1/*- 2 * SPDX-License-Identifier: BSD-2-Clause 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 30#ifndef _SYS_JAIL_H_ 31#define _SYS_JAIL_H_ 32 33#ifdef _KERNEL 34struct jail_v0 { 35 u_int32_t version; 36 char *path; 37 char *hostname; 38 u_int32_t ip_number; 39}; 40#endif 41 42struct jail { 43 uint32_t version; 44 char *path; 45 char *hostname; 46 char *jailname; 47 uint32_t ip4s; 48 uint32_t ip6s; 49 struct in_addr *ip4; 50 struct in6_addr *ip6; 51}; 52#define JAIL_API_VERSION 2 53 54/* 55 * For all xprison structs, always keep the pr_version an int and 56 * the first variable so userspace can easily distinguish them. 57 */ 58#ifndef _KERNEL 59struct xprison_v1 { 60 int pr_version; 61 int pr_id; 62 char pr_path[MAXPATHLEN]; 63 char pr_host[MAXHOSTNAMELEN]; 64 u_int32_t pr_ip; 65}; 66#endif 67 68struct xprison { 69 int pr_version; 70 int pr_id; 71 int pr_state; 72 cpusetid_t pr_cpusetid; 73 char pr_path[MAXPATHLEN]; 74 char pr_host[MAXHOSTNAMELEN]; 75 char pr_name[MAXHOSTNAMELEN]; 76 uint32_t pr_ip4s; 77 uint32_t pr_ip6s; 78#if 0 79 /* 80 * sizeof(xprison) will be malloced + size needed for all 81 * IPv4 and IPv6 addesses. Offsets are based numbers of addresses. 82 */ 83 struct in_addr pr_ip4[]; 84 struct in6_addr pr_ip6[]; 85#endif 86}; 87#define XPRISON_VERSION 3 88 89enum prison_state { 90 PRISON_STATE_INVALID = 0, /* New prison, not ready to be seen */ 91 PRISON_STATE_ALIVE, /* Current prison, visible to all */ 92 PRISON_STATE_DYING /* Removed but holding resources, */ 93}; /* optionally visible. */ 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 /* JAIL_DYING is deprecated/ignored here */ 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 111struct iovec; 112 113__BEGIN_DECLS 114int jail(struct jail *); 115int jail_set(struct iovec *, unsigned int, int); 116int jail_get(struct iovec *, unsigned int, int); 117int jail_attach(int); 118int 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 132MALLOC_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 DEFAULT_HOSTUUID "00000000-0000-0000-0000-000000000000" 142#define OSRELEASELEN 32 143 144struct racct; 145struct prison_racct; 146 147typedef enum { 148 PR_INET = 0, 149 PR_INET6 = 1, 150 PR_FAMILY_MAX = 2, 151} pr_family_t; 152 153/* 154 * This structure describes a prison. It is pointed to by all struct 155 * ucreds's of the inmates. pr_ref keeps track of them and is used to 156 * delete the structure when the last inmate is dead. 157 * 158 * Lock key: 159 * (a) allprison_lock 160 * (A) allproc_lock 161 * (c) set only during creation before the structure is shared, no mutex 162 * required to read 163 * (m) locked by pr_mtx 164 * (p) locked by pr_mtx, and also at least shared allprison_lock required 165 * to update 166 * (q) locked by both pr_mtx and allprison_lock 167 * (r) atomic via refcount(9), pr_mtx and allprison_lock required to 168 * decrement to zero 169 * (n) read access granted with the network epoch 170 */ 171struct prison { 172 TAILQ_ENTRY(prison) pr_list; /* (a) all prisons */ 173 int pr_id; /* (c) prison id */ 174 volatile u_int pr_ref; /* (r) refcount */ 175 volatile u_int pr_uref; /* (r) user (alive) refcount */ 176 unsigned pr_flags; /* (p) PR_* flags */ 177 LIST_HEAD(, prison) pr_children; /* (a) list of child jails */ 178 LIST_HEAD(, proc) pr_proclist; /* (A) list of jailed processes */ 179 LIST_ENTRY(prison) pr_sibling; /* (a) next in parent's list */ 180 struct prison *pr_parent; /* (c) containing jail */ 181 struct mtx pr_mtx; 182 struct task pr_task; /* (c) destroy task */ 183 struct osd pr_osd; /* (p) additional data */ 184 struct cpuset *pr_cpuset; /* (p) cpuset */ 185 struct vnet *pr_vnet; /* (c) network stack */ 186 struct vnode *pr_root; /* (c) vnode to rdir */ 187 struct prison_ip *pr_addrs[PR_FAMILY_MAX]; /* (p,n) IPs of jail */ 188 struct prison_racct *pr_prison_racct; /* (c) racct jail proxy */ 189 void *pr_sparep[3]; 190 int pr_childcount; /* (a) number of child jails */ 191 int pr_childmax; /* (p) maximum child jails */ 192 unsigned pr_allow; /* (p) PR_ALLOW_* flags */ 193 int pr_securelevel; /* (p) securelevel */ 194 int pr_enforce_statfs; /* (p) statfs permission */ 195 int pr_devfs_rsnum; /* (p) devfs ruleset */ 196 enum prison_state pr_state; /* (q) state in life cycle */ 197 volatile int pr_exportcnt; /* (r) count of mount exports */ 198 int pr_spare; 199 int pr_osreldate; /* (c) kern.osreldate value */ 200 unsigned long pr_hostid; /* (p) jail hostid */ 201 char pr_name[MAXHOSTNAMELEN]; /* (p) admin jail name */ 202 char pr_path[MAXPATHLEN]; /* (c) chroot path */ 203 char pr_hostname[MAXHOSTNAMELEN]; /* (p) jail hostname */ 204 char pr_domainname[MAXHOSTNAMELEN]; /* (p) jail domainname */ 205 char pr_hostuuid[HOSTUUIDLEN]; /* (p) jail hostuuid */ 206 char pr_osrelease[OSRELEASELEN]; /* (c) kern.osrelease value */ 207}; 208 209struct prison_racct { 210 LIST_ENTRY(prison_racct) prr_next; 211 char prr_name[MAXHOSTNAMELEN]; 212 u_int prr_refcount; 213 struct racct *prr_racct; 214}; 215#endif /* _KERNEL || _WANT_PRISON */ 216 217#ifdef _KERNEL 218/* Flag bits set via options */ 219#define PR_PERSIST 0x00000001 /* Can exist without processes */ 220#define PR_HOST 0x00000002 /* Virtualize hostname et al */ 221#define PR_IP4_USER 0x00000004 /* Restrict IPv4 addresses */ 222#define PR_IP6_USER 0x00000008 /* Restrict IPv6 addresses */ 223#define PR_VNET 0x00000010 /* Virtual network stack */ 224#define PR_IP4_SADDRSEL 0x00000080 /* Do IPv4 src addr sel. or use the */ 225 /* primary jail address. */ 226#define PR_IP6_SADDRSEL 0x00000100 /* Do IPv6 src addr sel. or use the */ 227 /* primary jail address. */ 228 229/* Internal flag bits */ 230#define PR_REMOVE 0x01000000 /* In process of being removed */ 231#define PR_IP4 0x02000000 /* IPv4 restricted or disabled */ 232 /* by this jail or an ancestor */ 233#define PR_IP6 0x04000000 /* IPv6 restricted or disabled */ 234 /* by this jail or an ancestor */ 235#define PR_COMPLETE_PROC 0x08000000 /* prison_complete called from */ 236 /* prison_proc_free, releases uref */ 237 238/* 239 * Flags for pr_allow 240 * Bits not noted here may be used for dynamic allow.mount.xxxfs. 241 */ 242#define PR_ALLOW_SET_HOSTNAME 0x00000001 243#define PR_ALLOW_SYSVIPC 0x00000002 244#define PR_ALLOW_RAW_SOCKETS 0x00000004 245#define PR_ALLOW_CHFLAGS 0x00000008 246#define PR_ALLOW_MOUNT 0x00000010 247#define PR_ALLOW_QUOTAS 0x00000020 248#define PR_ALLOW_SOCKET_AF 0x00000040 249#define PR_ALLOW_MLOCK 0x00000080 250#define PR_ALLOW_READ_MSGBUF 0x00000100 251#define PR_ALLOW_UNPRIV_DEBUG 0x00000200 252#define PR_ALLOW_SUSER 0x00000400 253#define PR_ALLOW_RESERVED_PORTS 0x00008000 254#define PR_ALLOW_KMEM_ACCESS 0x00010000 /* reserved, not used yet */ 255#define PR_ALLOW_NFSD 0x00020000 256#define PR_ALLOW_EXTATTR 0x00040000 257#define PR_ALLOW_ALL_STATIC 0x000787ff 258 259/* 260 * PR_ALLOW_DIFFERENCES determines which flags are able to be 261 * different between the parent and child jail upon creation. 262 */ 263#define PR_ALLOW_DIFFERENCES (PR_ALLOW_UNPRIV_DEBUG) 264 265/* 266 * OSD methods 267 */ 268#define PR_METHOD_CREATE 0 269#define PR_METHOD_GET 1 270#define PR_METHOD_SET 2 271#define PR_METHOD_CHECK 3 272#define PR_METHOD_ATTACH 4 273#define PR_METHOD_REMOVE 5 274#define PR_MAXMETHOD 6 275 276/* 277 * Lock/unlock a prison. 278 * XXX These exist not so much for general convenience, but to be useable in 279 * the FOREACH_PRISON_DESCENDANT_LOCKED macro which can't handle them in 280 * non-function form as currently defined. 281 */ 282static __inline void 283prison_lock(struct prison *pr) 284{ 285 286 mtx_lock(&pr->pr_mtx); 287} 288 289static __inline void 290prison_unlock(struct prison *pr) 291{ 292 293 mtx_unlock(&pr->pr_mtx); 294} 295 296/* Traverse a prison's immediate children. */ 297#define FOREACH_PRISON_CHILD(ppr, cpr) \ 298 LIST_FOREACH(cpr, &(ppr)->pr_children, pr_sibling) 299 300/* 301 * Preorder traversal of all of a prison's descendants. 302 * This ugly loop allows the macro to be followed by a single block 303 * as expected in a looping primitive. 304 */ 305#define FOREACH_PRISON_DESCENDANT(ppr, cpr, descend) \ 306 for ((cpr) = (ppr), (descend) = 1; \ 307 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \ 308 ? LIST_FIRST(&(cpr)->pr_children) \ 309 : ((cpr) == (ppr) \ 310 ? NULL \ 311 : (((descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \ 312 ? LIST_NEXT(cpr, pr_sibling) \ 313 : (cpr)->pr_parent))));) \ 314 if (!(descend)) \ 315 ; \ 316 else 317 318/* 319 * As above, but lock descendants on the way down and unlock on the way up. 320 */ 321#define FOREACH_PRISON_DESCENDANT_LOCKED(ppr, cpr, descend) \ 322 for ((cpr) = (ppr), (descend) = 1; \ 323 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \ 324 ? LIST_FIRST(&(cpr)->pr_children) \ 325 : ((cpr) == (ppr) \ 326 ? NULL \ 327 : ((prison_unlock(cpr), \ 328 (descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \ 329 ? LIST_NEXT(cpr, pr_sibling) \ 330 : (cpr)->pr_parent))));) \ 331 if ((descend) ? (prison_lock(cpr), 0) : 1) \ 332 ; \ 333 else 334 335/* 336 * As above, but also keep track of the level descended to. 337 */ 338#define FOREACH_PRISON_DESCENDANT_LOCKED_LEVEL(ppr, cpr, descend, level)\ 339 for ((cpr) = (ppr), (descend) = 1, (level) = 0; \ 340 ((cpr) = (((descend) && !LIST_EMPTY(&(cpr)->pr_children)) \ 341 ? (level++, LIST_FIRST(&(cpr)->pr_children)) \ 342 : ((cpr) == (ppr) \ 343 ? NULL \ 344 : ((prison_unlock(cpr), \ 345 (descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \ 346 ? LIST_NEXT(cpr, pr_sibling) \ 347 : (level--, (cpr)->pr_parent)))));) \ 348 if ((descend) ? (prison_lock(cpr), 0) : 1) \ 349 ; \ 350 else 351 352/* 353 * Traverse a prison's descendants, visiting both preorder and postorder. 354 */ 355#define FOREACH_PRISON_DESCENDANT_PRE_POST(ppr, cpr, descend) \ 356 for ((cpr) = (ppr), (descend) = 1; \ 357 ((cpr) = (descend) \ 358 ? ((descend) = !LIST_EMPTY(&(cpr)->pr_children)) \ 359 ? LIST_FIRST(&(cpr)->pr_children) \ 360 : (cpr) \ 361 : ((descend) = LIST_NEXT(cpr, pr_sibling) != NULL) \ 362 ? LIST_NEXT(cpr, pr_sibling) \ 363 : cpr->pr_parent) != (ppr);) 364 365/* 366 * Attributes of the physical system, and the root of the jail tree. 367 */ 368extern struct prison prison0; 369 370TAILQ_HEAD(prisonlist, prison); 371extern struct prisonlist allprison; 372extern struct sx allprison_lock; 373 374/* 375 * Sysctls to describe jail parameters. 376 */ 377SYSCTL_DECL(_security_jail_param); 378 379#define SYSCTL_JAIL_PARAM(module, param, type, fmt, descr) \ 380 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \ 381 (type) | CTLFLAG_MPSAFE, NULL, 0, sysctl_jail_param, fmt, descr) 382#define SYSCTL_JAIL_PARAM_STRING(module, param, access, len, descr) \ 383 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \ 384 CTLTYPE_STRING | CTLFLAG_MPSAFE | (access), NULL, len, \ 385 sysctl_jail_param, "A", descr) 386#define SYSCTL_JAIL_PARAM_STRUCT(module, param, access, len, fmt, descr)\ 387 SYSCTL_PROC(_security_jail_param ## module, OID_AUTO, param, \ 388 CTLTYPE_STRUCT | CTLFLAG_MPSAFE | (access), NULL, len, \ 389 sysctl_jail_param, fmt, descr) 390#define SYSCTL_JAIL_PARAM_NODE(module, descr) \ 391 SYSCTL_NODE(_security_jail_param, OID_AUTO, module, CTLFLAG_MPSAFE, \ 392 0, descr) 393#define SYSCTL_JAIL_PARAM_SUBNODE(parent, module, descr) \ 394 SYSCTL_NODE(_security_jail_param_##parent, OID_AUTO, module, \ 395 CTLFLAG_MPSAFE, 0, descr) 396#define SYSCTL_JAIL_PARAM_SYS_NODE(module, access, descr) \ 397 SYSCTL_JAIL_PARAM_NODE(module, descr); \ 398 SYSCTL_JAIL_PARAM(_##module, , CTLTYPE_INT | (access), "E,jailsys", \ 399 descr) 400 401/* 402 * Kernel support functions for jail(). 403 */ 404struct ucred; 405struct mount; 406struct sockaddr; 407struct statfs; 408struct vfsconf; 409 410/* 411 * Return 1 if the passed credential is in a jail, otherwise 0. 412 */ 413#define jailed(cred) (cred->cr_prison != &prison0) 414 415bool jailed_without_vnet(struct ucred *); 416void getcredhostname(struct ucred *, char *, size_t); 417void getcreddomainname(struct ucred *, char *, size_t); 418void getcredhostuuid(struct ucred *, char *, size_t); 419void getcredhostid(struct ucred *, unsigned long *); 420void getjailname(struct ucred *cred, char *name, size_t len); 421void prison0_init(void); 422bool prison_allow(struct ucred *, unsigned); 423int prison_check(struct ucred *cred1, struct ucred *cred2); 424bool prison_check_nfsd(struct ucred *cred); 425bool prison_owns_vnet(struct ucred *); 426int prison_canseemount(struct ucred *cred, struct mount *mp); 427void prison_enforce_statfs(struct ucred *cred, struct mount *mp, 428 struct statfs *sp); 429struct prison *prison_find(int prid); 430struct prison *prison_find_child(struct prison *, int); 431struct prison *prison_find_name(struct prison *, const char *); 432bool prison_flag(struct ucred *, unsigned); 433void prison_free(struct prison *pr); 434void prison_free_locked(struct prison *pr); 435void prison_hold(struct prison *pr); 436void prison_hold_locked(struct prison *pr); 437void prison_proc_hold(struct prison *); 438void prison_proc_free(struct prison *); 439void prison_proc_link(struct prison *, struct proc *); 440void prison_proc_unlink(struct prison *, struct proc *); 441void prison_proc_iterate(struct prison *, void (*)(struct proc *, void *), void *); 442void prison_set_allow(struct ucred *cred, unsigned flag, int enable); 443bool prison_ischild(struct prison *, struct prison *); 444bool prison_isalive(const struct prison *); 445bool prison_isvalid(struct prison *); 446#if defined(INET) || defined(INET6) 447int prison_ip_check(const struct prison *, const pr_family_t, const void *); 448const void *prison_ip_get0(const struct prison *, const pr_family_t); 449u_int prison_ip_cnt(const struct prison *, const pr_family_t); 450#endif 451#ifdef INET 452bool prison_equal_ip4(struct prison *, struct prison *); 453int prison_get_ip4(struct ucred *cred, struct in_addr *ia); 454int prison_local_ip4(struct ucred *cred, struct in_addr *ia); 455int prison_remote_ip4(struct ucred *cred, struct in_addr *ia); 456int prison_check_ip4(const struct ucred *, const struct in_addr *); 457int prison_check_ip4_locked(const struct prison *, const struct in_addr *); 458bool prison_saddrsel_ip4(struct ucred *, struct in_addr *); 459int prison_qcmp_v4(const void *, const void *); 460bool prison_valid_v4(const void *); 461#endif 462#ifdef INET6 463bool prison_equal_ip6(struct prison *, struct prison *); 464int prison_get_ip6(struct ucred *, struct in6_addr *); 465int prison_local_ip6(struct ucred *, struct in6_addr *, int); 466int prison_remote_ip6(struct ucred *, struct in6_addr *); 467int prison_check_ip6(const struct ucred *, const struct in6_addr *); 468int prison_check_ip6_locked(const struct prison *, const struct in6_addr *); 469bool prison_saddrsel_ip6(struct ucred *, struct in6_addr *); 470int prison_qcmp_v6(const void *, const void *); 471bool prison_valid_v6(const void *); 472#endif 473int prison_check_af(struct ucred *cred, int af); 474int prison_if(struct ucred *cred, const struct sockaddr *sa); 475char *prison_name(struct prison *, struct prison *); 476int prison_priv_check(struct ucred *cred, int priv); 477int sysctl_jail_param(SYSCTL_HANDLER_ARGS); 478unsigned prison_add_allow(const char *prefix, const char *name, 479 const char *prefix_descr, const char *descr); 480void prison_add_vfs(struct vfsconf *vfsp); 481void prison_racct_foreach(void (*callback)(struct racct *racct, 482 void *arg2, void *arg3), void (*pre)(void), void (*post)(void), 483 void *arg2, void *arg3); 484struct prison_racct *prison_racct_find(const char *name); 485void prison_racct_hold(struct prison_racct *prr); 486void prison_racct_free(struct prison_racct *prr); 487 488#endif /* _KERNEL */ 489#endif /* !_SYS_JAIL_H_ */ 490