xref: /illumos-gate/usr/src/uts/common/sys/zone.h (revision 00fc50d191cf90458c41a077d59fe2f81223f00a)
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 2018 Joyent, Inc.
24  * Copyright 2019 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/param.h>
33 #include <sys/tsol/label.h>
34 #include <sys/uadmin.h>
35 #include <netinet/in.h>
36 
37 #ifdef _KERNEL
38 /*
39  * Many includes are kernel-only to reduce namespace pollution of
40  * userland applications.
41  */
42 #include <sys/mutex.h>
43 #include <sys/rctl.h>
44 #include <sys/ipc_rctl.h>
45 #include <sys/pset.h>
46 #include <sys/cred.h>
47 #include <sys/netstack.h>
48 #include <sys/ksynch.h>
49 #include <sys/socket_impl.h>
50 #include <sys/secflags.h>
51 #include <sys/cpu_uarray.h>
52 #include <sys/list.h>
53 #include <sys/loadavg.h>
54 #endif	/* _KERNEL */
55 
56 #ifdef	__cplusplus
57 extern "C" {
58 #endif
59 
60 /*
61  * NOTE
62  *
63  * The contents of this file are private to the implementation of
64  * Solaris and are subject to change at any time without notice.
65  * Applications and drivers using these interfaces may fail to
66  * run on future releases.
67  */
68 
69 /* Available both in kernel and for user space */
70 
71 /* zone id restrictions and special ids */
72 #define	MAX_ZONEID	9999
73 #define	MIN_USERZONEID	1	/* lowest user-creatable zone ID */
74 #define	MIN_ZONEID	0	/* minimum zone ID on system */
75 #define	GLOBAL_ZONEID	0
76 #define	ZONEID_WIDTH	4	/* for printf */
77 
78 /*
79  * Special zoneid_t token to refer to all zones.
80  */
81 #define	ALL_ZONES	(-1)
82 
83 /* system call subcodes */
84 #define	ZONE_CREATE		0
85 #define	ZONE_DESTROY		1
86 #define	ZONE_GETATTR		2
87 #define	ZONE_ENTER		3
88 #define	ZONE_LIST		4
89 #define	ZONE_SHUTDOWN		5
90 #define	ZONE_LOOKUP		6
91 #define	ZONE_BOOT		7
92 #define	ZONE_VERSION		8
93 #define	ZONE_SETATTR		9
94 #define	ZONE_ADD_DATALINK	10
95 #define	ZONE_DEL_DATALINK	11
96 #define	ZONE_CHECK_DATALINK	12
97 #define	ZONE_LIST_DATALINK	13
98 
99 /* zone attributes */
100 #define	ZONE_ATTR_ROOT		1
101 #define	ZONE_ATTR_NAME		2
102 #define	ZONE_ATTR_STATUS	3
103 #define	ZONE_ATTR_PRIVSET	4
104 #define	ZONE_ATTR_UNIQID	5
105 #define	ZONE_ATTR_POOLID	6
106 #define	ZONE_ATTR_INITPID	7
107 #define	ZONE_ATTR_SLBL		8
108 #define	ZONE_ATTR_INITNAME	9
109 #define	ZONE_ATTR_BOOTARGS	10
110 #define	ZONE_ATTR_BRAND		11
111 #define	ZONE_ATTR_PHYS_MCAP	12
112 #define	ZONE_ATTR_SCHED_CLASS	13
113 #define	ZONE_ATTR_FLAGS		14
114 #define	ZONE_ATTR_HOSTID	15
115 #define	ZONE_ATTR_FS_ALLOWED	16
116 #define	ZONE_ATTR_NETWORK	17
117 #define	ZONE_ATTR_INITNORESTART	20
118 #define	ZONE_ATTR_SECFLAGS	21
119 
120 /* Start of the brand-specific attribute namespace */
121 #define	ZONE_ATTR_BRAND_ATTRS	32768
122 
123 #define	ZONE_FS_ALLOWED_MAX	1024
124 
125 #define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
126 #define	ZONE_EVENT_STATUS_CLASS	"status"
127 #define	ZONE_EVENT_STATUS_SUBCLASS	"change"
128 
129 #define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
130 #define	ZONE_EVENT_INITIALIZED		"initialized"
131 #define	ZONE_EVENT_READY		"ready"
132 #define	ZONE_EVENT_RUNNING		"running"
133 #define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
134 
135 #define	ZONE_CB_NAME		"zonename"
136 #define	ZONE_CB_NEWSTATE	"newstate"
137 #define	ZONE_CB_OLDSTATE	"oldstate"
138 #define	ZONE_CB_TIMESTAMP	"when"
139 #define	ZONE_CB_ZONEID		"zoneid"
140 
141 /*
142  * Exit values that may be returned by scripts or programs invoked by various
143  * zone commands.
144  *
145  * These are defined as:
146  *
147  *	ZONE_SUBPROC_OK
148  *	===============
149  *	The subprocess completed successfully.
150  *
151  *	ZONE_SUBPROC_USAGE
152  *	==================
153  *	The subprocess failed with a usage message, or a usage message should
154  *	be output in its behalf.
155  *
156  *	ZONE_SUBPROC_NOTCOMPLETE
157  *	========================
158  *	The subprocess did not complete, but the actions performed by the
159  *	subprocess require no recovery actions by the user.
160  *
161  *	For example, if the subprocess were called by "zoneadm install," the
162  *	installation of the zone did not succeed but the user need not perform
163  *	a "zoneadm uninstall" before attempting another install.
164  *
165  *	ZONE_SUBPROC_FATAL
166  *	==================
167  *	The subprocess failed in a fatal manner, usually one that will require
168  *	some type of recovery action by the user.
169  *
170  *	For example, if the subprocess were called by "zoneadm install," the
171  *	installation of the zone did not succeed and the user will need to
172  *	perform a "zoneadm uninstall" before another install attempt is
173  *	possible.
174  *
175  *	The non-success exit values are large to avoid accidental collision
176  *	with values used internally by some commands (e.g. "Z_ERR" and
177  *	"Z_USAGE" as used by zoneadm.)
178  */
179 #define	ZONE_SUBPROC_OK			0
180 #define	ZONE_SUBPROC_USAGE		253
181 #define	ZONE_SUBPROC_NOTCOMPLETE	254
182 #define	ZONE_SUBPROC_FATAL		255
183 
184 #ifdef _SYSCALL32
185 typedef struct {
186 	caddr32_t zone_name;
187 	caddr32_t zone_root;
188 	caddr32_t zone_privs;
189 	size32_t zone_privssz;
190 	caddr32_t rctlbuf;
191 	size32_t rctlbufsz;
192 	caddr32_t extended_error;
193 	caddr32_t zfsbuf;
194 	size32_t  zfsbufsz;
195 	int match;			/* match level */
196 	uint32_t doi;			/* DOI for label */
197 	caddr32_t label;		/* label associated with zone */
198 	int flags;
199 } zone_def32;
200 #endif
201 typedef struct {
202 	const char *zone_name;
203 	const char *zone_root;
204 	const struct priv_set *zone_privs;
205 	size_t zone_privssz;
206 	const char *rctlbuf;
207 	size_t rctlbufsz;
208 	int *extended_error;
209 	const char *zfsbuf;
210 	size_t zfsbufsz;
211 	int match;			/* match level */
212 	uint32_t doi;			/* DOI for label */
213 	const bslabel_t *label;		/* label associated with zone */
214 	int flags;
215 } zone_def;
216 
217 /* extended error information */
218 #define	ZE_UNKNOWN	0	/* No extended error info */
219 #define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
220 #define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
221 #define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
222 
223 /*
224  * zone_status values
225  *
226  * You must modify zone_status_names in mdb(1M)'s genunix module
227  * (genunix/zone.c) when you modify this enum.
228  */
229 typedef enum {
230 	ZONE_IS_UNINITIALIZED = 0,
231 	ZONE_IS_INITIALIZED,
232 	ZONE_IS_READY,
233 	ZONE_IS_BOOTING,
234 	ZONE_IS_RUNNING,
235 	ZONE_IS_SHUTTING_DOWN,
236 	ZONE_IS_EMPTY,
237 	ZONE_IS_DOWN,
238 	ZONE_IS_DYING,
239 	ZONE_IS_DEAD
240 } zone_status_t;
241 #define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
242 #define	ZONE_MAX_STATE		ZONE_IS_DEAD
243 
244 /*
245  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
246  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
247  */
248 typedef enum zone_cmd {
249 	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
250 	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
251 } zone_cmd_t;
252 
253 /*
254  * The structure of a request to zoneadmd.
255  */
256 typedef struct zone_cmd_arg {
257 	uint64_t	uniqid;		/* unique "generation number" */
258 	zone_cmd_t	cmd;		/* requested action */
259 	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
260 	char locale[MAXPATHLEN];	/* locale in which to render messages */
261 	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
262 } zone_cmd_arg_t;
263 
264 /*
265  * Structure of zoneadmd's response to a request.  A NULL return value means
266  * the caller should attempt to restart zoneadmd and retry.
267  */
268 typedef struct zone_cmd_rval {
269 	int rval;			/* return value of request */
270 	char errbuf[1];	/* variable-sized buffer containing error messages */
271 } zone_cmd_rval_t;
272 
273 /*
274  * The zone support infrastructure uses the zone name as a component
275  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
276  * in length, so ZONENAME_MAX is limited by that.
277  */
278 #define	ZONENAME_MAX		64
279 
280 #define	GLOBAL_ZONENAME		"global"
281 
282 /*
283  * Extended Regular expression (see regex(5)) which matches all valid zone
284  * names.
285  */
286 #define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
287 
288 /*
289  * Where the zones support infrastructure places temporary files.
290  */
291 #define	ZONES_TMPDIR		"/var/run/zones"
292 
293 /*
294  * The path to the door used by clients to communicate with zoneadmd.
295  */
296 #define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
297 
298 
299 /* zone_flags */
300 /*
301  * Threads that read or write the following flag must hold zone_lock.
302  */
303 #define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
304 
305 /*
306  * The following threads are set when the zone is created and never changed.
307  * Threads that test for these flags don't have to hold zone_lock.
308  */
309 #define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
310 #define	ZF_IS_SCRATCH		0x4	/* scratch zone */
311 #define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
312 
313 
314 /* zone_create flags */
315 #define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
316 
317 /* zone network properties */
318 #define	ZONE_NETWORK_ADDRESS	1
319 #define	ZONE_NETWORK_DEFROUTER	2
320 
321 #define	ZONE_NET_ADDRNAME	"address"
322 #define	ZONE_NET_RTRNAME	"route"
323 
324 typedef struct zone_net_data {
325 	int zn_type;
326 	int zn_len;
327 	datalink_id_t zn_linkid;
328 	uint8_t zn_val[1];
329 } zone_net_data_t;
330 
331 
332 #ifdef _KERNEL
333 
334 #define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
335 
336 /*
337  * Indexes into ->zone_ustate array, summing the micro state of all threads in a
338  * particular zone.
339  */
340 #define	ZONE_USTATE_STIME (0)
341 #define	ZONE_USTATE_UTIME (1)
342 #define	ZONE_USTATE_WTIME (2)
343 #define	ZONE_USTATE_MAX (3)
344 
345 struct pool;
346 struct brand;
347 
348 /*
349  * Each of these constants identifies a kernel subsystem that acquires and
350  * releases zone references.  Each subsystem that invokes
351  * zone_hold_ref() and zone_rele_ref() should specify the
352  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
353  * help users and developers quickly identify subsystems that stall zone
354  * shutdowns indefinitely.
355  *
356  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
357  * when you modify this enumeration.
358  */
359 typedef enum zone_ref_subsys {
360 	ZONE_REF_NFS,			/* NFS */
361 	ZONE_REF_NFSV4,			/* NFSv4 */
362 	ZONE_REF_SMBFS,			/* SMBFS */
363 	ZONE_REF_MNTFS,			/* MNTFS */
364 	ZONE_REF_LOFI,			/* LOFI devices */
365 	ZONE_REF_VFS,			/* VFS infrastructure */
366 	ZONE_REF_IPC,			/* IPC infrastructure */
367 	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
368 } zone_ref_subsys_t;
369 
370 /*
371  * zone_ref represents a general-purpose references to a zone.  Each zone's
372  * references are linked into the zone's zone_t::zone_ref_list.  This allows
373  * debuggers to walk zones' references.
374  */
375 typedef struct zone_ref {
376 	struct zone	*zref_zone; /* the zone to which the reference refers */
377 	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
378 } zone_ref_t;
379 
380 /*
381  * Structure to record list of ZFS datasets exported to a zone.
382  */
383 typedef struct zone_dataset {
384 	char		*zd_dataset;
385 	list_node_t	zd_linkage;
386 } zone_dataset_t;
387 
388 /*
389  * structure for zone kstats
390  */
391 typedef struct zone_kstat {
392 	kstat_named_t zk_zonename;
393 	kstat_named_t zk_usage;
394 	kstat_named_t zk_value;
395 } zone_kstat_t;
396 
397 struct cpucap;
398 
399 typedef struct {
400 	kstat_named_t	zm_zonename;
401 	kstat_named_t	zm_pgpgin;
402 	kstat_named_t	zm_anonpgin;
403 	kstat_named_t	zm_execpgin;
404 	kstat_named_t	zm_fspgin;
405 	kstat_named_t	zm_anon_alloc_fail;
406 } zone_mcap_kstat_t;
407 
408 typedef struct {
409 	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
410 	kstat_named_t	zm_utime;
411 	kstat_named_t	zm_stime;
412 	kstat_named_t	zm_wtime;
413 	kstat_named_t	zm_avenrun1;
414 	kstat_named_t	zm_avenrun5;
415 	kstat_named_t	zm_avenrun15;
416 	kstat_named_t	zm_ffcap;
417 	kstat_named_t	zm_ffnoproc;
418 	kstat_named_t	zm_ffnomem;
419 	kstat_named_t	zm_ffmisc;
420 	kstat_named_t	zm_nested_intp;
421 	kstat_named_t	zm_init_pid;
422 	kstat_named_t	zm_boot_time;
423 } zone_misc_kstat_t;
424 
425 typedef struct zone {
426 	/*
427 	 * zone_name is never modified once set.
428 	 */
429 	char		*zone_name;	/* zone's configuration name */
430 	/*
431 	 * zone_nodename and zone_domain are never freed once allocated.
432 	 */
433 	char		*zone_nodename;	/* utsname.nodename equivalent */
434 	char		*zone_domain;	/* srpc_domain equivalent */
435 	/*
436 	 * zone_hostid is used for per-zone hostid emulation.
437 	 * Currently it isn't modified after it's set (so no locks protect
438 	 * accesses), but that might have to change when we allow
439 	 * administrators to change running zones' properties.
440 	 *
441 	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
442 	 * that zone_get_hostid() will function correctly.
443 	 */
444 	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
445 					/* if not emulated */
446 	/*
447 	 * zone_lock protects the following fields of a zone_t:
448 	 *	zone_ref
449 	 *	zone_cred_ref
450 	 *	zone_subsys_ref
451 	 *	zone_ref_list
452 	 *	zone_ntasks
453 	 *	zone_flags
454 	 *	zone_zsd
455 	 *	zone_pfexecd
456 	 */
457 	kmutex_t	zone_lock;
458 	/*
459 	 * zone_linkage is the zone's linkage into the active or
460 	 * death-row list.  The field is protected by zonehash_lock.
461 	 */
462 	list_node_t	zone_linkage;
463 	zoneid_t	zone_id;	/* ID of zone */
464 	uint_t		zone_ref;	/* count of zone_hold()s on zone */
465 	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
466 	/*
467 	 * Fixed-sized array of subsystem-specific reference counts
468 	 * The sum of all of the counts must be less than or equal to zone_ref.
469 	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
470 	 * constants.
471 	 */
472 	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
473 	list_t		zone_ref_list;	/* list of zone_ref_t structs */
474 	/*
475 	 * zone_rootvp and zone_rootpath can never be modified once set.
476 	 */
477 	struct vnode	*zone_rootvp;	/* zone's root vnode */
478 	char		*zone_rootpath;	/* Path to zone's root + '/' */
479 	ushort_t	zone_flags;	/* misc flags */
480 	zone_status_t	zone_status;	/* protected by zone_status_lock */
481 	uint_t		zone_ntasks;	/* number of tasks executing in zone */
482 	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
483 					/* counters in projects and tasks */
484 					/* that are within the zone */
485 	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
486 	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
487 	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
488 	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
489 
490 	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
491 	uint32_t	zone_shares;	/* FSS shares allocated to zone */
492 	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
493 	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
494 					/* kpd_locked_mem for all */
495 					/* projects in zone. */
496 					/* Also protects zone_max_swap */
497 					/* grab after p_lock, before rcs_lock */
498 	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
499 						/* zone */
500 	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
501 						/* limit.  Protected by */
502 						/* zone_rctls->rcs_lock */
503 	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
504 	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
505 						/* Protected by */
506 						/* zone_rctls->rcs_lock */
507 	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
508 	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
509 	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
510 						/* Protected by */
511 						/* zone_rctls->rcs_lock */
512 	list_t		zone_zsd;	/* list of Zone-Specific Data values */
513 	kcondvar_t	zone_cv;	/* used to signal state changes */
514 	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
515 	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
516 	char		*zone_initname;	/* fs path to 'init' */
517 	int		zone_boot_err;  /* for zone_boot() if boot fails */
518 	char		*zone_bootargs;	/* arguments passed via zone_boot() */
519 	uint64_t	zone_phys_mcap;	/* physical memory cap */
520 	/*
521 	 * zone_kthreads is protected by zone_status_lock.
522 	 */
523 	kthread_t	*zone_kthreads;	/* kernel threads in zone */
524 	struct priv_set	*zone_privset;	/* limit set for zone */
525 	/*
526 	 * zone_vfslist is protected by vfs_list_lock().
527 	 */
528 	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
529 	uint64_t	zone_uniqid;	/* unique zone generation number */
530 	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
531 	/*
532 	 * zone_pool is protected by pool_lock().
533 	 */
534 	struct pool	*zone_pool;	/* pool the zone is bound to */
535 	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
536 	/* zone_psetid is protected by cpu_lock */
537 	psetid_t	zone_psetid;	/* pset the zone is bound to */
538 
539 	time_t		zone_boot_time;	/* Similar to boot_time */
540 
541 	/*
542 	 * The following two can be read without holding any locks.  They are
543 	 * updated under cpu_lock.
544 	 */
545 	int		zone_ncpus;  /* zone's idea of ncpus */
546 	int		zone_ncpus_online; /* zone's idea of ncpus_online */
547 	/*
548 	 * List of ZFS datasets exported to this zone.
549 	 */
550 	list_t		zone_datasets;	/* list of datasets */
551 
552 	ts_label_t	*zone_slabel;	/* zone sensitivity label */
553 	int		zone_match;	/* require label match for packets */
554 	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
555 
556 	boolean_t	zone_restart_init;	/* Restart init if it dies? */
557 	struct brand	*zone_brand;		/* zone's brand */
558 	void		*zone_brand_data;	/* store brand specific data */
559 	id_t		zone_defaultcid;	/* dflt scheduling class id */
560 	kstat_t		*zone_swapresv_kstat;
561 	kstat_t		*zone_lockedmem_kstat;
562 	/*
563 	 * zone_dl_list is protected by zone_lock
564 	 */
565 	list_t		zone_dl_list;
566 	netstack_t	*zone_netstack;
567 	struct cpucap	*zone_cpucap;	/* CPU caps data */
568 	/*
569 	 * Solaris Auditing per-zone audit context
570 	 */
571 	struct au_kcontext	*zone_audit_kctxt;
572 	/*
573 	 * For private use by mntfs.
574 	 */
575 	struct mntelem	*zone_mntfs_db;
576 	krwlock_t	zone_mntfs_db_lock;
577 
578 	struct klpd_reg		*zone_pfexecd;
579 
580 	char		*zone_fs_allowed;
581 	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
582 	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
583 						/* zone_rctls->rcs_lock */
584 	kstat_t		*zone_nprocs_kstat;
585 
586 	kmutex_t	zone_mcap_lock;	/* protects mcap statistics */
587 	kstat_t		*zone_mcap_ksp;
588 	zone_mcap_kstat_t *zone_mcap_stats;
589 	uint64_t	zone_pgpgin;		/* pages paged in */
590 	uint64_t	zone_anonpgin;		/* anon pages paged in */
591 	uint64_t	zone_execpgin;		/* exec pages paged in */
592 	uint64_t	zone_fspgin;		/* fs pages paged in */
593 	uint64_t	zone_anon_alloc_fail;	/* cnt of anon alloc fails */
594 
595 	psecflags_t	zone_secflags; /* default zone security-flags */
596 
597 	/*
598 	 * Misc. kstats and counters for zone cpu-usage aggregation.
599 	 */
600 	kmutex_t	zone_misc_lock;		/* protects misc statistics */
601 	kstat_t		*zone_misc_ksp;
602 	zone_misc_kstat_t *zone_misc_stats;
603 	/* Accumulated microstate for all threads in this zone. */
604 	cpu_uarray_t	*zone_ustate;
605 	/* fork-fail kstat tracking */
606 	uint32_t	zone_ffcap;		/* hit an rctl cap */
607 	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
608 	uint32_t	zone_ffnomem;		/* as_dup/memory error */
609 	uint32_t	zone_ffmisc;		/* misc. other error */
610 
611 	uint32_t	zone_nested_intp;	/* nested interp. kstat */
612 
613 	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
614 	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
615 	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
616 
617 	/*
618 	 * FSS stats updated once per second by fss_decay_usage.
619 	 */
620 	uint32_t	zone_fss_gen;		/* FSS generation cntr */
621 	uint64_t	zone_run_ticks;		/* tot # of ticks running */
622 
623 	/*
624 	 * DTrace-private per-zone state
625 	 */
626 	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
627 
628 	/*
629 	 * Synchronization primitives used to synchronize between mounts and
630 	 * zone creation/destruction.
631 	 */
632 	int		zone_mounts_in_progress;
633 	kcondvar_t	zone_mount_cv;
634 	kmutex_t	zone_mount_lock;
635 } zone_t;
636 
637 /*
638  * Special value of zone_psetid to indicate that pools are disabled.
639  */
640 #define	ZONE_PS_INVAL	PS_MYID
641 
642 
643 extern zone_t zone0;
644 extern zone_t *global_zone;
645 extern uint_t maxzones;
646 extern rctl_hndl_t rc_zone_nlwps;
647 extern rctl_hndl_t rc_zone_nprocs;
648 
649 extern long zone(int, void *, void *, void *, void *);
650 extern void zone_zsd_init(void);
651 extern void zone_init(void);
652 extern void zone_hold(zone_t *);
653 extern void zone_rele(zone_t *);
654 extern void zone_init_ref(zone_ref_t *);
655 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
656 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
657 extern void zone_cred_hold(zone_t *);
658 extern void zone_cred_rele(zone_t *);
659 extern void zone_task_hold(zone_t *);
660 extern void zone_task_rele(zone_t *);
661 extern zone_t *zone_find_by_id(zoneid_t);
662 extern zone_t *zone_find_by_label(const ts_label_t *);
663 extern zone_t *zone_find_by_name(char *);
664 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
665 extern zone_t *zone_find_by_path(const char *);
666 extern zoneid_t getzoneid(void);
667 extern zone_t *zone_find_by_id_nolock(zoneid_t);
668 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
669 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
670 extern void zone_loadavg_update();
671 
672 /*
673  * Zone-specific data (ZSD) APIs
674  */
675 /*
676  * The following is what code should be initializing its zone_key_t to if it
677  * calls zone_getspecific() without necessarily knowing that zone_key_create()
678  * has been called on the key.
679  */
680 #define	ZONE_KEY_UNINITIALIZED	0
681 
682 typedef uint_t zone_key_t;
683 
684 extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
685     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
686 extern int	zone_key_delete(zone_key_t);
687 extern void	*zone_getspecific(zone_key_t, zone_t *);
688 extern int	zone_setspecific(zone_key_t, zone_t *, const void *);
689 
690 /*
691  * The definition of a zsd_entry is truly private to zone.c and is only
692  * placed here so it can be shared with mdb.
693  *
694  * State maintained for each zone times each registered key, which tracks
695  * the state of the create, shutdown and destroy callbacks.
696  *
697  * zsd_flags is used to keep track of pending actions to avoid holding locks
698  * when calling the create/shutdown/destroy callbacks, since doing so
699  * could lead to deadlocks.
700  */
701 struct zsd_entry {
702 	zone_key_t		zsd_key;	/* Key used to lookup value */
703 	void			*zsd_data;	/* Caller-managed value */
704 	/*
705 	 * Callbacks to be executed when a zone is created, shutdown, and
706 	 * destroyed, respectively.
707 	 */
708 	void			*(*zsd_create)(zoneid_t);
709 	void			(*zsd_shutdown)(zoneid_t, void *);
710 	void			(*zsd_destroy)(zoneid_t, void *);
711 	list_node_t		zsd_linkage;
712 	uint16_t		zsd_flags;	/* See below */
713 	kcondvar_t		zsd_cv;
714 };
715 
716 /*
717  * zsd_flags
718  */
719 #define	ZSD_CREATE_NEEDED	0x0001
720 #define	ZSD_CREATE_INPROGRESS	0x0002
721 #define	ZSD_CREATE_COMPLETED	0x0004
722 #define	ZSD_SHUTDOWN_NEEDED	0x0010
723 #define	ZSD_SHUTDOWN_INPROGRESS	0x0020
724 #define	ZSD_SHUTDOWN_COMPLETED	0x0040
725 #define	ZSD_DESTROY_NEEDED	0x0100
726 #define	ZSD_DESTROY_INPROGRESS	0x0200
727 #define	ZSD_DESTROY_COMPLETED	0x0400
728 
729 #define	ZSD_CREATE_ALL	\
730 	(ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
731 #define	ZSD_SHUTDOWN_ALL	\
732 	(ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
733 #define	ZSD_DESTROY_ALL	\
734 	(ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
735 
736 #define	ZSD_ALL_INPROGRESS \
737 	(ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
738 
739 /*
740  * Macros to help with zone visibility restrictions.
741  */
742 
743 /*
744  * Is process in the global zone?
745  */
746 #define	INGLOBALZONE(p) \
747 	((p)->p_zone == global_zone)
748 
749 /*
750  * Can process view objects in given zone?
751  */
752 #define	HASZONEACCESS(p, zoneid) \
753 	((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
754 
755 /*
756  * Convenience macro to see if a resolved path is visible from within a
757  * given zone.
758  *
759  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
760  * two strings must be equal.  Since the rootpathlen has a trailing '/',
761  * we want to skip everything in the path up to (but not including) the
762  * trailing '/'.
763  */
764 #define	ZONE_PATH_VISIBLE(path, zone) \
765 	(strncmp((path), (zone)->zone_rootpath,		\
766 	    (zone)->zone_rootpathlen - 1) == 0)
767 
768 /*
769  * Convenience macro to go from the global view of a path to that seen
770  * from within said zone.  It is the responsibility of the caller to
771  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
772  * in fact visible from within the zone.
773  */
774 #define	ZONE_PATH_TRANSLATE(path, zone)	\
775 	(ASSERT(ZONE_PATH_VISIBLE(path, zone)),	\
776 	(path) + (zone)->zone_rootpathlen - 2)
777 
778 /*
779  * Special processes visible in all zones.
780  */
781 #define	ZONE_SPECIALPID(x)	 ((x) == 0 || (x) == 1)
782 
783 /*
784  * Zone-safe version of thread_create() to be used when the caller wants to
785  * create a kernel thread to run within the current zone's context.
786  */
787 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
788     pri_t);
789 extern void zthread_exit(void);
790 
791 /*
792  * Functions for an external observer to register interest in a zone's status
793  * change.  Observers will be woken up when the zone status equals the status
794  * argument passed in (in the case of zone_status_timedwait, the function may
795  * also return because of a timeout; zone_status_wait_sig may return early due
796  * to a signal being delivered; zone_status_timedwait_sig may return for any of
797  * the above reasons).
798  *
799  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
800  * cv_wait_sig() respectively.
801  */
802 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
803 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
804 extern void zone_status_wait(zone_t *, zone_status_t);
805 extern int zone_status_wait_sig(zone_t *, zone_status_t);
806 
807 /*
808  * Get the status  of the zone (at the time it was called).  The state may
809  * have progressed by the time it is returned.
810  */
811 extern zone_status_t zone_status_get(zone_t *);
812 
813 /*
814  * Safely get the hostid of the specified zone (defaults to machine's hostid
815  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
816  * the global zone's (i.e., physical system's) hostid.
817  */
818 extern uint32_t zone_get_hostid(zone_t *);
819 
820 /*
821  * Get the "kcred" credentials corresponding to the given zone.
822  */
823 extern struct cred *zone_get_kcred(zoneid_t);
824 
825 /*
826  * Get/set the pool the zone is currently bound to.
827  */
828 extern struct pool *zone_pool_get(zone_t *);
829 extern void zone_pool_set(zone_t *, struct pool *);
830 
831 /*
832  * Get/set the pset the zone is currently using.
833  */
834 extern psetid_t zone_pset_get(zone_t *);
835 extern void zone_pset_set(zone_t *, psetid_t);
836 
837 /*
838  * Get the number of cpus/online-cpus visible from the given zone.
839  */
840 extern int zone_ncpus_get(zone_t *);
841 extern int zone_ncpus_online_get(zone_t *);
842 
843 /*
844  * Returns true if the named pool/dataset is visible in the current zone.
845  */
846 extern int zone_dataset_visible(const char *, int *);
847 
848 /*
849  * zone version of kadmin()
850  */
851 extern int zone_kadmin(int, int, const char *, cred_t *);
852 extern void zone_shutdown_global(void);
853 
854 extern void mount_in_progress(zone_t *);
855 extern void mount_completed(zone_t *);
856 
857 extern int zone_walk(int (*)(zone_t *, void *), void *);
858 
859 extern rctl_hndl_t rc_zone_locked_mem;
860 extern rctl_hndl_t rc_zone_max_swap;
861 extern rctl_hndl_t rc_zone_max_lofi;
862 
863 #endif	/* _KERNEL */
864 
865 #ifdef	__cplusplus
866 }
867 #endif
868 
869 #endif	/* _SYS_ZONE_H */
870