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 (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 /* 22 * Copyright 2007 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _SYS_ZONE_H 27 #define _SYS_ZONE_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #include <sys/types.h> 32 #include <sys/mutex.h> 33 #include <sys/param.h> 34 #include <sys/rctl.h> 35 #include <sys/ipc_rctl.h> 36 #include <sys/pset.h> 37 #include <sys/tsol/label.h> 38 #include <sys/cred.h> 39 #include <sys/netstack.h> 40 #include <sys/uadmin.h> 41 42 #ifdef __cplusplus 43 extern "C" { 44 #endif 45 46 /* 47 * NOTE 48 * 49 * The contents of this file are private to the implementation of 50 * Solaris and are subject to change at any time without notice. 51 * Applications and drivers using these interfaces may fail to 52 * run on future releases. 53 */ 54 55 /* Available both in kernel and for user space */ 56 57 /* zone id restrictions and special ids */ 58 #define MAX_ZONEID 9999 59 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */ 60 #define MIN_ZONEID 0 /* minimum zone ID on system */ 61 #define GLOBAL_ZONEID 0 62 #define ZONEID_WIDTH 4 /* for printf */ 63 64 /* 65 * Special zoneid_t token to refer to all zones. 66 */ 67 #define ALL_ZONES (-1) 68 69 /* system call subcodes */ 70 #define ZONE_CREATE 0 71 #define ZONE_DESTROY 1 72 #define ZONE_GETATTR 2 73 #define ZONE_ENTER 3 74 #define ZONE_LIST 4 75 #define ZONE_SHUTDOWN 5 76 #define ZONE_LOOKUP 6 77 #define ZONE_BOOT 7 78 #define ZONE_VERSION 8 79 #define ZONE_SETATTR 9 80 #define ZONE_ADD_DATALINK 10 81 #define ZONE_DEL_DATALINK 11 82 #define ZONE_CHECK_DATALINK 12 83 #define ZONE_LIST_DATALINK 13 84 85 /* zone attributes */ 86 #define ZONE_ATTR_ROOT 1 87 #define ZONE_ATTR_NAME 2 88 #define ZONE_ATTR_STATUS 3 89 #define ZONE_ATTR_PRIVSET 4 90 #define ZONE_ATTR_UNIQID 5 91 #define ZONE_ATTR_POOLID 6 92 #define ZONE_ATTR_INITPID 7 93 #define ZONE_ATTR_SLBL 8 94 #define ZONE_ATTR_INITNAME 9 95 #define ZONE_ATTR_BOOTARGS 10 96 #define ZONE_ATTR_BRAND 11 97 #define ZONE_ATTR_PHYS_MCAP 12 98 #define ZONE_ATTR_SCHED_CLASS 13 99 #define ZONE_ATTR_FLAGS 14 100 101 /* Start of the brand-specific attribute namespace */ 102 #define ZONE_ATTR_BRAND_ATTRS 32768 103 104 #define ZONE_EVENT_CHANNEL "com.sun:zones:status" 105 #define ZONE_EVENT_STATUS_CLASS "status" 106 #define ZONE_EVENT_STATUS_SUBCLASS "change" 107 108 #define ZONE_EVENT_UNINITIALIZED "uninitialized" 109 #define ZONE_EVENT_READY "ready" 110 #define ZONE_EVENT_RUNNING "running" 111 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down" 112 113 #define ZONE_CB_NAME "zonename" 114 #define ZONE_CB_NEWSTATE "newstate" 115 #define ZONE_CB_OLDSTATE "oldstate" 116 #define ZONE_CB_TIMESTAMP "when" 117 #define ZONE_CB_ZONEID "zoneid" 118 119 /* 120 * Exit values that may be returned by scripts or programs invoked by various 121 * zone commands. 122 * 123 * These are defined as: 124 * 125 * ZONE_SUBPROC_OK 126 * =============== 127 * The subprocess completed successfully. 128 * 129 * ZONE_SUBPROC_USAGE 130 * ================== 131 * The subprocess failed with a usage message, or a usage message should 132 * be output in its behalf. 133 * 134 * ZONE_SUBPROC_NOTCOMPLETE 135 * ======================== 136 * The subprocess did not complete, but the actions performed by the 137 * subprocess require no recovery actions by the user. 138 * 139 * For example, if the subprocess were called by "zoneadm install," the 140 * installation of the zone did not succeed but the user need not perform 141 * a "zoneadm uninstall" before attempting another install. 142 * 143 * ZONE_SUBPROC_FATAL 144 * ================== 145 * The subprocess failed in a fatal manner, usually one that will require 146 * some type of recovery action by the user. 147 * 148 * For example, if the subprocess were called by "zoneadm install," the 149 * installation of the zone did not succeed and the user will need to 150 * perform a "zoneadm uninstall" before another install attempt is 151 * possible. 152 * 153 * The non-success exit values are large to avoid accidental collision 154 * with values used internally by some commands (e.g. "Z_ERR" and 155 * "Z_USAGE" as used by zoneadm.) 156 */ 157 #define ZONE_SUBPROC_OK 0 158 #define ZONE_SUBPROC_USAGE 253 159 #define ZONE_SUBPROC_NOTCOMPLETE 254 160 #define ZONE_SUBPROC_FATAL 255 161 162 #ifdef _SYSCALL32 163 typedef struct { 164 caddr32_t zone_name; 165 caddr32_t zone_root; 166 caddr32_t zone_privs; 167 size32_t zone_privssz; 168 caddr32_t rctlbuf; 169 size32_t rctlbufsz; 170 caddr32_t extended_error; 171 caddr32_t zfsbuf; 172 size32_t zfsbufsz; 173 int match; /* match level */ 174 uint32_t doi; /* DOI for label */ 175 caddr32_t label; /* label associated with zone */ 176 int flags; 177 } zone_def32; 178 #endif 179 typedef struct { 180 const char *zone_name; 181 const char *zone_root; 182 const struct priv_set *zone_privs; 183 size_t zone_privssz; 184 const char *rctlbuf; 185 size_t rctlbufsz; 186 int *extended_error; 187 const char *zfsbuf; 188 size_t zfsbufsz; 189 int match; /* match level */ 190 uint32_t doi; /* DOI for label */ 191 const bslabel_t *label; /* label associated with zone */ 192 int flags; 193 } zone_def; 194 195 /* extended error information */ 196 #define ZE_UNKNOWN 0 /* No extended error info */ 197 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */ 198 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */ 199 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */ 200 201 /* zone_status */ 202 typedef enum { 203 ZONE_IS_UNINITIALIZED = 0, 204 ZONE_IS_READY, 205 ZONE_IS_BOOTING, 206 ZONE_IS_RUNNING, 207 ZONE_IS_SHUTTING_DOWN, 208 ZONE_IS_EMPTY, 209 ZONE_IS_DOWN, 210 ZONE_IS_DYING, 211 ZONE_IS_DEAD 212 } zone_status_t; 213 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED 214 #define ZONE_MAX_STATE ZONE_IS_DEAD 215 216 /* 217 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also 218 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT. 219 */ 220 typedef enum zone_cmd { 221 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING, 222 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT 223 } zone_cmd_t; 224 225 /* 226 * The structure of a request to zoneadmd. 227 */ 228 typedef struct zone_cmd_arg { 229 uint64_t uniqid; /* unique "generation number" */ 230 zone_cmd_t cmd; /* requested action */ 231 uint32_t _pad; /* need consistent 32/64 bit alignmt */ 232 char locale[MAXPATHLEN]; /* locale in which to render messages */ 233 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */ 234 } zone_cmd_arg_t; 235 236 /* 237 * Structure of zoneadmd's response to a request. A NULL return value means 238 * the caller should attempt to restart zoneadmd and retry. 239 */ 240 typedef struct zone_cmd_rval { 241 int rval; /* return value of request */ 242 char errbuf[1]; /* variable-sized buffer containing error messages */ 243 } zone_cmd_rval_t; 244 245 /* 246 * The zone support infrastructure uses the zone name as a component 247 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters 248 * in length, so ZONENAME_MAX is limited by that. 249 */ 250 #define ZONENAME_MAX 64 251 252 #define GLOBAL_ZONENAME "global" 253 254 /* 255 * Extended Regular expression (see regex(5)) which matches all valid zone 256 * names. 257 */ 258 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}" 259 260 /* 261 * Where the zones support infrastructure places temporary files. 262 */ 263 #define ZONES_TMPDIR "/var/run/zones" 264 265 /* 266 * The path to the door used by clients to communicate with zoneadmd. 267 */ 268 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door" 269 270 /* zone_flags */ 271 #define ZF_DESTROYED 0x1 /* ZSD destructor callbacks run */ 272 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */ 273 #define ZF_IS_SCRATCH 0x4 /* scratch zone */ 274 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */ 275 276 /* zone_create flags */ 277 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */ 278 279 #ifdef _KERNEL 280 /* 281 * We need to protect the definition of 'list_t' from userland applications and 282 * libraries which may be defining ther own versions. 283 */ 284 #include <sys/list.h> 285 286 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */ 287 288 struct pool; 289 struct brand; 290 struct dlnamelist; 291 292 /* 293 * Structure to record list of ZFS datasets exported to a zone. 294 */ 295 typedef struct zone_dataset { 296 char *zd_dataset; 297 list_node_t zd_linkage; 298 } zone_dataset_t; 299 300 /* 301 * structure for zone kstats 302 */ 303 typedef struct zone_kstat { 304 kstat_named_t zk_zonename; 305 kstat_named_t zk_usage; 306 kstat_named_t zk_value; 307 } zone_kstat_t; 308 309 struct cpucap; 310 311 typedef struct zone { 312 /* 313 * zone_name is never modified once set. 314 */ 315 char *zone_name; /* zone's configuration name */ 316 /* 317 * zone_nodename and zone_domain are never freed once allocated. 318 */ 319 char *zone_nodename; /* utsname.nodename equivalent */ 320 char *zone_domain; /* srpc_domain equivalent */ 321 /* 322 * zone_lock protects the following fields of a zone_t: 323 * zone_ref 324 * zone_cred_ref 325 * zone_ntasks 326 * zone_flags 327 * zone_zsd 328 */ 329 kmutex_t zone_lock; 330 /* 331 * zone_linkage is the zone's linkage into the active or 332 * death-row list. The field is protected by zonehash_lock. 333 */ 334 list_node_t zone_linkage; 335 zoneid_t zone_id; /* ID of zone */ 336 uint_t zone_ref; /* count of zone_hold()s on zone */ 337 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */ 338 /* 339 * zone_rootvp and zone_rootpath can never be modified once set. 340 */ 341 struct vnode *zone_rootvp; /* zone's root vnode */ 342 char *zone_rootpath; /* Path to zone's root + '/' */ 343 ushort_t zone_flags; /* misc flags */ 344 zone_status_t zone_status; /* protected by zone_status_lock */ 345 uint_t zone_ntasks; /* number of tasks executing in zone */ 346 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */ 347 /* counters in projects and tasks */ 348 /* that are within the zone */ 349 rctl_qty_t zone_nlwps; /* number of lwps in zone */ 350 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */ 351 rctl_qty_t zone_shmmax; /* System V shared memory usage */ 352 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */ 353 354 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */ 355 uint32_t zone_shares; /* FSS shares allocated to zone */ 356 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */ 357 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */ 358 /* kpd_locked_mem for all */ 359 /* projects in zone. */ 360 /* Also protects zone_max_swap */ 361 /* grab after p_lock, before rcs_lock */ 362 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */ 363 /* zone */ 364 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */ 365 /* limit. Protected by */ 366 /* zone_rctls->rcs_lock */ 367 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */ 368 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */ 369 /* Protected by */ 370 /* zone_rctls->rcs_lock */ 371 list_t zone_zsd; /* list of Zone-Specific Data values */ 372 kcondvar_t zone_cv; /* used to signal state changes */ 373 struct proc *zone_zsched; /* Dummy kernel "zsched" process */ 374 pid_t zone_proc_initpid; /* pid of "init" for this zone */ 375 char *zone_initname; /* fs path to 'init' */ 376 int zone_boot_err; /* for zone_boot() if boot fails */ 377 char *zone_bootargs; /* arguments passed via zone_boot() */ 378 uint64_t zone_phys_mcap; /* physical memory cap */ 379 /* 380 * zone_kthreads is protected by zone_status_lock. 381 */ 382 kthread_t *zone_kthreads; /* kernel threads in zone */ 383 struct priv_set *zone_privset; /* limit set for zone */ 384 /* 385 * zone_vfslist is protected by vfs_list_lock(). 386 */ 387 struct vfs *zone_vfslist; /* list of FS's mounted in zone */ 388 uint64_t zone_uniqid; /* unique zone generation number */ 389 struct cred *zone_kcred; /* kcred-like, zone-limited cred */ 390 /* 391 * zone_pool is protected by pool_lock(). 392 */ 393 struct pool *zone_pool; /* pool the zone is bound to */ 394 hrtime_t zone_pool_mod; /* last pool bind modification time */ 395 /* zone_psetid is protected by cpu_lock */ 396 psetid_t zone_psetid; /* pset the zone is bound to */ 397 /* 398 * The following two can be read without holding any locks. They are 399 * updated under cpu_lock. 400 */ 401 int zone_ncpus; /* zone's idea of ncpus */ 402 int zone_ncpus_online; /* zone's idea of ncpus_online */ 403 /* 404 * List of ZFS datasets exported to this zone. 405 */ 406 list_t zone_datasets; /* list of datasets */ 407 408 ts_label_t *zone_slabel; /* zone sensitivity label */ 409 int zone_match; /* require label match for packets */ 410 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */ 411 412 boolean_t zone_restart_init; /* Restart init if it dies? */ 413 struct brand *zone_brand; /* zone's brand */ 414 id_t zone_defaultcid; /* dflt scheduling class id */ 415 kstat_t *zone_swapresv_kstat; 416 kstat_t *zone_lockedmem_kstat; 417 /* 418 * zone_dl_list is protected by zone_lock 419 */ 420 struct dlnamelist *zone_dl_list; 421 netstack_t *zone_netstack; 422 struct cpucap *zone_cpucap; /* CPU caps data */ 423 /* 424 * Solaris Auditing per-zone audit context 425 */ 426 struct au_kcontext *zone_audit_kctxt; 427 } zone_t; 428 429 /* 430 * Special value of zone_psetid to indicate that pools are disabled. 431 */ 432 #define ZONE_PS_INVAL PS_MYID 433 434 435 extern zone_t zone0; 436 extern zone_t *global_zone; 437 extern uint_t maxzones; 438 extern rctl_hndl_t rc_zone_nlwps; 439 440 extern long zone(int, void *, void *, void *, void *); 441 extern void zone_zsd_init(void); 442 extern void zone_init(void); 443 extern void zone_hold(zone_t *); 444 extern void zone_rele(zone_t *); 445 extern void zone_cred_hold(zone_t *); 446 extern void zone_cred_rele(zone_t *); 447 extern void zone_task_hold(zone_t *); 448 extern void zone_task_rele(zone_t *); 449 extern zone_t *zone_find_by_id(zoneid_t); 450 extern zone_t *zone_find_by_label(const ts_label_t *); 451 extern zone_t *zone_find_by_name(char *); 452 extern zone_t *zone_find_by_any_path(const char *, boolean_t); 453 extern zone_t *zone_find_by_path(const char *); 454 extern zoneid_t getzoneid(void); 455 extern zone_t *zone_find_by_id_nolock(zoneid_t); 456 457 /* 458 * Zone-specific data (ZSD) APIs 459 */ 460 /* 461 * The following is what code should be initializing its zone_key_t to if it 462 * calls zone_getspecific() without necessarily knowing that zone_key_create() 463 * has been called on the key. 464 */ 465 #define ZONE_KEY_UNINITIALIZED 0 466 467 typedef uint_t zone_key_t; 468 469 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t), 470 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *)); 471 extern int zone_key_delete(zone_key_t); 472 extern void *zone_getspecific(zone_key_t, zone_t *); 473 extern int zone_setspecific(zone_key_t, zone_t *, const void *); 474 475 /* 476 * The definition of a zsd_entry is truly private to zone.c and is only 477 * placed here so it can be shared with mdb. 478 */ 479 struct zsd_entry { 480 zone_key_t zsd_key; /* Key used to lookup value */ 481 void *zsd_data; /* Caller-managed value */ 482 /* 483 * Callbacks to be executed when a zone is created, shutdown, and 484 * destroyed, respectively. 485 */ 486 void *(*zsd_create)(zoneid_t); 487 void (*zsd_shutdown)(zoneid_t, void *); 488 void (*zsd_destroy)(zoneid_t, void *); 489 list_node_t zsd_linkage; 490 }; 491 492 /* 493 * Macros to help with zone visibility restrictions. 494 */ 495 496 /* 497 * Is process in the global zone? 498 */ 499 #define INGLOBALZONE(p) \ 500 ((p)->p_zone == global_zone) 501 502 /* 503 * Can process view objects in given zone? 504 */ 505 #define HASZONEACCESS(p, zoneid) \ 506 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p)) 507 508 /* 509 * Convenience macro to see if a resolved path is visible from within a 510 * given zone. 511 * 512 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the 513 * two strings must be equal. Since the rootpathlen has a trailing '/', 514 * we want to skip everything in the path up to (but not including) the 515 * trailing '/'. 516 */ 517 #define ZONE_PATH_VISIBLE(path, zone) \ 518 (strncmp((path), (zone)->zone_rootpath, \ 519 (zone)->zone_rootpathlen - 1) == 0) 520 521 /* 522 * Convenience macro to go from the global view of a path to that seen 523 * from within said zone. It is the responsibility of the caller to 524 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is 525 * in fact visible from within the zone. 526 */ 527 #define ZONE_PATH_TRANSLATE(path, zone) \ 528 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \ 529 (path) + (zone)->zone_rootpathlen - 2) 530 531 /* 532 * Special processes visible in all zones. 533 */ 534 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1) 535 536 /* 537 * Zone-safe version of thread_create() to be used when the caller wants to 538 * create a kernel thread to run within the current zone's context. 539 */ 540 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t, 541 pri_t); 542 extern void zthread_exit(void); 543 544 /* 545 * Functions for an external observer to register interest in a zone's status 546 * change. Observers will be woken up when the zone status equals the status 547 * argument passed in (in the case of zone_status_timedwait, the function may 548 * also return because of a timeout; zone_status_wait_sig may return early due 549 * to a signal being delivered; zone_status_timedwait_sig may return for any of 550 * the above reasons). 551 * 552 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and 553 * cv_wait_sig() respectively. 554 */ 555 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t); 556 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t); 557 extern void zone_status_wait(zone_t *, zone_status_t); 558 extern int zone_status_wait_sig(zone_t *, zone_status_t); 559 560 /* 561 * Get the status of the zone (at the time it was called). The state may 562 * have progressed by the time it is returned. 563 */ 564 extern zone_status_t zone_status_get(zone_t *); 565 566 /* 567 * Get the "kcred" credentials corresponding to the given zone. 568 */ 569 extern struct cred *zone_get_kcred(zoneid_t); 570 571 /* 572 * Get/set the pool the zone is currently bound to. 573 */ 574 extern struct pool *zone_pool_get(zone_t *); 575 extern void zone_pool_set(zone_t *, struct pool *); 576 577 /* 578 * Get/set the pset the zone is currently using. 579 */ 580 extern psetid_t zone_pset_get(zone_t *); 581 extern void zone_pset_set(zone_t *, psetid_t); 582 583 /* 584 * Get the number of cpus/online-cpus visible from the given zone. 585 */ 586 extern int zone_ncpus_get(zone_t *); 587 extern int zone_ncpus_online_get(zone_t *); 588 589 /* 590 * Returns true if the named pool/dataset is visible in the current zone. 591 */ 592 extern int zone_dataset_visible(const char *, int *); 593 594 /* 595 * zone version of kadmin() 596 */ 597 extern int zone_kadmin(int, int, const char *, cred_t *); 598 extern void zone_shutdown_global(void); 599 600 extern void mount_in_progress(void); 601 extern void mount_completed(void); 602 603 extern int zone_walk(int (*)(zone_t *, void *), void *); 604 605 extern rctl_hndl_t rc_zone_locked_mem; 606 extern rctl_hndl_t rc_zone_max_swap; 607 608 #endif /* _KERNEL */ 609 610 #ifdef __cplusplus 611 } 612 #endif 613 614 #endif /* _SYS_ZONE_H */ 615