1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License, Version 1.0 only 6 * (the "License"). You may not use this file except in compliance 7 * with the License. 8 * 9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 10 * or http://www.opensolaris.org/os/licensing. 11 * See the License for the specific language governing permissions 12 * and limitations under the License. 13 * 14 * When distributing Covered Code, include this CDDL HEADER in each 15 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 16 * If applicable, add the following below this CDDL HEADER, with the 17 * fields enclosed by brackets "[]" replaced with your own identifying 18 * information: Portions Copyright [yyyy] [name of copyright owner] 19 * 20 * CDDL HEADER END 21 */ 22 /* 23 * Copyright 2005 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #ifndef _SYS_ZONE_H 28 #define _SYS_ZONE_H 29 30 #pragma ident "%Z%%M% %I% %E% SMI" 31 32 #include <sys/types.h> 33 #include <sys/mutex.h> 34 #include <sys/param.h> 35 #include <sys/rctl.h> 36 #include <sys/pset.h> 37 38 #ifdef __cplusplus 39 extern "C" { 40 #endif 41 42 /* 43 * NOTE 44 * 45 * The contents of this file are private to the implementation of 46 * Solaris and are subject to change at any time without notice. 47 * Applications and drivers using these interfaces may fail to 48 * run on future releases. 49 */ 50 51 /* Available both in kernel and for user space */ 52 53 /* zone id restrictions and special ids */ 54 #define MAX_ZONEID 9999 55 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */ 56 #define MIN_ZONEID 0 /* minimum zone ID on system */ 57 #define GLOBAL_ZONEID 0 58 #define ZONEID_WIDTH 4 /* for printf */ 59 60 /* system call subcodes */ 61 #define ZONE_CREATE 0 62 #define ZONE_DESTROY 1 63 #define ZONE_GETATTR 2 64 #define ZONE_ENTER 3 65 #define ZONE_LIST 4 66 #define ZONE_SHUTDOWN 5 67 #define ZONE_LOOKUP 6 68 #define ZONE_BOOT 7 69 #define ZONE_VERSION 8 70 71 /* zone attributes */ 72 #define ZONE_ATTR_ROOT 1 73 #define ZONE_ATTR_NAME 2 74 #define ZONE_ATTR_STATUS 3 75 #define ZONE_ATTR_PRIVSET 4 76 #define ZONE_ATTR_UNIQID 5 77 #define ZONE_ATTR_POOLID 6 78 #define ZONE_ATTR_INITPID 7 79 80 #define ZONE_EVENT_CHANNEL "com.sun:zones:status" 81 #define ZONE_EVENT_STATUS_CLASS "status" 82 #define ZONE_EVENT_STATUS_SUBCLASS "change" 83 84 #define ZONE_EVENT_UNINITIALIZED "uninitialized" 85 #define ZONE_EVENT_READY "ready" 86 #define ZONE_EVENT_RUNNING "running" 87 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down" 88 89 #define ZONE_CB_NAME "zonename" 90 #define ZONE_CB_NEWSTATE "newstate" 91 #define ZONE_CB_OLDSTATE "oldstate" 92 #define ZONE_CB_TIMESTAMP "when" 93 #define ZONE_CB_ZONEID "zoneid" 94 95 #ifdef _SYSCALL32 96 typedef struct { 97 caddr32_t zone_name; 98 caddr32_t zone_root; 99 caddr32_t zone_privs; 100 size32_t zone_privssz; 101 caddr32_t rctlbuf; 102 size32_t rctlbufsz; 103 caddr32_t extended_error; 104 caddr32_t zfsbuf; 105 size32_t zfsbufsz; 106 } zone_def32; 107 #endif 108 typedef struct { 109 const char *zone_name; 110 const char *zone_root; 111 const struct priv_set *zone_privs; 112 size_t zone_privssz; 113 const char *rctlbuf; 114 size_t rctlbufsz; 115 int *extended_error; 116 const char *zfsbuf; 117 size_t zfsbufsz; 118 } zone_def; 119 120 /* extended error information */ 121 #define ZE_UNKNOWN 0 /* No extended error info */ 122 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */ 123 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */ 124 125 /* zone_status */ 126 typedef enum { 127 ZONE_IS_UNINITIALIZED = 0, 128 ZONE_IS_READY, 129 ZONE_IS_BOOTING, 130 ZONE_IS_RUNNING, 131 ZONE_IS_SHUTTING_DOWN, 132 ZONE_IS_EMPTY, 133 ZONE_IS_DOWN, 134 ZONE_IS_DYING, 135 ZONE_IS_DEAD 136 } zone_status_t; 137 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED 138 #define ZONE_MAX_STATE ZONE_IS_DEAD 139 140 /* 141 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also 142 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT. 143 */ 144 typedef enum zone_cmd { 145 Z_READY, Z_BOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING, 146 Z_MOUNT, Z_UNMOUNT 147 } zone_cmd_t; 148 149 #define ZONEBOOTARGS_MAX 257 /* uadmin()'s buffer is 257 bytes. */ 150 151 /* 152 * The structure of a request to zoneadmd. 153 */ 154 typedef struct zone_cmd_arg { 155 uint64_t uniqid; /* unique "generation number" */ 156 zone_cmd_t cmd; /* requested action */ 157 char locale[MAXPATHLEN]; /* locale in which to render messages */ 158 char bootbuf[ZONEBOOTARGS_MAX]; /* arguments passed to zone_boot() */ 159 } zone_cmd_arg_t; 160 161 /* 162 * Structure of zoneadmd's response to a request. A NULL return value means 163 * the caller should attempt to restart zoneadmd and retry. 164 */ 165 typedef struct zone_cmd_rval { 166 int rval; /* return value of request */ 167 char errbuf[1]; /* variable-sized buffer containing error messages */ 168 } zone_cmd_rval_t; 169 170 /* 171 * The zone support infrastructure uses the zone name as a component 172 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters 173 * in length, so ZONENAME_MAX is limited by that. 174 */ 175 #define ZONENAME_MAX 64 176 177 #define GLOBAL_ZONENAME "global" 178 179 /* 180 * Extended Regular expression (see regex(5)) which matches all valid zone 181 * names. 182 */ 183 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}" 184 185 /* 186 * Where the zones support infrastructure places temporary files. 187 */ 188 #define ZONES_TMPDIR "/var/run/zones" 189 190 /* 191 * The path to the door used by clients to communicate with zoneadmd. 192 */ 193 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door" 194 195 196 #ifdef _KERNEL 197 /* 198 * We need to protect the definition of 'list_t' from userland applications and 199 * libraries which may be defining ther own versions. 200 */ 201 #include <sys/list.h> 202 203 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */ 204 205 /* zone_flags */ 206 #define ZF_DESTROYED 0x1 /* ZSD destructor callbacks run */ 207 208 struct pool; 209 210 /* 211 * Structure to record list of ZFS datasets exported to a zone. 212 */ 213 typedef struct zone_dataset { 214 char *zd_dataset; 215 list_node_t zd_linkage; 216 } zone_dataset_t; 217 218 typedef struct zone { 219 /* 220 * zone_name is never modified once set. 221 */ 222 char *zone_name; /* zone's configuration name */ 223 /* 224 * zone_nodename and zone_domain are never freed once allocated. 225 */ 226 char *zone_nodename; /* utsname.nodename equivalent */ 227 char *zone_domain; /* srpc_domain equivalent */ 228 /* 229 * zone_lock protects the following fields of a zone_t: 230 * zone_ref 231 * zone_cred_ref 232 * zone_ntasks 233 * zone_flags 234 * zone_zsd 235 */ 236 kmutex_t zone_lock; 237 /* 238 * zone_linkage is the zone's linkage into the active or 239 * death-row list. The field is protected by zonehash_lock. 240 */ 241 list_node_t zone_linkage; 242 zoneid_t zone_id; /* ID of zone */ 243 uint_t zone_ref; /* count of zone_hold()s on zone */ 244 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */ 245 /* 246 * zone_rootvp and zone_rootpath can never be modified once set. 247 */ 248 struct vnode *zone_rootvp; /* zone's root vnode */ 249 char *zone_rootpath; /* Path to zone's root + '/' */ 250 ushort_t zone_flags; /* misc flags */ 251 zone_status_t zone_status; /* protected by zone_status_lock */ 252 uint_t zone_ntasks; /* number of tasks executing in zone */ 253 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */ 254 /* counters in projects and tasks */ 255 /* that are within the zone */ 256 rctl_qty_t zone_nlwps; /* number of lwps in zone */ 257 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */ 258 259 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */ 260 uint32_t zone_shares; /* FSS shares allocated to zone */ 261 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */ 262 list_t zone_zsd; /* list of Zone-Specific Data values */ 263 kcondvar_t zone_cv; /* used to signal state changes */ 264 struct proc *zone_zsched; /* Dummy kernel "zsched" process */ 265 pid_t zone_proc_initpid; /* pid of "init" for this zone */ 266 int zone_boot_err; /* for zone_boot() if boot fails */ 267 char *zone_bootargs; /* arguments passed via zone_boot() */ 268 /* 269 * zone_kthreads is protected by zone_status_lock. 270 */ 271 kthread_t *zone_kthreads; /* kernel threads in zone */ 272 struct priv_set *zone_privset; /* limit set for zone */ 273 /* 274 * zone_vfslist is protected by vfs_list_lock(). 275 */ 276 struct vfs *zone_vfslist; /* list of FS's mounted in zone */ 277 uint64_t zone_uniqid; /* unique zone generation number */ 278 struct cred *zone_kcred; /* kcred-like, zone-limited cred */ 279 /* 280 * zone_pool is protected by pool_lock(). 281 */ 282 struct pool *zone_pool; /* pool the zone is bound to */ 283 hrtime_t zone_pool_mod; /* last pool bind modification time */ 284 /* zone_psetid is protected by cpu_lock */ 285 psetid_t zone_psetid; /* pset the zone is bound to */ 286 /* 287 * The following two can be read without holding any locks. They are 288 * updated under cpu_lock. 289 */ 290 int zone_ncpus; /* zone's idea of ncpus */ 291 int zone_ncpus_online; /* zone's idea of ncpus_online */ 292 /* 293 * List of ZFS datasets exported to this zone. 294 */ 295 list_t zone_datasets; /* list of datasets */ 296 } zone_t; 297 298 /* 299 * Special value of zone_psetid to indicate that pools are disabled. 300 */ 301 #define ZONE_PS_INVAL PS_MYID 302 303 extern zone_t zone0; 304 extern zone_t *global_zone; 305 extern uint_t maxzones; 306 extern rctl_hndl_t rc_zone_nlwps; 307 308 extern const char * const zone_initname; 309 310 extern long zone(int, void *, void *, void *, void *); 311 extern void zone_zsd_init(void); 312 extern void zone_init(void); 313 extern void zone_hold(zone_t *); 314 extern void zone_rele(zone_t *); 315 extern void zone_cred_hold(zone_t *); 316 extern void zone_cred_rele(zone_t *); 317 extern void zone_task_hold(zone_t *); 318 extern void zone_task_rele(zone_t *); 319 extern zone_t *zone_find_by_id(zoneid_t); 320 extern zone_t *zone_find_by_name(char *); 321 extern zone_t *zone_find_by_path(const char *); 322 extern zoneid_t getzoneid(void); 323 324 /* 325 * Zone-specific data (ZSD) APIs 326 */ 327 /* 328 * The following is what code should be initializing its zone_key_t to if it 329 * calls zone_getspecific() without necessarily knowing that zone_key_create() 330 * has been called on the key. 331 */ 332 #define ZONE_KEY_UNINITIALIZED 0 333 334 typedef uint_t zone_key_t; 335 336 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t), 337 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *)); 338 extern int zone_key_delete(zone_key_t); 339 extern void *zone_getspecific(zone_key_t, zone_t *); 340 extern int zone_setspecific(zone_key_t, zone_t *, const void *); 341 342 /* 343 * The definition of a zsd_entry is truly private to zone.c and is only 344 * placed here so it can be shared with mdb. 345 */ 346 struct zsd_entry { 347 zone_key_t zsd_key; /* Key used to lookup value */ 348 void *zsd_data; /* Caller-managed value */ 349 /* 350 * Callbacks to be executed when a zone is created, shutdown, and 351 * destroyed, respectively. 352 */ 353 void *(*zsd_create)(zoneid_t); 354 void (*zsd_shutdown)(zoneid_t, void *); 355 void (*zsd_destroy)(zoneid_t, void *); 356 list_node_t zsd_linkage; 357 }; 358 359 /* 360 * Macros to help with zone visibility restrictions. 361 */ 362 363 /* 364 * Is process in the global zone? 365 */ 366 #define INGLOBALZONE(p) \ 367 ((p)->p_zone == global_zone) 368 369 /* 370 * Can process view objects in given zone? 371 */ 372 #define HASZONEACCESS(p, zoneid) \ 373 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p)) 374 375 /* 376 * Convenience macro to see if a resolved path is visible from within a 377 * given zone. 378 * 379 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the 380 * two strings must be equal. Since the rootpathlen has a trailing '/', 381 * we want to skip everything in the path up to (but not including) the 382 * trailing '/'. 383 */ 384 #define ZONE_PATH_VISIBLE(path, zone) \ 385 (strncmp((path), (zone)->zone_rootpath, \ 386 (zone)->zone_rootpathlen - 1) == 0) 387 388 /* 389 * Convenience macro to go from the global view of a path to that seen 390 * from within said zone. It is the responsibility of the caller to 391 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is 392 * in fact visible from within the zone. 393 */ 394 #define ZONE_PATH_TRANSLATE(path, zone) \ 395 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \ 396 (path) + (zone)->zone_rootpathlen - 2) 397 398 /* 399 * Special processes visible in all zones. 400 */ 401 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1) 402 403 /* 404 * Special zoneid_t token to refer to all zones. 405 */ 406 #define ALL_ZONES (-1) 407 408 /* 409 * Zone-safe version of thread_create() to be used when the caller wants to 410 * create a kernel thread to run within the current zone's context. 411 */ 412 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t, 413 pri_t); 414 extern void zthread_exit(void); 415 416 /* 417 * Functions for an external observer to register interest in a zone's status 418 * change. Observers will be woken up when the zone status equals the status 419 * argument passed in (in the case of zone_status_timedwait, the function may 420 * also return because of a timeout; zone_status_wait_sig may return early due 421 * to a signal being delivered; zone_status_timedwait_sig may return for any of 422 * the above reasons). 423 * 424 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and 425 * cv_wait_sig() respectively. 426 */ 427 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t); 428 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t); 429 extern void zone_status_wait(zone_t *, zone_status_t); 430 extern int zone_status_wait_sig(zone_t *, zone_status_t); 431 432 /* 433 * Get the status of the zone (at the time it was called). The state may 434 * have progressed by the time it is returned. 435 */ 436 extern zone_status_t zone_status_get(zone_t *); 437 438 /* 439 * Get the "kcred" credentials corresponding to the given zone. 440 */ 441 extern struct cred *zone_get_kcred(zoneid_t); 442 443 /* 444 * Get/set the pool the zone is currently bound to. 445 */ 446 extern struct pool *zone_pool_get(zone_t *); 447 extern void zone_pool_set(zone_t *, struct pool *); 448 449 /* 450 * Get/set the pset the zone is currently using. 451 */ 452 extern psetid_t zone_pset_get(zone_t *); 453 extern void zone_pset_set(zone_t *, psetid_t); 454 455 /* 456 * Get the number of cpus/online-cpus visible from the given zone. 457 */ 458 extern int zone_ncpus_get(zone_t *); 459 extern int zone_ncpus_online_get(zone_t *); 460 461 /* 462 * Returns true if the named pool/dataset is visible in the current zone. 463 */ 464 extern int zone_dataset_visible(const char *, int *); 465 466 /* 467 * zone version of uadmin() 468 */ 469 extern int zone_uadmin(int, int, struct cred *); 470 extern void zone_shutdown_global(void); 471 472 extern void mount_in_progress(void); 473 extern void mount_completed(void); 474 475 extern int zone_walk(int (*)(zone_t *, void *), void *); 476 477 #endif /* _KERNEL */ 478 479 #ifdef __cplusplus 480 } 481 #endif 482 483 #endif /* _SYS_ZONE_H */ 484