xref: /illumos-gate/usr/src/uts/common/sys/zone.h (revision 2dc692e0)
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 (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
23  * Copyright 2013, Joyent, Inc. All rights reserved.
24  * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
25  * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
26  */
27 
28 #ifndef _SYS_ZONE_H
29 #define	_SYS_ZONE_H
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 #include <sys/ksynch.h>
42 #include <sys/socket_impl.h>
43 #include <netinet/in.h>
44 
45 #ifdef	__cplusplus
46 extern "C" {
47 #endif
48 
49 /*
50  * NOTE
51  *
52  * The contents of this file are private to the implementation of
53  * Solaris and are subject to change at any time without notice.
54  * Applications and drivers using these interfaces may fail to
55  * run on future releases.
56  */
57 
58 /* Available both in kernel and for user space */
59 
60 /* zone id restrictions and special ids */
61 #define	MAX_ZONEID	9999
62 #define	MIN_USERZONEID	1	/* lowest user-creatable zone ID */
63 #define	MIN_ZONEID	0	/* minimum zone ID on system */
64 #define	GLOBAL_ZONEID	0
65 #define	ZONEID_WIDTH	4	/* for printf */
66 
67 /*
68  * Special zoneid_t token to refer to all zones.
69  */
70 #define	ALL_ZONES	(-1)
71 
72 /* system call subcodes */
73 #define	ZONE_CREATE		0
74 #define	ZONE_DESTROY		1
75 #define	ZONE_GETATTR		2
76 #define	ZONE_ENTER		3
77 #define	ZONE_LIST		4
78 #define	ZONE_SHUTDOWN		5
79 #define	ZONE_LOOKUP		6
80 #define	ZONE_BOOT		7
81 #define	ZONE_VERSION		8
82 #define	ZONE_SETATTR		9
83 #define	ZONE_ADD_DATALINK	10
84 #define	ZONE_DEL_DATALINK	11
85 #define	ZONE_CHECK_DATALINK	12
86 #define	ZONE_LIST_DATALINK	13
87 
88 /* zone attributes */
89 #define	ZONE_ATTR_ROOT		1
90 #define	ZONE_ATTR_NAME		2
91 #define	ZONE_ATTR_STATUS	3
92 #define	ZONE_ATTR_PRIVSET	4
93 #define	ZONE_ATTR_UNIQID	5
94 #define	ZONE_ATTR_POOLID	6
95 #define	ZONE_ATTR_INITPID	7
96 #define	ZONE_ATTR_SLBL		8
97 #define	ZONE_ATTR_INITNAME	9
98 #define	ZONE_ATTR_BOOTARGS	10
99 #define	ZONE_ATTR_BRAND		11
100 #define	ZONE_ATTR_PHYS_MCAP	12
101 #define	ZONE_ATTR_SCHED_CLASS	13
102 #define	ZONE_ATTR_FLAGS		14
103 #define	ZONE_ATTR_HOSTID	15
104 #define	ZONE_ATTR_FS_ALLOWED	16
105 #define	ZONE_ATTR_NETWORK	17
106 
107 /* Start of the brand-specific attribute namespace */
108 #define	ZONE_ATTR_BRAND_ATTRS	32768
109 
110 #define	ZONE_FS_ALLOWED_MAX	1024
111 
112 #define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
113 #define	ZONE_EVENT_STATUS_CLASS	"status"
114 #define	ZONE_EVENT_STATUS_SUBCLASS	"change"
115 
116 #define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
117 #define	ZONE_EVENT_INITIALIZED		"initialized"
118 #define	ZONE_EVENT_READY		"ready"
119 #define	ZONE_EVENT_RUNNING		"running"
120 #define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
121 
122 #define	ZONE_CB_NAME		"zonename"
123 #define	ZONE_CB_NEWSTATE	"newstate"
124 #define	ZONE_CB_OLDSTATE	"oldstate"
125 #define	ZONE_CB_TIMESTAMP	"when"
126 #define	ZONE_CB_ZONEID		"zoneid"
127 
128 /*
129  * Exit values that may be returned by scripts or programs invoked by various
130  * zone commands.
131  *
132  * These are defined as:
133  *
134  *	ZONE_SUBPROC_OK
135  *	===============
136  *	The subprocess completed successfully.
137  *
138  *	ZONE_SUBPROC_USAGE
139  *	==================
140  *	The subprocess failed with a usage message, or a usage message should
141  *	be output in its behalf.
142  *
143  *	ZONE_SUBPROC_NOTCOMPLETE
144  *	========================
145  *	The subprocess did not complete, but the actions performed by the
146  *	subprocess require no recovery actions by the user.
147  *
148  *	For example, if the subprocess were called by "zoneadm install," the
149  *	installation of the zone did not succeed but the user need not perform
150  *	a "zoneadm uninstall" before attempting another install.
151  *
152  *	ZONE_SUBPROC_FATAL
153  *	==================
154  *	The subprocess failed in a fatal manner, usually one that will require
155  *	some type of recovery action by the user.
156  *
157  *	For example, if the subprocess were called by "zoneadm install," the
158  *	installation of the zone did not succeed and the user will need to
159  *	perform a "zoneadm uninstall" before another install attempt is
160  *	possible.
161  *
162  *	The non-success exit values are large to avoid accidental collision
163  *	with values used internally by some commands (e.g. "Z_ERR" and
164  *	"Z_USAGE" as used by zoneadm.)
165  */
166 #define	ZONE_SUBPROC_OK			0
167 #define	ZONE_SUBPROC_USAGE		253
168 #define	ZONE_SUBPROC_NOTCOMPLETE	254
169 #define	ZONE_SUBPROC_FATAL		255
170 
171 #ifdef _SYSCALL32
172 typedef struct {
173 	caddr32_t zone_name;
174 	caddr32_t zone_root;
175 	caddr32_t zone_privs;
176 	size32_t zone_privssz;
177 	caddr32_t rctlbuf;
178 	size32_t rctlbufsz;
179 	caddr32_t extended_error;
180 	caddr32_t zfsbuf;
181 	size32_t  zfsbufsz;
182 	int match;			/* match level */
183 	uint32_t doi;			/* DOI for label */
184 	caddr32_t label;		/* label associated with zone */
185 	int flags;
186 } zone_def32;
187 #endif
188 typedef struct {
189 	const char *zone_name;
190 	const char *zone_root;
191 	const struct priv_set *zone_privs;
192 	size_t zone_privssz;
193 	const char *rctlbuf;
194 	size_t rctlbufsz;
195 	int *extended_error;
196 	const char *zfsbuf;
197 	size_t zfsbufsz;
198 	int match;			/* match level */
199 	uint32_t doi;			/* DOI for label */
200 	const bslabel_t *label;		/* label associated with zone */
201 	int flags;
202 } zone_def;
203 
204 /* extended error information */
205 #define	ZE_UNKNOWN	0	/* No extended error info */
206 #define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
207 #define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
208 #define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
209 
210 /*
211  * zone_status values
212  *
213  * You must modify zone_status_names in mdb(1M)'s genunix module
214  * (genunix/zone.c) when you modify this enum.
215  */
216 typedef enum {
217 	ZONE_IS_UNINITIALIZED = 0,
218 	ZONE_IS_INITIALIZED,
219 	ZONE_IS_READY,
220 	ZONE_IS_BOOTING,
221 	ZONE_IS_RUNNING,
222 	ZONE_IS_SHUTTING_DOWN,
223 	ZONE_IS_EMPTY,
224 	ZONE_IS_DOWN,
225 	ZONE_IS_DYING,
226 	ZONE_IS_DEAD
227 } zone_status_t;
228 #define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
229 #define	ZONE_MAX_STATE		ZONE_IS_DEAD
230 
231 /*
232  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
233  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
234  */
235 typedef enum zone_cmd {
236 	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
237 	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
238 } zone_cmd_t;
239 
240 /*
241  * The structure of a request to zoneadmd.
242  */
243 typedef struct zone_cmd_arg {
244 	uint64_t	uniqid;		/* unique "generation number" */
245 	zone_cmd_t	cmd;		/* requested action */
246 	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
247 	char locale[MAXPATHLEN];	/* locale in which to render messages */
248 	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
249 } zone_cmd_arg_t;
250 
251 /*
252  * Structure of zoneadmd's response to a request.  A NULL return value means
253  * the caller should attempt to restart zoneadmd and retry.
254  */
255 typedef struct zone_cmd_rval {
256 	int rval;			/* return value of request */
257 	char errbuf[1];	/* variable-sized buffer containing error messages */
258 } zone_cmd_rval_t;
259 
260 /*
261  * The zone support infrastructure uses the zone name as a component
262  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
263  * in length, so ZONENAME_MAX is limited by that.
264  */
265 #define	ZONENAME_MAX		64
266 
267 #define	GLOBAL_ZONENAME		"global"
268 
269 /*
270  * Extended Regular expression (see regex(5)) which matches all valid zone
271  * names.
272  */
273 #define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
274 
275 /*
276  * Where the zones support infrastructure places temporary files.
277  */
278 #define	ZONES_TMPDIR		"/var/run/zones"
279 
280 /*
281  * The path to the door used by clients to communicate with zoneadmd.
282  */
283 #define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
284 
285 
286 /* zone_flags */
287 /*
288  * Threads that read or write the following flag must hold zone_lock.
289  */
290 #define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
291 
292 /*
293  * The following threads are set when the zone is created and never changed.
294  * Threads that test for these flags don't have to hold zone_lock.
295  */
296 #define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
297 #define	ZF_IS_SCRATCH		0x4	/* scratch zone */
298 #define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
299 
300 
301 /* zone_create flags */
302 #define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
303 
304 /* zone network properties */
305 #define	ZONE_NETWORK_ADDRESS	1
306 #define	ZONE_NETWORK_DEFROUTER	2
307 
308 #define	ZONE_NET_ADDRNAME	"address"
309 #define	ZONE_NET_RTRNAME	"route"
310 
311 typedef struct zone_net_data {
312 	int zn_type;
313 	int zn_len;
314 	datalink_id_t zn_linkid;
315 	uint8_t zn_val[1];
316 } zone_net_data_t;
317 
318 
319 #ifdef _KERNEL
320 
321 /*
322  * We need to protect the definition of 'list_t' from userland applications and
323  * libraries which may be defining ther own versions.
324  */
325 #include <sys/list.h>
326 #include <sys/loadavg.h>
327 
328 #define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
329 
330 struct pool;
331 struct brand;
332 
333 /*
334  * Each of these constants identifies a kernel subsystem that acquires and
335  * releases zone references.  Each subsystem that invokes
336  * zone_hold_ref() and zone_rele_ref() should specify the
337  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
338  * help users and developers quickly identify subsystems that stall zone
339  * shutdowns indefinitely.
340  *
341  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
342  * when you modify this enumeration.
343  */
344 typedef enum zone_ref_subsys {
345 	ZONE_REF_NFS,			/* NFS */
346 	ZONE_REF_NFSV4,			/* NFSv4 */
347 	ZONE_REF_SMBFS,			/* SMBFS */
348 	ZONE_REF_MNTFS,			/* MNTFS */
349 	ZONE_REF_LOFI,			/* LOFI devices */
350 	ZONE_REF_VFS,			/* VFS infrastructure */
351 	ZONE_REF_IPC,			/* IPC infrastructure */
352 	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
353 } zone_ref_subsys_t;
354 
355 /*
356  * zone_ref represents a general-purpose references to a zone.  Each zone's
357  * references are linked into the zone's zone_t::zone_ref_list.  This allows
358  * debuggers to walk zones' references.
359  */
360 typedef struct zone_ref {
361 	struct zone	*zref_zone; /* the zone to which the reference refers */
362 	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
363 } zone_ref_t;
364 
365 /*
366  * Structure to record list of ZFS datasets exported to a zone.
367  */
368 typedef struct zone_dataset {
369 	char		*zd_dataset;
370 	list_node_t	zd_linkage;
371 } zone_dataset_t;
372 
373 /*
374  * structure for zone kstats
375  */
376 typedef struct zone_kstat {
377 	kstat_named_t zk_zonename;
378 	kstat_named_t zk_usage;
379 	kstat_named_t zk_value;
380 } zone_kstat_t;
381 
382 struct cpucap;
383 
384 typedef struct {
385 	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
386 	kstat_named_t	zm_utime;
387 	kstat_named_t	zm_stime;
388 	kstat_named_t	zm_wtime;
389 	kstat_named_t	zm_avenrun1;
390 	kstat_named_t	zm_avenrun5;
391 	kstat_named_t	zm_avenrun15;
392 	kstat_named_t	zm_run_ticks;
393 	kstat_named_t	zm_run_wait;
394 	kstat_named_t	zm_fss_shr_pct;
395 	kstat_named_t	zm_fss_pri_hi;
396 	kstat_named_t	zm_fss_pri_avg;
397 	kstat_named_t	zm_ffcap;
398 	kstat_named_t	zm_ffnoproc;
399 	kstat_named_t	zm_ffnomem;
400 	kstat_named_t	zm_ffmisc;
401 } zone_misc_kstat_t;
402 
403 typedef struct zone {
404 	/*
405 	 * zone_name is never modified once set.
406 	 */
407 	char		*zone_name;	/* zone's configuration name */
408 	/*
409 	 * zone_nodename and zone_domain are never freed once allocated.
410 	 */
411 	char		*zone_nodename;	/* utsname.nodename equivalent */
412 	char		*zone_domain;	/* srpc_domain equivalent */
413 	/*
414 	 * zone_hostid is used for per-zone hostid emulation.
415 	 * Currently it isn't modified after it's set (so no locks protect
416 	 * accesses), but that might have to change when we allow
417 	 * administrators to change running zones' properties.
418 	 *
419 	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
420 	 * that zone_get_hostid() will function correctly.
421 	 */
422 	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
423 					/* if not emulated */
424 	/*
425 	 * zone_lock protects the following fields of a zone_t:
426 	 * 	zone_ref
427 	 * 	zone_cred_ref
428 	 * 	zone_subsys_ref
429 	 * 	zone_ref_list
430 	 * 	zone_ntasks
431 	 * 	zone_flags
432 	 * 	zone_zsd
433 	 *	zone_pfexecd
434 	 */
435 	kmutex_t	zone_lock;
436 	/*
437 	 * zone_linkage is the zone's linkage into the active or
438 	 * death-row list.  The field is protected by zonehash_lock.
439 	 */
440 	list_node_t	zone_linkage;
441 	zoneid_t	zone_id;	/* ID of zone */
442 	uint_t		zone_ref;	/* count of zone_hold()s on zone */
443 	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
444 	/*
445 	 * Fixed-sized array of subsystem-specific reference counts
446 	 * The sum of all of the counts must be less than or equal to zone_ref.
447 	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
448 	 * constants.
449 	 */
450 	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
451 	list_t		zone_ref_list;	/* list of zone_ref_t structs */
452 	/*
453 	 * zone_rootvp and zone_rootpath can never be modified once set.
454 	 */
455 	struct vnode	*zone_rootvp;	/* zone's root vnode */
456 	char		*zone_rootpath;	/* Path to zone's root + '/' */
457 	ushort_t	zone_flags;	/* misc flags */
458 	zone_status_t	zone_status;	/* protected by zone_status_lock */
459 	uint_t		zone_ntasks;	/* number of tasks executing in zone */
460 	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
461 					/* counters in projects and tasks */
462 					/* that are within the zone */
463 	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
464 	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
465 	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
466 	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
467 
468 	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
469 	uint32_t	zone_shares;	/* FSS shares allocated to zone */
470 	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
471 	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
472 					/* kpd_locked_mem for all */
473 					/* projects in zone. */
474 					/* Also protects zone_max_swap */
475 					/* grab after p_lock, before rcs_lock */
476 	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
477 						/* zone */
478 	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
479 						/* limit.  Protected by */
480 						/* zone_rctls->rcs_lock */
481 	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
482 	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
483 						/* Protected by */
484 						/* zone_rctls->rcs_lock */
485 	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
486 	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
487 	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
488 						/* Protected by */
489 						/* zone_rctls->rcs_lock */
490 	list_t		zone_zsd;	/* list of Zone-Specific Data values */
491 	kcondvar_t	zone_cv;	/* used to signal state changes */
492 	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
493 	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
494 	char		*zone_initname;	/* fs path to 'init' */
495 	int		zone_boot_err;  /* for zone_boot() if boot fails */
496 	char		*zone_bootargs;	/* arguments passed via zone_boot() */
497 	uint64_t	zone_phys_mcap;	/* physical memory cap */
498 	/*
499 	 * zone_kthreads is protected by zone_status_lock.
500 	 */
501 	kthread_t	*zone_kthreads;	/* kernel threads in zone */
502 	struct priv_set	*zone_privset;	/* limit set for zone */
503 	/*
504 	 * zone_vfslist is protected by vfs_list_lock().
505 	 */
506 	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
507 	uint64_t	zone_uniqid;	/* unique zone generation number */
508 	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
509 	/*
510 	 * zone_pool is protected by pool_lock().
511 	 */
512 	struct pool	*zone_pool;	/* pool the zone is bound to */
513 	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
514 	/* zone_psetid is protected by cpu_lock */
515 	psetid_t	zone_psetid;	/* pset the zone is bound to */
516 
517 	time_t		zone_boot_time;	/* Similar to boot_time */
518 
519 	/*
520 	 * The following two can be read without holding any locks.  They are
521 	 * updated under cpu_lock.
522 	 */
523 	int		zone_ncpus;  /* zone's idea of ncpus */
524 	int		zone_ncpus_online; /* zone's idea of ncpus_online */
525 	/*
526 	 * List of ZFS datasets exported to this zone.
527 	 */
528 	list_t		zone_datasets;	/* list of datasets */
529 
530 	ts_label_t	*zone_slabel;	/* zone sensitivity label */
531 	int		zone_match;	/* require label match for packets */
532 	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
533 
534 	boolean_t	zone_restart_init;	/* Restart init if it dies? */
535 	struct brand	*zone_brand;		/* zone's brand */
536 	void 		*zone_brand_data;	/* store brand specific data */
537 	id_t		zone_defaultcid;	/* dflt scheduling class id */
538 	kstat_t		*zone_swapresv_kstat;
539 	kstat_t		*zone_lockedmem_kstat;
540 	/*
541 	 * zone_dl_list is protected by zone_lock
542 	 */
543 	list_t		zone_dl_list;
544 	netstack_t	*zone_netstack;
545 	struct cpucap	*zone_cpucap;	/* CPU caps data */
546 	/*
547 	 * Solaris Auditing per-zone audit context
548 	 */
549 	struct au_kcontext	*zone_audit_kctxt;
550 	/*
551 	 * For private use by mntfs.
552 	 */
553 	struct mntelem	*zone_mntfs_db;
554 	krwlock_t	zone_mntfs_db_lock;
555 
556 	struct klpd_reg		*zone_pfexecd;
557 
558 	char		*zone_fs_allowed;
559 	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
560 	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
561 						/* zone_rctls->rcs_lock */
562 	kstat_t		*zone_nprocs_kstat;
563 
564 	/*
565 	 * Misc. kstats and counters for zone cpu-usage aggregation.
566 	 * The zone_Xtime values are the sum of the micro-state accounting
567 	 * values for all threads that are running or have run in the zone.
568 	 * This is tracked in msacct.c as threads change state.
569 	 * The zone_stime is the sum of the LMS_SYSTEM times.
570 	 * The zone_utime is the sum of the LMS_USER times.
571 	 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
572 	 * As with per-thread micro-state accounting values, these values are
573 	 * not scaled to nanosecs.  The scaling is done by the
574 	 * zone_misc_kstat_update function when kstats are requested.
575 	 */
576 	kmutex_t	zone_misc_lock;		/* protects misc statistics */
577 	kstat_t		*zone_misc_ksp;
578 	zone_misc_kstat_t *zone_misc_stats;
579 	uint64_t	zone_stime;		/* total system time */
580 	uint64_t	zone_utime;		/* total user time */
581 	uint64_t	zone_wtime;		/* total time waiting in runq */
582 	/* fork-fail kstat tracking */
583 	uint32_t	zone_ffcap;		/* hit an rctl cap */
584 	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
585 	uint32_t	zone_ffnomem;		/* as_dup/memory error */
586 	uint32_t	zone_ffmisc;		/* misc. other error */
587 
588 	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
589 	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
590 	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
591 
592 	/*
593 	 * FSS stats updated once per second by fss_decay_usage.
594 	 */
595 	uint32_t	zone_fss_gen;		/* FSS generation cntr */
596 	uint64_t	zone_run_ticks;		/* tot # of ticks running */
597 
598 	/*
599 	 * DTrace-private per-zone state
600 	 */
601 	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
602 } zone_t;
603 
604 /*
605  * Special value of zone_psetid to indicate that pools are disabled.
606  */
607 #define	ZONE_PS_INVAL	PS_MYID
608 
609 
610 extern zone_t zone0;
611 extern zone_t *global_zone;
612 extern uint_t maxzones;
613 extern rctl_hndl_t rc_zone_nlwps;
614 extern rctl_hndl_t rc_zone_nprocs;
615 
616 extern long zone(int, void *, void *, void *, void *);
617 extern void zone_zsd_init(void);
618 extern void zone_init(void);
619 extern void zone_hold(zone_t *);
620 extern void zone_rele(zone_t *);
621 extern void zone_init_ref(zone_ref_t *);
622 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
623 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
624 extern void zone_cred_hold(zone_t *);
625 extern void zone_cred_rele(zone_t *);
626 extern void zone_task_hold(zone_t *);
627 extern void zone_task_rele(zone_t *);
628 extern zone_t *zone_find_by_id(zoneid_t);
629 extern zone_t *zone_find_by_label(const ts_label_t *);
630 extern zone_t *zone_find_by_name(char *);
631 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
632 extern zone_t *zone_find_by_path(const char *);
633 extern zoneid_t getzoneid(void);
634 extern zone_t *zone_find_by_id_nolock(zoneid_t);
635 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
636 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
637 extern void zone_loadavg_update();
638 
639 /*
640  * Zone-specific data (ZSD) APIs
641  */
642 /*
643  * The following is what code should be initializing its zone_key_t to if it
644  * calls zone_getspecific() without necessarily knowing that zone_key_create()
645  * has been called on the key.
646  */
647 #define	ZONE_KEY_UNINITIALIZED	0
648 
649 typedef uint_t zone_key_t;
650 
651 extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
652     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
653 extern int 	zone_key_delete(zone_key_t);
654 extern void	*zone_getspecific(zone_key_t, zone_t *);
655 extern int	zone_setspecific(zone_key_t, zone_t *, const void *);
656 
657 /*
658  * The definition of a zsd_entry is truly private to zone.c and is only
659  * placed here so it can be shared with mdb.
660  *
661  * State maintained for each zone times each registered key, which tracks
662  * the state of the create, shutdown and destroy callbacks.
663  *
664  * zsd_flags is used to keep track of pending actions to avoid holding locks
665  * when calling the create/shutdown/destroy callbacks, since doing so
666  * could lead to deadlocks.
667  */
668 struct zsd_entry {
669 	zone_key_t		zsd_key;	/* Key used to lookup value */
670 	void			*zsd_data;	/* Caller-managed value */
671 	/*
672 	 * Callbacks to be executed when a zone is created, shutdown, and
673 	 * destroyed, respectively.
674 	 */
675 	void			*(*zsd_create)(zoneid_t);
676 	void			(*zsd_shutdown)(zoneid_t, void *);
677 	void			(*zsd_destroy)(zoneid_t, void *);
678 	list_node_t		zsd_linkage;
679 	uint16_t 		zsd_flags;	/* See below */
680 	kcondvar_t		zsd_cv;
681 };
682 
683 /*
684  * zsd_flags
685  */
686 #define	ZSD_CREATE_NEEDED	0x0001
687 #define	ZSD_CREATE_INPROGRESS	0x0002
688 #define	ZSD_CREATE_COMPLETED	0x0004
689 #define	ZSD_SHUTDOWN_NEEDED	0x0010
690 #define	ZSD_SHUTDOWN_INPROGRESS	0x0020
691 #define	ZSD_SHUTDOWN_COMPLETED	0x0040
692 #define	ZSD_DESTROY_NEEDED	0x0100
693 #define	ZSD_DESTROY_INPROGRESS	0x0200
694 #define	ZSD_DESTROY_COMPLETED	0x0400
695 
696 #define	ZSD_CREATE_ALL	\
697 	(ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
698 #define	ZSD_SHUTDOWN_ALL	\
699 	(ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
700 #define	ZSD_DESTROY_ALL	\
701 	(ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
702 
703 #define	ZSD_ALL_INPROGRESS \
704 	(ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
705 
706 /*
707  * Macros to help with zone visibility restrictions.
708  */
709 
710 /*
711  * Is process in the global zone?
712  */
713 #define	INGLOBALZONE(p) \
714 	((p)->p_zone == global_zone)
715 
716 /*
717  * Can process view objects in given zone?
718  */
719 #define	HASZONEACCESS(p, zoneid) \
720 	((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
721 
722 /*
723  * Convenience macro to see if a resolved path is visible from within a
724  * given zone.
725  *
726  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
727  * two strings must be equal.  Since the rootpathlen has a trailing '/',
728  * we want to skip everything in the path up to (but not including) the
729  * trailing '/'.
730  */
731 #define	ZONE_PATH_VISIBLE(path, zone) \
732 	(strncmp((path), (zone)->zone_rootpath,		\
733 	    (zone)->zone_rootpathlen - 1) == 0)
734 
735 /*
736  * Convenience macro to go from the global view of a path to that seen
737  * from within said zone.  It is the responsibility of the caller to
738  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
739  * in fact visible from within the zone.
740  */
741 #define	ZONE_PATH_TRANSLATE(path, zone)	\
742 	(ASSERT(ZONE_PATH_VISIBLE(path, zone)),	\
743 	(path) + (zone)->zone_rootpathlen - 2)
744 
745 /*
746  * Special processes visible in all zones.
747  */
748 #define	ZONE_SPECIALPID(x)	 ((x) == 0 || (x) == 1)
749 
750 /*
751  * Zone-safe version of thread_create() to be used when the caller wants to
752  * create a kernel thread to run within the current zone's context.
753  */
754 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
755     pri_t);
756 extern void zthread_exit(void);
757 
758 /*
759  * Functions for an external observer to register interest in a zone's status
760  * change.  Observers will be woken up when the zone status equals the status
761  * argument passed in (in the case of zone_status_timedwait, the function may
762  * also return because of a timeout; zone_status_wait_sig may return early due
763  * to a signal being delivered; zone_status_timedwait_sig may return for any of
764  * the above reasons).
765  *
766  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
767  * cv_wait_sig() respectively.
768  */
769 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
770 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
771 extern void zone_status_wait(zone_t *, zone_status_t);
772 extern int zone_status_wait_sig(zone_t *, zone_status_t);
773 
774 /*
775  * Get the status  of the zone (at the time it was called).  The state may
776  * have progressed by the time it is returned.
777  */
778 extern zone_status_t zone_status_get(zone_t *);
779 
780 /*
781  * Safely get the hostid of the specified zone (defaults to machine's hostid
782  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
783  * the global zone's (i.e., physical system's) hostid.
784  */
785 extern uint32_t zone_get_hostid(zone_t *);
786 
787 /*
788  * Get the "kcred" credentials corresponding to the given zone.
789  */
790 extern struct cred *zone_get_kcred(zoneid_t);
791 
792 /*
793  * Get/set the pool the zone is currently bound to.
794  */
795 extern struct pool *zone_pool_get(zone_t *);
796 extern void zone_pool_set(zone_t *, struct pool *);
797 
798 /*
799  * Get/set the pset the zone is currently using.
800  */
801 extern psetid_t zone_pset_get(zone_t *);
802 extern void zone_pset_set(zone_t *, psetid_t);
803 
804 /*
805  * Get the number of cpus/online-cpus visible from the given zone.
806  */
807 extern int zone_ncpus_get(zone_t *);
808 extern int zone_ncpus_online_get(zone_t *);
809 
810 /*
811  * Returns true if the named pool/dataset is visible in the current zone.
812  */
813 extern int zone_dataset_visible(const char *, int *);
814 
815 /*
816  * zone version of kadmin()
817  */
818 extern int zone_kadmin(int, int, const char *, cred_t *);
819 extern void zone_shutdown_global(void);
820 
821 extern void mount_in_progress(void);
822 extern void mount_completed(void);
823 
824 extern int zone_walk(int (*)(zone_t *, void *), void *);
825 
826 extern rctl_hndl_t rc_zone_locked_mem;
827 extern rctl_hndl_t rc_zone_max_swap;
828 extern rctl_hndl_t rc_zone_max_lofi;
829 
830 #endif	/* _KERNEL */
831 
832 #ifdef	__cplusplus
833 }
834 #endif
835 
836 #endif	/* _SYS_ZONE_H */
837