xref: /illumos-gate/usr/src/uts/common/sys/zone.h (revision 3ba944265c4ae1fcf23ef758537c2e4f4feec16e)
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 2015 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 #define	ZONE_ATTR_INITNORESTART	20
107 
108 /* Start of the brand-specific attribute namespace */
109 #define	ZONE_ATTR_BRAND_ATTRS	32768
110 
111 #define	ZONE_FS_ALLOWED_MAX	1024
112 
113 #define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
114 #define	ZONE_EVENT_STATUS_CLASS	"status"
115 #define	ZONE_EVENT_STATUS_SUBCLASS	"change"
116 
117 #define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
118 #define	ZONE_EVENT_INITIALIZED		"initialized"
119 #define	ZONE_EVENT_READY		"ready"
120 #define	ZONE_EVENT_RUNNING		"running"
121 #define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
122 
123 #define	ZONE_CB_NAME		"zonename"
124 #define	ZONE_CB_NEWSTATE	"newstate"
125 #define	ZONE_CB_OLDSTATE	"oldstate"
126 #define	ZONE_CB_TIMESTAMP	"when"
127 #define	ZONE_CB_ZONEID		"zoneid"
128 
129 /*
130  * Exit values that may be returned by scripts or programs invoked by various
131  * zone commands.
132  *
133  * These are defined as:
134  *
135  *	ZONE_SUBPROC_OK
136  *	===============
137  *	The subprocess completed successfully.
138  *
139  *	ZONE_SUBPROC_USAGE
140  *	==================
141  *	The subprocess failed with a usage message, or a usage message should
142  *	be output in its behalf.
143  *
144  *	ZONE_SUBPROC_NOTCOMPLETE
145  *	========================
146  *	The subprocess did not complete, but the actions performed by the
147  *	subprocess require no recovery actions by the user.
148  *
149  *	For example, if the subprocess were called by "zoneadm install," the
150  *	installation of the zone did not succeed but the user need not perform
151  *	a "zoneadm uninstall" before attempting another install.
152  *
153  *	ZONE_SUBPROC_FATAL
154  *	==================
155  *	The subprocess failed in a fatal manner, usually one that will require
156  *	some type of recovery action by the user.
157  *
158  *	For example, if the subprocess were called by "zoneadm install," the
159  *	installation of the zone did not succeed and the user will need to
160  *	perform a "zoneadm uninstall" before another install attempt is
161  *	possible.
162  *
163  *	The non-success exit values are large to avoid accidental collision
164  *	with values used internally by some commands (e.g. "Z_ERR" and
165  *	"Z_USAGE" as used by zoneadm.)
166  */
167 #define	ZONE_SUBPROC_OK			0
168 #define	ZONE_SUBPROC_USAGE		253
169 #define	ZONE_SUBPROC_NOTCOMPLETE	254
170 #define	ZONE_SUBPROC_FATAL		255
171 
172 #ifdef _SYSCALL32
173 typedef struct {
174 	caddr32_t zone_name;
175 	caddr32_t zone_root;
176 	caddr32_t zone_privs;
177 	size32_t zone_privssz;
178 	caddr32_t rctlbuf;
179 	size32_t rctlbufsz;
180 	caddr32_t extended_error;
181 	caddr32_t zfsbuf;
182 	size32_t  zfsbufsz;
183 	int match;			/* match level */
184 	uint32_t doi;			/* DOI for label */
185 	caddr32_t label;		/* label associated with zone */
186 	int flags;
187 } zone_def32;
188 #endif
189 typedef struct {
190 	const char *zone_name;
191 	const char *zone_root;
192 	const struct priv_set *zone_privs;
193 	size_t zone_privssz;
194 	const char *rctlbuf;
195 	size_t rctlbufsz;
196 	int *extended_error;
197 	const char *zfsbuf;
198 	size_t zfsbufsz;
199 	int match;			/* match level */
200 	uint32_t doi;			/* DOI for label */
201 	const bslabel_t *label;		/* label associated with zone */
202 	int flags;
203 } zone_def;
204 
205 /* extended error information */
206 #define	ZE_UNKNOWN	0	/* No extended error info */
207 #define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
208 #define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
209 #define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
210 
211 /*
212  * zone_status values
213  *
214  * You must modify zone_status_names in mdb(1M)'s genunix module
215  * (genunix/zone.c) when you modify this enum.
216  */
217 typedef enum {
218 	ZONE_IS_UNINITIALIZED = 0,
219 	ZONE_IS_INITIALIZED,
220 	ZONE_IS_READY,
221 	ZONE_IS_BOOTING,
222 	ZONE_IS_RUNNING,
223 	ZONE_IS_SHUTTING_DOWN,
224 	ZONE_IS_EMPTY,
225 	ZONE_IS_DOWN,
226 	ZONE_IS_DYING,
227 	ZONE_IS_DEAD
228 } zone_status_t;
229 #define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
230 #define	ZONE_MAX_STATE		ZONE_IS_DEAD
231 
232 /*
233  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
234  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
235  */
236 typedef enum zone_cmd {
237 	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
238 	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
239 } zone_cmd_t;
240 
241 /*
242  * The structure of a request to zoneadmd.
243  */
244 typedef struct zone_cmd_arg {
245 	uint64_t	uniqid;		/* unique "generation number" */
246 	zone_cmd_t	cmd;		/* requested action */
247 	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
248 	char locale[MAXPATHLEN];	/* locale in which to render messages */
249 	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
250 } zone_cmd_arg_t;
251 
252 /*
253  * Structure of zoneadmd's response to a request.  A NULL return value means
254  * the caller should attempt to restart zoneadmd and retry.
255  */
256 typedef struct zone_cmd_rval {
257 	int rval;			/* return value of request */
258 	char errbuf[1];	/* variable-sized buffer containing error messages */
259 } zone_cmd_rval_t;
260 
261 /*
262  * The zone support infrastructure uses the zone name as a component
263  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
264  * in length, so ZONENAME_MAX is limited by that.
265  */
266 #define	ZONENAME_MAX		64
267 
268 #define	GLOBAL_ZONENAME		"global"
269 
270 /*
271  * Extended Regular expression (see regex(5)) which matches all valid zone
272  * names.
273  */
274 #define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
275 
276 /*
277  * Where the zones support infrastructure places temporary files.
278  */
279 #define	ZONES_TMPDIR		"/var/run/zones"
280 
281 /*
282  * The path to the door used by clients to communicate with zoneadmd.
283  */
284 #define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
285 
286 
287 /* zone_flags */
288 /*
289  * Threads that read or write the following flag must hold zone_lock.
290  */
291 #define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
292 
293 /*
294  * The following threads are set when the zone is created and never changed.
295  * Threads that test for these flags don't have to hold zone_lock.
296  */
297 #define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
298 #define	ZF_IS_SCRATCH		0x4	/* scratch zone */
299 #define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
300 
301 
302 /* zone_create flags */
303 #define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
304 
305 /* zone network properties */
306 #define	ZONE_NETWORK_ADDRESS	1
307 #define	ZONE_NETWORK_DEFROUTER	2
308 
309 #define	ZONE_NET_ADDRNAME	"address"
310 #define	ZONE_NET_RTRNAME	"route"
311 
312 typedef struct zone_net_data {
313 	int zn_type;
314 	int zn_len;
315 	datalink_id_t zn_linkid;
316 	uint8_t zn_val[1];
317 } zone_net_data_t;
318 
319 
320 #ifdef _KERNEL
321 
322 /*
323  * We need to protect the definition of 'list_t' from userland applications and
324  * libraries which may be defining ther own versions.
325  */
326 #include <sys/list.h>
327 #include <sys/loadavg.h>
328 
329 #define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
330 
331 struct pool;
332 struct brand;
333 
334 /*
335  * Each of these constants identifies a kernel subsystem that acquires and
336  * releases zone references.  Each subsystem that invokes
337  * zone_hold_ref() and zone_rele_ref() should specify the
338  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
339  * help users and developers quickly identify subsystems that stall zone
340  * shutdowns indefinitely.
341  *
342  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
343  * when you modify this enumeration.
344  */
345 typedef enum zone_ref_subsys {
346 	ZONE_REF_NFS,			/* NFS */
347 	ZONE_REF_NFSV4,			/* NFSv4 */
348 	ZONE_REF_SMBFS,			/* SMBFS */
349 	ZONE_REF_MNTFS,			/* MNTFS */
350 	ZONE_REF_LOFI,			/* LOFI devices */
351 	ZONE_REF_VFS,			/* VFS infrastructure */
352 	ZONE_REF_IPC,			/* IPC infrastructure */
353 	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
354 } zone_ref_subsys_t;
355 
356 /*
357  * zone_ref represents a general-purpose references to a zone.  Each zone's
358  * references are linked into the zone's zone_t::zone_ref_list.  This allows
359  * debuggers to walk zones' references.
360  */
361 typedef struct zone_ref {
362 	struct zone	*zref_zone; /* the zone to which the reference refers */
363 	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
364 } zone_ref_t;
365 
366 /*
367  * Structure to record list of ZFS datasets exported to a zone.
368  */
369 typedef struct zone_dataset {
370 	char		*zd_dataset;
371 	list_node_t	zd_linkage;
372 } zone_dataset_t;
373 
374 /*
375  * structure for zone kstats
376  */
377 typedef struct zone_kstat {
378 	kstat_named_t zk_zonename;
379 	kstat_named_t zk_usage;
380 	kstat_named_t zk_value;
381 } zone_kstat_t;
382 
383 struct cpucap;
384 
385 typedef struct {
386 	kstat_named_t	zm_zonename;
387 	kstat_named_t	zm_pgpgin;
388 	kstat_named_t	zm_anonpgin;
389 	kstat_named_t	zm_execpgin;
390 	kstat_named_t	zm_fspgin;
391 	kstat_named_t	zm_anon_alloc_fail;
392 } zone_mcap_kstat_t;
393 
394 typedef struct {
395 	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
396 	kstat_named_t	zm_utime;
397 	kstat_named_t	zm_stime;
398 	kstat_named_t	zm_wtime;
399 	kstat_named_t	zm_avenrun1;
400 	kstat_named_t	zm_avenrun5;
401 	kstat_named_t	zm_avenrun15;
402 	kstat_named_t	zm_ffcap;
403 	kstat_named_t	zm_ffnoproc;
404 	kstat_named_t	zm_ffnomem;
405 	kstat_named_t	zm_ffmisc;
406 	kstat_named_t	zm_nested_intp;
407 	kstat_named_t	zm_init_pid;
408 	kstat_named_t	zm_boot_time;
409 } zone_misc_kstat_t;
410 
411 typedef struct zone {
412 	/*
413 	 * zone_name is never modified once set.
414 	 */
415 	char		*zone_name;	/* zone's configuration name */
416 	/*
417 	 * zone_nodename and zone_domain are never freed once allocated.
418 	 */
419 	char		*zone_nodename;	/* utsname.nodename equivalent */
420 	char		*zone_domain;	/* srpc_domain equivalent */
421 	/*
422 	 * zone_hostid is used for per-zone hostid emulation.
423 	 * Currently it isn't modified after it's set (so no locks protect
424 	 * accesses), but that might have to change when we allow
425 	 * administrators to change running zones' properties.
426 	 *
427 	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
428 	 * that zone_get_hostid() will function correctly.
429 	 */
430 	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
431 					/* if not emulated */
432 	/*
433 	 * zone_lock protects the following fields of a zone_t:
434 	 * 	zone_ref
435 	 * 	zone_cred_ref
436 	 * 	zone_subsys_ref
437 	 * 	zone_ref_list
438 	 * 	zone_ntasks
439 	 * 	zone_flags
440 	 * 	zone_zsd
441 	 *	zone_pfexecd
442 	 */
443 	kmutex_t	zone_lock;
444 	/*
445 	 * zone_linkage is the zone's linkage into the active or
446 	 * death-row list.  The field is protected by zonehash_lock.
447 	 */
448 	list_node_t	zone_linkage;
449 	zoneid_t	zone_id;	/* ID of zone */
450 	uint_t		zone_ref;	/* count of zone_hold()s on zone */
451 	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
452 	/*
453 	 * Fixed-sized array of subsystem-specific reference counts
454 	 * The sum of all of the counts must be less than or equal to zone_ref.
455 	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
456 	 * constants.
457 	 */
458 	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
459 	list_t		zone_ref_list;	/* list of zone_ref_t structs */
460 	/*
461 	 * zone_rootvp and zone_rootpath can never be modified once set.
462 	 */
463 	struct vnode	*zone_rootvp;	/* zone's root vnode */
464 	char		*zone_rootpath;	/* Path to zone's root + '/' */
465 	ushort_t	zone_flags;	/* misc flags */
466 	zone_status_t	zone_status;	/* protected by zone_status_lock */
467 	uint_t		zone_ntasks;	/* number of tasks executing in zone */
468 	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
469 					/* counters in projects and tasks */
470 					/* that are within the zone */
471 	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
472 	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
473 	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
474 	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
475 
476 	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
477 	uint32_t	zone_shares;	/* FSS shares allocated to zone */
478 	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
479 	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
480 					/* kpd_locked_mem for all */
481 					/* projects in zone. */
482 					/* Also protects zone_max_swap */
483 					/* grab after p_lock, before rcs_lock */
484 	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
485 						/* zone */
486 	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
487 						/* limit.  Protected by */
488 						/* zone_rctls->rcs_lock */
489 	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
490 	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
491 						/* Protected by */
492 						/* zone_rctls->rcs_lock */
493 	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
494 	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
495 	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
496 						/* Protected by */
497 						/* zone_rctls->rcs_lock */
498 	list_t		zone_zsd;	/* list of Zone-Specific Data values */
499 	kcondvar_t	zone_cv;	/* used to signal state changes */
500 	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
501 	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
502 	char		*zone_initname;	/* fs path to 'init' */
503 	int		zone_boot_err;  /* for zone_boot() if boot fails */
504 	char		*zone_bootargs;	/* arguments passed via zone_boot() */
505 	uint64_t	zone_phys_mcap;	/* physical memory cap */
506 	/*
507 	 * zone_kthreads is protected by zone_status_lock.
508 	 */
509 	kthread_t	*zone_kthreads;	/* kernel threads in zone */
510 	struct priv_set	*zone_privset;	/* limit set for zone */
511 	/*
512 	 * zone_vfslist is protected by vfs_list_lock().
513 	 */
514 	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
515 	uint64_t	zone_uniqid;	/* unique zone generation number */
516 	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
517 	/*
518 	 * zone_pool is protected by pool_lock().
519 	 */
520 	struct pool	*zone_pool;	/* pool the zone is bound to */
521 	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
522 	/* zone_psetid is protected by cpu_lock */
523 	psetid_t	zone_psetid;	/* pset the zone is bound to */
524 
525 	time_t		zone_boot_time;	/* Similar to boot_time */
526 
527 	/*
528 	 * The following two can be read without holding any locks.  They are
529 	 * updated under cpu_lock.
530 	 */
531 	int		zone_ncpus;  /* zone's idea of ncpus */
532 	int		zone_ncpus_online; /* zone's idea of ncpus_online */
533 	/*
534 	 * List of ZFS datasets exported to this zone.
535 	 */
536 	list_t		zone_datasets;	/* list of datasets */
537 
538 	ts_label_t	*zone_slabel;	/* zone sensitivity label */
539 	int		zone_match;	/* require label match for packets */
540 	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
541 
542 	boolean_t	zone_restart_init;	/* Restart init if it dies? */
543 	struct brand	*zone_brand;		/* zone's brand */
544 	void 		*zone_brand_data;	/* store brand specific data */
545 	id_t		zone_defaultcid;	/* dflt scheduling class id */
546 	kstat_t		*zone_swapresv_kstat;
547 	kstat_t		*zone_lockedmem_kstat;
548 	/*
549 	 * zone_dl_list is protected by zone_lock
550 	 */
551 	list_t		zone_dl_list;
552 	netstack_t	*zone_netstack;
553 	struct cpucap	*zone_cpucap;	/* CPU caps data */
554 	/*
555 	 * Solaris Auditing per-zone audit context
556 	 */
557 	struct au_kcontext	*zone_audit_kctxt;
558 	/*
559 	 * For private use by mntfs.
560 	 */
561 	struct mntelem	*zone_mntfs_db;
562 	krwlock_t	zone_mntfs_db_lock;
563 
564 	struct klpd_reg		*zone_pfexecd;
565 
566 	char		*zone_fs_allowed;
567 	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
568 	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
569 						/* zone_rctls->rcs_lock */
570 	kstat_t		*zone_nprocs_kstat;
571 
572 	kmutex_t	zone_mcap_lock;	/* protects mcap statistics */
573 	kstat_t		*zone_mcap_ksp;
574 	zone_mcap_kstat_t *zone_mcap_stats;
575 	uint64_t	zone_pgpgin;		/* pages paged in */
576 	uint64_t	zone_anonpgin;		/* anon pages paged in */
577 	uint64_t	zone_execpgin;		/* exec pages paged in */
578 	uint64_t	zone_fspgin;		/* fs pages paged in */
579 	uint64_t	zone_anon_alloc_fail;	/* cnt of anon alloc fails */
580 
581 	/*
582 	 * Misc. kstats and counters for zone cpu-usage aggregation.
583 	 * The zone_Xtime values are the sum of the micro-state accounting
584 	 * values for all threads that are running or have run in the zone.
585 	 * This is tracked in msacct.c as threads change state.
586 	 * The zone_stime is the sum of the LMS_SYSTEM times.
587 	 * The zone_utime is the sum of the LMS_USER times.
588 	 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
589 	 * As with per-thread micro-state accounting values, these values are
590 	 * not scaled to nanosecs.  The scaling is done by the
591 	 * zone_misc_kstat_update function when kstats are requested.
592 	 */
593 	kmutex_t	zone_misc_lock;		/* protects misc statistics */
594 	kstat_t		*zone_misc_ksp;
595 	zone_misc_kstat_t *zone_misc_stats;
596 	uint64_t	zone_stime;		/* total system time */
597 	uint64_t	zone_utime;		/* total user time */
598 	uint64_t	zone_wtime;		/* total time waiting in runq */
599 	/* fork-fail kstat tracking */
600 	uint32_t	zone_ffcap;		/* hit an rctl cap */
601 	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
602 	uint32_t	zone_ffnomem;		/* as_dup/memory error */
603 	uint32_t	zone_ffmisc;		/* misc. other error */
604 
605 	uint32_t	zone_nested_intp;	/* nested interp. kstat */
606 
607 	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
608 	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
609 	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
610 
611 	/*
612 	 * FSS stats updated once per second by fss_decay_usage.
613 	 */
614 	uint32_t	zone_fss_gen;		/* FSS generation cntr */
615 	uint64_t	zone_run_ticks;		/* tot # of ticks running */
616 
617 	/*
618 	 * DTrace-private per-zone state
619 	 */
620 	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
621 
622 	/*
623 	 * Synchronization primitives used to synchronize between mounts and
624 	 * zone creation/destruction.
625 	 */
626 	int		zone_mounts_in_progress;
627 	kcondvar_t	zone_mount_cv;
628 	kmutex_t	zone_mount_lock;
629 } zone_t;
630 
631 /*
632  * Special value of zone_psetid to indicate that pools are disabled.
633  */
634 #define	ZONE_PS_INVAL	PS_MYID
635 
636 
637 extern zone_t zone0;
638 extern zone_t *global_zone;
639 extern uint_t maxzones;
640 extern rctl_hndl_t rc_zone_nlwps;
641 extern rctl_hndl_t rc_zone_nprocs;
642 
643 extern long zone(int, void *, void *, void *, void *);
644 extern void zone_zsd_init(void);
645 extern void zone_init(void);
646 extern void zone_hold(zone_t *);
647 extern void zone_rele(zone_t *);
648 extern void zone_init_ref(zone_ref_t *);
649 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
650 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
651 extern void zone_cred_hold(zone_t *);
652 extern void zone_cred_rele(zone_t *);
653 extern void zone_task_hold(zone_t *);
654 extern void zone_task_rele(zone_t *);
655 extern zone_t *zone_find_by_id(zoneid_t);
656 extern zone_t *zone_find_by_label(const ts_label_t *);
657 extern zone_t *zone_find_by_name(char *);
658 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
659 extern zone_t *zone_find_by_path(const char *);
660 extern zoneid_t getzoneid(void);
661 extern zone_t *zone_find_by_id_nolock(zoneid_t);
662 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
663 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
664 extern void zone_loadavg_update();
665 
666 /*
667  * Zone-specific data (ZSD) APIs
668  */
669 /*
670  * The following is what code should be initializing its zone_key_t to if it
671  * calls zone_getspecific() without necessarily knowing that zone_key_create()
672  * has been called on the key.
673  */
674 #define	ZONE_KEY_UNINITIALIZED	0
675 
676 typedef uint_t zone_key_t;
677 
678 extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
679     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
680 extern int 	zone_key_delete(zone_key_t);
681 extern void	*zone_getspecific(zone_key_t, zone_t *);
682 extern int	zone_setspecific(zone_key_t, zone_t *, const void *);
683 
684 /*
685  * The definition of a zsd_entry is truly private to zone.c and is only
686  * placed here so it can be shared with mdb.
687  *
688  * State maintained for each zone times each registered key, which tracks
689  * the state of the create, shutdown and destroy callbacks.
690  *
691  * zsd_flags is used to keep track of pending actions to avoid holding locks
692  * when calling the create/shutdown/destroy callbacks, since doing so
693  * could lead to deadlocks.
694  */
695 struct zsd_entry {
696 	zone_key_t		zsd_key;	/* Key used to lookup value */
697 	void			*zsd_data;	/* Caller-managed value */
698 	/*
699 	 * Callbacks to be executed when a zone is created, shutdown, and
700 	 * destroyed, respectively.
701 	 */
702 	void			*(*zsd_create)(zoneid_t);
703 	void			(*zsd_shutdown)(zoneid_t, void *);
704 	void			(*zsd_destroy)(zoneid_t, void *);
705 	list_node_t		zsd_linkage;
706 	uint16_t 		zsd_flags;	/* See below */
707 	kcondvar_t		zsd_cv;
708 };
709 
710 /*
711  * zsd_flags
712  */
713 #define	ZSD_CREATE_NEEDED	0x0001
714 #define	ZSD_CREATE_INPROGRESS	0x0002
715 #define	ZSD_CREATE_COMPLETED	0x0004
716 #define	ZSD_SHUTDOWN_NEEDED	0x0010
717 #define	ZSD_SHUTDOWN_INPROGRESS	0x0020
718 #define	ZSD_SHUTDOWN_COMPLETED	0x0040
719 #define	ZSD_DESTROY_NEEDED	0x0100
720 #define	ZSD_DESTROY_INPROGRESS	0x0200
721 #define	ZSD_DESTROY_COMPLETED	0x0400
722 
723 #define	ZSD_CREATE_ALL	\
724 	(ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
725 #define	ZSD_SHUTDOWN_ALL	\
726 	(ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
727 #define	ZSD_DESTROY_ALL	\
728 	(ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
729 
730 #define	ZSD_ALL_INPROGRESS \
731 	(ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
732 
733 /*
734  * Macros to help with zone visibility restrictions.
735  */
736 
737 /*
738  * Is process in the global zone?
739  */
740 #define	INGLOBALZONE(p) \
741 	((p)->p_zone == global_zone)
742 
743 /*
744  * Can process view objects in given zone?
745  */
746 #define	HASZONEACCESS(p, zoneid) \
747 	((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
748 
749 /*
750  * Convenience macro to see if a resolved path is visible from within a
751  * given zone.
752  *
753  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
754  * two strings must be equal.  Since the rootpathlen has a trailing '/',
755  * we want to skip everything in the path up to (but not including) the
756  * trailing '/'.
757  */
758 #define	ZONE_PATH_VISIBLE(path, zone) \
759 	(strncmp((path), (zone)->zone_rootpath,		\
760 	    (zone)->zone_rootpathlen - 1) == 0)
761 
762 /*
763  * Convenience macro to go from the global view of a path to that seen
764  * from within said zone.  It is the responsibility of the caller to
765  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
766  * in fact visible from within the zone.
767  */
768 #define	ZONE_PATH_TRANSLATE(path, zone)	\
769 	(ASSERT(ZONE_PATH_VISIBLE(path, zone)),	\
770 	(path) + (zone)->zone_rootpathlen - 2)
771 
772 /*
773  * Special processes visible in all zones.
774  */
775 #define	ZONE_SPECIALPID(x)	 ((x) == 0 || (x) == 1)
776 
777 /*
778  * Zone-safe version of thread_create() to be used when the caller wants to
779  * create a kernel thread to run within the current zone's context.
780  */
781 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
782     pri_t);
783 extern void zthread_exit(void);
784 
785 /*
786  * Functions for an external observer to register interest in a zone's status
787  * change.  Observers will be woken up when the zone status equals the status
788  * argument passed in (in the case of zone_status_timedwait, the function may
789  * also return because of a timeout; zone_status_wait_sig may return early due
790  * to a signal being delivered; zone_status_timedwait_sig may return for any of
791  * the above reasons).
792  *
793  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
794  * cv_wait_sig() respectively.
795  */
796 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
797 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
798 extern void zone_status_wait(zone_t *, zone_status_t);
799 extern int zone_status_wait_sig(zone_t *, zone_status_t);
800 
801 /*
802  * Get the status  of the zone (at the time it was called).  The state may
803  * have progressed by the time it is returned.
804  */
805 extern zone_status_t zone_status_get(zone_t *);
806 
807 /*
808  * Safely get the hostid of the specified zone (defaults to machine's hostid
809  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
810  * the global zone's (i.e., physical system's) hostid.
811  */
812 extern uint32_t zone_get_hostid(zone_t *);
813 
814 /*
815  * Get the "kcred" credentials corresponding to the given zone.
816  */
817 extern struct cred *zone_get_kcred(zoneid_t);
818 
819 /*
820  * Get/set the pool the zone is currently bound to.
821  */
822 extern struct pool *zone_pool_get(zone_t *);
823 extern void zone_pool_set(zone_t *, struct pool *);
824 
825 /*
826  * Get/set the pset the zone is currently using.
827  */
828 extern psetid_t zone_pset_get(zone_t *);
829 extern void zone_pset_set(zone_t *, psetid_t);
830 
831 /*
832  * Get the number of cpus/online-cpus visible from the given zone.
833  */
834 extern int zone_ncpus_get(zone_t *);
835 extern int zone_ncpus_online_get(zone_t *);
836 
837 /*
838  * Returns true if the named pool/dataset is visible in the current zone.
839  */
840 extern int zone_dataset_visible(const char *, int *);
841 
842 /*
843  * zone version of kadmin()
844  */
845 extern int zone_kadmin(int, int, const char *, cred_t *);
846 extern void zone_shutdown_global(void);
847 
848 extern void mount_in_progress(zone_t *);
849 extern void mount_completed(zone_t *);
850 
851 extern int zone_walk(int (*)(zone_t *, void *), void *);
852 
853 extern rctl_hndl_t rc_zone_locked_mem;
854 extern rctl_hndl_t rc_zone_max_swap;
855 extern rctl_hndl_t rc_zone_max_lofi;
856 
857 #endif	/* _KERNEL */
858 
859 #ifdef	__cplusplus
860 }
861 #endif
862 
863 #endif	/* _SYS_ZONE_H */
864