr_id;
73 int pr_state;
74 cpusetid_t pr_cpusetid;
75 char pr_path[MAXPATHLEN];
76 char pr_host[MAXHOSTNAMELEN];
77 char pr_name[MAXHOSTNAMELEN];
78 uint32_t pr_ip4s;
79 uint32_t pr_ip6s;
80 #if 0
81 /*
82 * sizeof(xprison) will be malloced + size needed for all
83 * IPv4 and IPv6 addesses. Offsets are based numbers of addresses.
84 */
85 struct in_addr pr_ip4[];
86 struct in6_addr pr_ip6[];
87 #endif
88 };
89 #define XPRISON_VERSION 3
90
91 #define PRISON_STATE_INVALID 0
92 #define PRISON_STATE_ALIVE 1
93 #define PRISON_STATE_DYING 2
94
95 /*
96 * Flags for jail_set and jail_get.
97 */
98 #define JAIL_CREATE 0x01 /* Create jail if it doesn't exist */
99 #define JAIL_UPDATE 0x02 /* Update parameters of existing jail */
100 #define JAIL_ATTACH 0x04 /* Attach to jail upon creation */
101 #define JAIL_DYING 0x08 /* Allow getting a dying jail */
102 #define JAIL_SET_MASK 0x0f
103 #define JAIL_GET_MASK 0x08
104
105 #define JAIL_SYS_DISABLE 0
106 #define JAIL_SYS_NEW 1
107 #define JAIL_SYS_INHERIT 2
108
109 #ifndef _KERNEL
110
111 struct iovec;
112
113 __BEGIN_DECLS
114 int jail(struct jail *);
115 int jail_set(struct iovec *, unsigned int, int);
116 int jail_get(struct iovec *, unsigned int, int);
117 int jail_attach(int);
118 int jail_remove(int);
119 __END_DECLS
120
121 #else /* _KERNEL */
122
123 #include <sys/queue.h>
124 #include <sys/sysctl.h>
125 #include <sys/lock.h>
126 #include <sys/mutex.h>
127 #include <sys/_task.h>
128
129 #define JAIL_MAX 999999
130
131 #ifdef MALLOC_DECLARE
132 MALLOC_DECLARE(M_PRISON);
133 #endif
134 #endif /* _KERNEL */
135
136 #if defined(_KERNEL) || defined(_WANT_PRISON)
137
138 #include <sys/osd.h>
139
140 #define HOSTUUIDLEN 64
141 #define OSRELEASELEN 32
142
143 struct racct;
144 struct prison_racct;
145
146 /*
147 * This structure describes a prison. It is pointed to by all struct
148 * ucreds's of the inmates. pr_ref keeps track of them and is used to
149 * delete the struture when the last inmate is dead.
150 *
151 * Lock key:
152 * (a) allprison_lock
153 * (c) set only during creation before the structure is shared, no mutex
154 * required to read
155 * (m) locked by pr_mtx
156 * (p) locked by pr_mtx, and also at least shared allprison_lock required
157 * to update
158 * (r) atomic via refcount(9), pr_mtx required to decrement to zero
159 */
160 struct prison {
161 TAILQ_ENTRY(prison) pr_list; /* (a) all prisons */
162 int pr_id; /* (c) prison id */
163 volatile u_int pr_ref; /* (r) refcount */
164 volatile u_int pr_uref; /* (r) user (alive) refcount */
165 unsigned pr_flags; /* (p) PR_* flags */
166 LIST_HEAD(, prison) pr_children; /* (a) list of child jails */
167 LIST_ENTRY(prison) pr_sibling; /* (a) next in parent's list */
168 struct prison *pr_parent; /* (c) containing jail */
169 struct mtx pr_mtx;
170 struct task pr_task; /* (c) destroy task */
171 struct osd pr_osd; /* (p) additional data */
172 struct cpuset *pr_cpuset; /* (p) cpuset */
173 struct vnet *pr_vnet; /* (c) network stack */
174 struct vnode *pr_root; /* (c) vnode to rdir */
175 int pr_ip4s; /* (p) number of v4 IPs */
176 int pr_ip6s; /* (p) number of v6 IPs */
177 struct in_addr *pr_ip4; /* (p) v4 IPs of jail */
178 struct in6_addr *pr_ip6; /* (p) v6 IPs of jail */
179 struct prison_racct *pr_prison_racct; /* (c) racct jail proxy */
180 void *pr_sparep[3];
181 int pr_childcount; /* (a) number of child jails */
182 int pr_childmax; /* (p) maximum child jails */
183 unsigned pr_allow; /* (p) PR_ALLOW_* flags */
184 int pr_securelevel; /* (p) securelevel */
185 int pr_enforce_statfs; /* (p) statfs permission */
186 int pr_devfs_rsnum; /* (p) devfs ruleset */
187 int pr_spare[3];
188 int pr_osreldate; /* (c) kern.osreldate value */
189 unsigned long pr_hostid; /* (p) jail hostid */
190 char pr_name[MAXHOSTNAMELEN]; /* (p) admin jail name */
191 char pr_path[MAXPATHLEN]; /* (c) chroot path */
192 char pr_hostname[MAXHOSTNAMELEN]; /* (p) jail hostname */
193 char pr_domainname[MAXHOSTNAMELEN]; /* (p) jail domainname */
194 char pr_hostuuid[HOSTUUIDLEN]; /* (p) jail hostuuid */
195 char pr_osrelease[OSRELEASELEN]; /* (c) kern.osrelease value */
196 };
197
198 struct prison_racct {
199 LIST_ENTRY(prison_racct) prr_next;
200 char prr_name[MAXHOSTNAMELEN];
201 u_int prr_refcount;
202 struct racct *prr_racct;
203 };
204 #endif /* _KERNEL || _WANT_PRISON */
205
206 #ifdef _KERNEL
207 /* Flag bits set via options */
208 #define PR_PERSIST 0x00000001 /* Can exist without processes */
209 #define PR_HOST 0x00000002 /* Virtualize hostname et al */
210 #define PR_IP4_USER 0x00000004 /* Restrict IPv4 addresses */
211 #define PR_IP6_USER 0x00000008 /* Restrict IPv6 addresses */
212 #define PR_VNET 0x00000010 /* Virtual network stack */
213 #define PR_IP4_SADDRSEL 0x00000080 /* Do IPv4 src addr sel. or use the */
214 /* primary jail address. */
215 #define PR_IP6_SADDRSEL 0x00000100 /* Do IPv6 src addr sel. or use the */
216 /* primary jail address. */
217
218 /* Internal flag bits */
219 #define PR_REMOVE 0x01000000 /* In process of being removed */
220 #define PR_IP4 0x02000000 /* IPv4 restricted or disabled */
221 /* by this jail or an ancestor */
222 #define PR_IP6 0x04000000 /* IPv6 restricted or disabled */
223 /* by this jail or an ancestor */
224
225 /*
226 * Flags for pr_allow
227 * Bits not noted here may be used for dynamic allow.mount.xxxfs.
228 */
229 #define PR_ALLOW_SET_HOSTNAME 0x00000001
230 #define PR_ALLOW_SYSVIPC 0x00000002
231 #define PR_ALLOW_RAW_SOCKETS 0x00000004
232 #define PR_ALLOW_CHFLAGS 0x00000008
233 #define PR_ALLOW_MOUNT 0x00000010
234 #define PR_ALLOW_QUOTAS 0x00000020
235 #define PR_ALLOW_SOCKET_AF 0x00000040
236 #define PR_ALLOW_MLOCK 0x00000080
237 #define PR_ALLOW_READ_MSGBUF 0x00000100
238 #define PR_ALLOW_UNPRIV_DEBUG 0x00000200
239 #define PR_ALLOW_SUSER 0x00000400
240 #define PR_ALLOW_RESERVED_PORTS 0x00008000
241 #define PR_ALLOW_KMEM_ACCESS 0x00010000 /* reserved, not used yet */
242 #define PR_ALLOW_ALL_STATIC 0x000187ff
243
244 /*
245 * PR_ALLOW_DIFFERENCES determines which flags are able to be
246 * different between the parent and child jail upon creation.
247 */
248 #define PR_ALLOW_DIFFERENCES (PR_ALLOW_UNPRIV_DEBUG)
249
250 /*
251 * OSD methods
252 */
253 #define PR_METHOD_CREATE 0
254 #define PR_METHOD_GET 1
255 #define PR_METHOD_SET 2
256 #define PR_METHOD_CHECK 3
257 #define PR_METHOD_ATTACH 4
258 #define PR_METHOD_REMOVE 5
259 #define PR_MAXMETHOD 6
260
261 /*
262 * Lock/unlock a prison.
263 * XXX These exist not so much for general convenience, but to be useable in
264 * the FOREACH_PRISON_DESCENDANT_LOCKED macro which can't handle them in
265 * non-function form as currently defined.
266 */
267 static __inline void
prison_lock(struct prison * pr)268 prison_lock(struct prison *pr)
269 {
270
271 mtx_lock(&pr->pr_mtx);
272 }
273
274 static __inline void
prison_unlock(struct prison * pr)275 prison_unlock(struct prison *pr)
276 {
277
278 mtx_unlock(&pr->pr_mtx);
279 }
280
281 /* Traverse a prison's immediate children. */
282 #define FOREACH_PRISON_CHILD(ppr, cpr) \
283 LIST_FOREACH(cpr, &(ppr)->pr_children, pr_sibling)
284
285 /*
286 * Preorder traversal of all of a prison's descendants.
287 * This ugly loop allows the macro to be followed by a single block
288 * as expected in a looping primitive.
289 */
290 #define FOREACH_PRISON_DESCENDANT(ppr, cpr, descend) \
291 for ((cpr) = (ppr), (descend) = 1; \
292 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \
293 ? LIST_FIRST(&(cpr)->pr_children) \
294 : ((cpr) == (ppr) \
295 ? NULL \
296 : (((descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \
297 ? LIST_NEXT(cpr, pr_sibling) \
298 : (cpr)->pr_parent))));) \
299 if (!(descend)) \
300 ; \
301 else
302
303 /*
304 * As above, but lock descendants on the way down and unlock on the way up.
305 */
306 #define FOREACH_PRISON_DESCENDANT_LOCKED(ppr, cpr, descend) \
307 for ((cpr) = (ppr), (descend) = 1; \
308 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \
309 ? LIST_FIRST(&(cpr)->pr_children) \
310 : ((cpr) == (ppr) \
311 ? NULL \
312 : ((prison_unlock(cpr), \
313 (descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \
314 ? LIST_NEXT(cpr, pr_sibling) \
315 : (cpr)->pr_parent))));) \
316 if ((descend) ? (prison_lock(cpr), 0) : 1) \
317 ; \
318 else
319
320 /*
321 * As above, but also keep track of the level descended to.
322 */
323 #define FOREACH_PRISON_DESCENDANT_LOCKED_LEVEL(ppr, cpr, descend, level)\
324 for ((cpr) = (ppr), (descend) = 1, (level) = 0; \
325 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \
326 ? (level++, LIST_FIRST(&(cpr)->pr_children)) \
327 : ((cpr) == (ppr) \
328 ? NULL \
329 : ((prison_unlock(cpr), \
330 (descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \
331 ? LIST_NEXT(cpr, pr_sibling) \
332 : (level--, (cpr)->pr_parent)))));) \
333 if ((descend) ? (prison_lock(cpr), 0) : 1) \
334 ; \
335 else
336
337 /*
338 * Attributes of the physical system, and the root of the jail tree.
339 */
340 extern struct prison prison0;
341
342 TAILQ_HEAD(prisonlist, prison);
343 extern struct prisonlist allprison;
344 extern struct sx allprison_lock;
345
346 /*
347 * Sysctls to describe jail parameters.
348 */
349 SYSCTL_DECL(_security_jail_param);
350
351 #define SYSCTL_JAIL_PARAM(module, param, type, fmt, descr) \
352 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \
353 (type) | CTLFLAG_MPSAFE, NULL, 0, sysctl_jail_param, fmt, descr)
354 #define SYSCTL_JAIL_PARAM_STRING(module, param, access, len, descr) \
355 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \
356 CTLTYPE_STRING | CTLFLAG_MPSAFE | (access), NULL, len, \
357 sysctl_jail_param, "A", descr)
358 #define SYSCTL_JAIL_PARAM_STRUCT(module, param, access, len, fmt, descr)\
359 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \
360 CTLTYPE_STRUCT | CTLFLAG_MPSAFE | (access), NULL, len, \
361 sysctl_jail_param, fmt, descr)
362 #define SYSCTL_JAIL_PARAM_NODE(module, descr) \
363 SYSCTL_NODE(_security_jail_param, OID_AUTO, module, CTLFLAG_MPSAFE, \
364 0, descr)
365 #define SYSCTL_JAIL_PARAM_SUBNODE(parent, module, descr) \
366 SYSCTL_NODE(_security_jail_param_##parent, OID_AUTO, module, \
367 CTLFLAG_MPSAFE, 0, descr)
368 #define SYSCTL_JAIL_PARAM_SYS_NODE(module, access, descr) \
369 SYSCTL_JAIL_PARAM_NODE(module, descr); \
370 SYSCTL_JAIL_PARAM(_##module, , CTLTYPE_INT | (access), "E,jailsys", \
371 descr)
372
373 /*
374 * Kernel support functions for jail().
375 */
376 struct ucred;
377 struct mount;
378 struct sockaddr;
379 struct statfs;
380 struct vfsconf;
381
382 /*
383 * Return 1 if the passed credential is in a jail, otherwise 0.
384 */
385 #define jailed(cred) (cred->cr_prison != &prison0)
386
387 int jailed_without_vnet(struct ucred *);
388 void getcredhostname(struct ucred *, char *, size_t);
389 void getcreddomainname(struct ucred *, char *, size_t);
390 void getcredhostuuid(struct ucred *, char *, size_t);
391 void getcredhostid(struct ucred *, unsigned long *);
392 void getjailname(struct ucred *cred, char *name, size_t len);
393 void prison0_init(void);
394 int prison_allow(struct ucred *, unsigned);
395 int prison_check(struct ucred *cred1, struct ucred *cred2);
396 int prison_owns_vnet(struct ucred *);
397 int prison_canseemount(struct ucred *cred, struct mount *mp);
398 void prison_enforce_statfs(struct ucred *cred, struct mount *mp,
399 struct statfs *sp);
400 struct prison *prison_find(int prid);
401 struct prison *prison_find_child(struct prison *, int);
402 struct prison *prison_find_name(struct prison *, const char *);
403 int prison_flag(struct ucred *, unsigned);
404 void prison_free(struct prison *pr);
405 void prison_free_locked(struct prison *pr);
406 void prison_hold(struct prison *pr);
407 void prison_hold_locked(struct prison *pr);
408 void prison_proc_hold(struct prison *);
409 void prison_proc_free(struct prison *);
410 void prison_set_allow(struct ucred *cred, unsigned flag, int enable);
411 int prison_ischild(struct prison *, struct prison *);
412 bool prison_isalive(struct prison *);
413 bool prison_isvalid(struct prison *);
414 int prison_equal_ip4(struct prison *, struct prison *);
415 int prison_get_ip4(struct ucred *cred, struct in_addr *ia);
416 int prison_local_ip4(struct ucred *cred, struct in_addr *ia);
417 int prison_remote_ip4(struct ucred *cred, struct in_addr *ia);
418 int prison_check_ip4(const struct ucred *, const struct in_addr *);
419 int prison_check_ip4_locked(const struct prison *, const struct in_addr *);
420 int prison_saddrsel_ip4(struct ucred *, struct in_addr *);
421 int prison_restrict_ip4(struct prison *, struct in_addr *);
422 int prison_qcmp_v4(const void *, const void *);
423 #ifdef INET6
424 int prison_equal_ip6(struct prison *, struct prison *);
425 int prison_get_ip6(struct ucred *, struct in6_addr *);
426 int prison_local_ip6(struct ucred *, struct in6_addr *, int);
427 int prison_remote_ip6(struct ucred *, struct in6_addr *);
428 int prison_check_ip6(const struct ucred *, const struct in6_addr *);
429 int prison_check_ip6_locked(const struct prison *, const struct in6_addr *);
430 int prison_saddrsel_ip6(struct ucred *, struct in6_addr *);
431 int prison_restrict_ip6(struct prison *, struct in6_addr *);
432 int prison_qcmp_v6(const void *, const void *);
433 #endif
434 int prison_check_af(struct ucred *cred, int af);
435 int prison_if(struct ucred *cred, const struct sockaddr *sa);
436 char *prison_name(struct prison *, struct prison *);
437 int prison_priv_check(struct ucred *cred, int priv);
438 int sysctl_jail_param(SYSCTL_HANDLER_ARGS);
439 unsigned prison_add_allow(const char *prefix, const char *name,
440 const char *prefix_descr, const char *descr);
441 void prison_add_vfs(struct vfsconf *vfsp);
442 void prison_racct_foreach(void (*callback)(struct racct *racct,
443 void *arg2, void *arg3), void (*pre)(void), void (*post)(void),
444 void *arg2, void *arg3);
445 struct prison_racct *prison_racct_find(const char *name);
446 void prison_racct_hold(struct prison_racct *prr);
447 void prison_racct_free(struct prison_racct *prr);
448
449 #endif /* _KERNEL */
450 #endif /* !_SYS_JAIL_H_ */
451