xref: /illumos-gate/usr/src/uts/common/sys/zone.h (revision dcdeca0a948b9d3139743db085d193c9db8ff2a3)
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 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 <sys/secflags.h>
44 #include <netinet/in.h>
45 #include <sys/cpu_uarray.h>
46 
47 #ifdef	__cplusplus
48 extern "C" {
49 #endif
50 
51 /*
52  * NOTE
53  *
54  * The contents of this file are private to the implementation of
55  * Solaris and are subject to change at any time without notice.
56  * Applications and drivers using these interfaces may fail to
57  * run on future releases.
58  */
59 
60 /* Available both in kernel and for user space */
61 
62 /* zone id restrictions and special ids */
63 #define	MAX_ZONEID	9999
64 #define	MIN_USERZONEID	1	/* lowest user-creatable zone ID */
65 #define	MIN_ZONEID	0	/* minimum zone ID on system */
66 #define	GLOBAL_ZONEID	0
67 #define	ZONEID_WIDTH	4	/* for printf */
68 
69 /*
70  * Special zoneid_t token to refer to all zones.
71  */
72 #define	ALL_ZONES	(-1)
73 
74 /* system call subcodes */
75 #define	ZONE_CREATE		0
76 #define	ZONE_DESTROY		1
77 #define	ZONE_GETATTR		2
78 #define	ZONE_ENTER		3
79 #define	ZONE_LIST		4
80 #define	ZONE_SHUTDOWN		5
81 #define	ZONE_LOOKUP		6
82 #define	ZONE_BOOT		7
83 #define	ZONE_VERSION		8
84 #define	ZONE_SETATTR		9
85 #define	ZONE_ADD_DATALINK	10
86 #define	ZONE_DEL_DATALINK	11
87 #define	ZONE_CHECK_DATALINK	12
88 #define	ZONE_LIST_DATALINK	13
89 
90 /* zone attributes */
91 #define	ZONE_ATTR_ROOT		1
92 #define	ZONE_ATTR_NAME		2
93 #define	ZONE_ATTR_STATUS	3
94 #define	ZONE_ATTR_PRIVSET	4
95 #define	ZONE_ATTR_UNIQID	5
96 #define	ZONE_ATTR_POOLID	6
97 #define	ZONE_ATTR_INITPID	7
98 #define	ZONE_ATTR_SLBL		8
99 #define	ZONE_ATTR_INITNAME	9
100 #define	ZONE_ATTR_BOOTARGS	10
101 #define	ZONE_ATTR_BRAND		11
102 #define	ZONE_ATTR_PHYS_MCAP	12
103 #define	ZONE_ATTR_SCHED_CLASS	13
104 #define	ZONE_ATTR_FLAGS		14
105 #define	ZONE_ATTR_HOSTID	15
106 #define	ZONE_ATTR_FS_ALLOWED	16
107 #define	ZONE_ATTR_NETWORK	17
108 #define	ZONE_ATTR_INITNORESTART	20
109 #define	ZONE_ATTR_SECFLAGS	21
110 
111 /* Start of the brand-specific attribute namespace */
112 #define	ZONE_ATTR_BRAND_ATTRS	32768
113 
114 #define	ZONE_FS_ALLOWED_MAX	1024
115 
116 #define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
117 #define	ZONE_EVENT_STATUS_CLASS	"status"
118 #define	ZONE_EVENT_STATUS_SUBCLASS	"change"
119 
120 #define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
121 #define	ZONE_EVENT_INITIALIZED		"initialized"
122 #define	ZONE_EVENT_READY		"ready"
123 #define	ZONE_EVENT_RUNNING		"running"
124 #define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
125 
126 #define	ZONE_CB_NAME		"zonename"
127 #define	ZONE_CB_NEWSTATE	"newstate"
128 #define	ZONE_CB_OLDSTATE	"oldstate"
129 #define	ZONE_CB_TIMESTAMP	"when"
130 #define	ZONE_CB_ZONEID		"zoneid"
131 
132 /*
133  * Exit values that may be returned by scripts or programs invoked by various
134  * zone commands.
135  *
136  * These are defined as:
137  *
138  *	ZONE_SUBPROC_OK
139  *	===============
140  *	The subprocess completed successfully.
141  *
142  *	ZONE_SUBPROC_USAGE
143  *	==================
144  *	The subprocess failed with a usage message, or a usage message should
145  *	be output in its behalf.
146  *
147  *	ZONE_SUBPROC_NOTCOMPLETE
148  *	========================
149  *	The subprocess did not complete, but the actions performed by the
150  *	subprocess require no recovery actions by the user.
151  *
152  *	For example, if the subprocess were called by "zoneadm install," the
153  *	installation of the zone did not succeed but the user need not perform
154  *	a "zoneadm uninstall" before attempting another install.
155  *
156  *	ZONE_SUBPROC_FATAL
157  *	==================
158  *	The subprocess failed in a fatal manner, usually one that will require
159  *	some type of recovery action by the user.
160  *
161  *	For example, if the subprocess were called by "zoneadm install," the
162  *	installation of the zone did not succeed and the user will need to
163  *	perform a "zoneadm uninstall" before another install attempt is
164  *	possible.
165  *
166  *	The non-success exit values are large to avoid accidental collision
167  *	with values used internally by some commands (e.g. "Z_ERR" and
168  *	"Z_USAGE" as used by zoneadm.)
169  */
170 #define	ZONE_SUBPROC_OK			0
171 #define	ZONE_SUBPROC_USAGE		253
172 #define	ZONE_SUBPROC_NOTCOMPLETE	254
173 #define	ZONE_SUBPROC_FATAL		255
174 
175 #ifdef _SYSCALL32
176 typedef struct {
177 	caddr32_t zone_name;
178 	caddr32_t zone_root;
179 	caddr32_t zone_privs;
180 	size32_t zone_privssz;
181 	caddr32_t rctlbuf;
182 	size32_t rctlbufsz;
183 	caddr32_t extended_error;
184 	caddr32_t zfsbuf;
185 	size32_t  zfsbufsz;
186 	int match;			/* match level */
187 	uint32_t doi;			/* DOI for label */
188 	caddr32_t label;		/* label associated with zone */
189 	int flags;
190 } zone_def32;
191 #endif
192 typedef struct {
193 	const char *zone_name;
194 	const char *zone_root;
195 	const struct priv_set *zone_privs;
196 	size_t zone_privssz;
197 	const char *rctlbuf;
198 	size_t rctlbufsz;
199 	int *extended_error;
200 	const char *zfsbuf;
201 	size_t zfsbufsz;
202 	int match;			/* match level */
203 	uint32_t doi;			/* DOI for label */
204 	const bslabel_t *label;		/* label associated with zone */
205 	int flags;
206 } zone_def;
207 
208 /* extended error information */
209 #define	ZE_UNKNOWN	0	/* No extended error info */
210 #define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
211 #define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
212 #define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
213 
214 /*
215  * zone_status values
216  *
217  * You must modify zone_status_names in mdb(1M)'s genunix module
218  * (genunix/zone.c) when you modify this enum.
219  */
220 typedef enum {
221 	ZONE_IS_UNINITIALIZED = 0,
222 	ZONE_IS_INITIALIZED,
223 	ZONE_IS_READY,
224 	ZONE_IS_BOOTING,
225 	ZONE_IS_RUNNING,
226 	ZONE_IS_SHUTTING_DOWN,
227 	ZONE_IS_EMPTY,
228 	ZONE_IS_DOWN,
229 	ZONE_IS_DYING,
230 	ZONE_IS_DEAD
231 } zone_status_t;
232 #define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
233 #define	ZONE_MAX_STATE		ZONE_IS_DEAD
234 
235 /*
236  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
237  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
238  */
239 typedef enum zone_cmd {
240 	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
241 	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
242 } zone_cmd_t;
243 
244 /*
245  * The structure of a request to zoneadmd.
246  */
247 typedef struct zone_cmd_arg {
248 	uint64_t	uniqid;		/* unique "generation number" */
249 	zone_cmd_t	cmd;		/* requested action */
250 	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
251 	char locale[MAXPATHLEN];	/* locale in which to render messages */
252 	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
253 } zone_cmd_arg_t;
254 
255 /*
256  * Structure of zoneadmd's response to a request.  A NULL return value means
257  * the caller should attempt to restart zoneadmd and retry.
258  */
259 typedef struct zone_cmd_rval {
260 	int rval;			/* return value of request */
261 	char errbuf[1];	/* variable-sized buffer containing error messages */
262 } zone_cmd_rval_t;
263 
264 /*
265  * The zone support infrastructure uses the zone name as a component
266  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
267  * in length, so ZONENAME_MAX is limited by that.
268  */
269 #define	ZONENAME_MAX		64
270 
271 #define	GLOBAL_ZONENAME		"global"
272 
273 /*
274  * Extended Regular expression (see regex(5)) which matches all valid zone
275  * names.
276  */
277 #define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
278 
279 /*
280  * Where the zones support infrastructure places temporary files.
281  */
282 #define	ZONES_TMPDIR		"/var/run/zones"
283 
284 /*
285  * The path to the door used by clients to communicate with zoneadmd.
286  */
287 #define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
288 
289 
290 /* zone_flags */
291 /*
292  * Threads that read or write the following flag must hold zone_lock.
293  */
294 #define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
295 
296 /*
297  * The following threads are set when the zone is created and never changed.
298  * Threads that test for these flags don't have to hold zone_lock.
299  */
300 #define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
301 #define	ZF_IS_SCRATCH		0x4	/* scratch zone */
302 #define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
303 
304 
305 /* zone_create flags */
306 #define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
307 
308 /* zone network properties */
309 #define	ZONE_NETWORK_ADDRESS	1
310 #define	ZONE_NETWORK_DEFROUTER	2
311 
312 #define	ZONE_NET_ADDRNAME	"address"
313 #define	ZONE_NET_RTRNAME	"route"
314 
315 typedef struct zone_net_data {
316 	int zn_type;
317 	int zn_len;
318 	datalink_id_t zn_linkid;
319 	uint8_t zn_val[1];
320 } zone_net_data_t;
321 
322 
323 #ifdef _KERNEL
324 
325 /*
326  * We need to protect the definition of 'list_t' from userland applications and
327  * libraries which may be defining ther own versions.
328  */
329 #include <sys/list.h>
330 #include <sys/loadavg.h>
331 
332 #define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
333 
334 /*
335  * Indexes into ->zone_ustate array, summing the micro state of all threads in a
336  * particular zone.
337  */
338 #define	ZONE_USTATE_STIME (0)
339 #define	ZONE_USTATE_UTIME (1)
340 #define	ZONE_USTATE_WTIME (2)
341 #define	ZONE_USTATE_MAX (3)
342 
343 struct pool;
344 struct brand;
345 
346 /*
347  * Each of these constants identifies a kernel subsystem that acquires and
348  * releases zone references.  Each subsystem that invokes
349  * zone_hold_ref() and zone_rele_ref() should specify the
350  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
351  * help users and developers quickly identify subsystems that stall zone
352  * shutdowns indefinitely.
353  *
354  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
355  * when you modify this enumeration.
356  */
357 typedef enum zone_ref_subsys {
358 	ZONE_REF_NFS,			/* NFS */
359 	ZONE_REF_NFSV4,			/* NFSv4 */
360 	ZONE_REF_SMBFS,			/* SMBFS */
361 	ZONE_REF_MNTFS,			/* MNTFS */
362 	ZONE_REF_LOFI,			/* LOFI devices */
363 	ZONE_REF_VFS,			/* VFS infrastructure */
364 	ZONE_REF_IPC,			/* IPC infrastructure */
365 	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
366 } zone_ref_subsys_t;
367 
368 /*
369  * zone_ref represents a general-purpose references to a zone.  Each zone's
370  * references are linked into the zone's zone_t::zone_ref_list.  This allows
371  * debuggers to walk zones' references.
372  */
373 typedef struct zone_ref {
374 	struct zone	*zref_zone; /* the zone to which the reference refers */
375 	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
376 } zone_ref_t;
377 
378 /*
379  * Structure to record list of ZFS datasets exported to a zone.
380  */
381 typedef struct zone_dataset {
382 	char		*zd_dataset;
383 	list_node_t	zd_linkage;
384 } zone_dataset_t;
385 
386 /*
387  * structure for zone kstats
388  */
389 typedef struct zone_kstat {
390 	kstat_named_t zk_zonename;
391 	kstat_named_t zk_usage;
392 	kstat_named_t zk_value;
393 } zone_kstat_t;
394 
395 struct cpucap;
396 
397 typedef struct {
398 	kstat_named_t	zm_zonename;
399 	kstat_named_t	zm_pgpgin;
400 	kstat_named_t	zm_anonpgin;
401 	kstat_named_t	zm_execpgin;
402 	kstat_named_t	zm_fspgin;
403 	kstat_named_t	zm_anon_alloc_fail;
404 } zone_mcap_kstat_t;
405 
406 typedef struct {
407 	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
408 	kstat_named_t	zm_utime;
409 	kstat_named_t	zm_stime;
410 	kstat_named_t	zm_wtime;
411 	kstat_named_t	zm_avenrun1;
412 	kstat_named_t	zm_avenrun5;
413 	kstat_named_t	zm_avenrun15;
414 	kstat_named_t	zm_ffcap;
415 	kstat_named_t	zm_ffnoproc;
416 	kstat_named_t	zm_ffnomem;
417 	kstat_named_t	zm_ffmisc;
418 	kstat_named_t	zm_nested_intp;
419 	kstat_named_t	zm_init_pid;
420 	kstat_named_t	zm_boot_time;
421 } zone_misc_kstat_t;
422 
423 typedef struct zone {
424 	/*
425 	 * zone_name is never modified once set.
426 	 */
427 	char		*zone_name;	/* zone's configuration name */
428 	/*
429 	 * zone_nodename and zone_domain are never freed once allocated.
430 	 */
431 	char		*zone_nodename;	/* utsname.nodename equivalent */
432 	char		*zone_domain;	/* srpc_domain equivalent */
433 	/*
434 	 * zone_hostid is used for per-zone hostid emulation.
435 	 * Currently it isn't modified after it's set (so no locks protect
436 	 * accesses), but that might have to change when we allow
437 	 * administrators to change running zones' properties.
438 	 *
439 	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
440 	 * that zone_get_hostid() will function correctly.
441 	 */
442 	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
443 					/* if not emulated */
444 	/*
445 	 * zone_lock protects the following fields of a zone_t:
446 	 *	zone_ref
447 	 *	zone_cred_ref
448 	 *	zone_subsys_ref
449 	 *	zone_ref_list
450 	 *	zone_ntasks
451 	 *	zone_flags
452 	 *	zone_zsd
453 	 *	zone_pfexecd
454 	 */
455 	kmutex_t	zone_lock;
456 	/*
457 	 * zone_linkage is the zone's linkage into the active or
458 	 * death-row list.  The field is protected by zonehash_lock.
459 	 */
460 	list_node_t	zone_linkage;
461 	zoneid_t	zone_id;	/* ID of zone */
462 	uint_t		zone_ref;	/* count of zone_hold()s on zone */
463 	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
464 	/*
465 	 * Fixed-sized array of subsystem-specific reference counts
466 	 * The sum of all of the counts must be less than or equal to zone_ref.
467 	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
468 	 * constants.
469 	 */
470 	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
471 	list_t		zone_ref_list;	/* list of zone_ref_t structs */
472 	/*
473 	 * zone_rootvp and zone_rootpath can never be modified once set.
474 	 */
475 	struct vnode	*zone_rootvp;	/* zone's root vnode */
476 	char		*zone_rootpath;	/* Path to zone's root + '/' */
477 	ushort_t	zone_flags;	/* misc flags */
478 	zone_status_t	zone_status;	/* protected by zone_status_lock */
479 	uint_t		zone_ntasks;	/* number of tasks executing in zone */
480 	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
481 					/* counters in projects and tasks */
482 					/* that are within the zone */
483 	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
484 	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
485 	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
486 	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
487 
488 	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
489 	uint32_t	zone_shares;	/* FSS shares allocated to zone */
490 	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
491 	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
492 					/* kpd_locked_mem for all */
493 					/* projects in zone. */
494 					/* Also protects zone_max_swap */
495 					/* grab after p_lock, before rcs_lock */
496 	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
497 						/* zone */
498 	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
499 						/* limit.  Protected by */
500 						/* zone_rctls->rcs_lock */
501 	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
502 	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
503 						/* Protected by */
504 						/* zone_rctls->rcs_lock */
505 	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
506 	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
507 	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
508 						/* Protected by */
509 						/* zone_rctls->rcs_lock */
510 	list_t		zone_zsd;	/* list of Zone-Specific Data values */
511 	kcondvar_t	zone_cv;	/* used to signal state changes */
512 	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
513 	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
514 	char		*zone_initname;	/* fs path to 'init' */
515 	int		zone_boot_err;  /* for zone_boot() if boot fails */
516 	char		*zone_bootargs;	/* arguments passed via zone_boot() */
517 	uint64_t	zone_phys_mcap;	/* physical memory cap */
518 	/*
519 	 * zone_kthreads is protected by zone_status_lock.
520 	 */
521 	kthread_t	*zone_kthreads;	/* kernel threads in zone */
522 	struct priv_set	*zone_privset;	/* limit set for zone */
523 	/*
524 	 * zone_vfslist is protected by vfs_list_lock().
525 	 */
526 	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
527 	uint64_t	zone_uniqid;	/* unique zone generation number */
528 	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
529 	/*
530 	 * zone_pool is protected by pool_lock().
531 	 */
532 	struct pool	*zone_pool;	/* pool the zone is bound to */
533 	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
534 	/* zone_psetid is protected by cpu_lock */
535 	psetid_t	zone_psetid;	/* pset the zone is bound to */
536 
537 	time_t		zone_boot_time;	/* Similar to boot_time */
538 
539 	/*
540 	 * The following two can be read without holding any locks.  They are
541 	 * updated under cpu_lock.
542 	 */
543 	int		zone_ncpus;  /* zone's idea of ncpus */
544 	int		zone_ncpus_online; /* zone's idea of ncpus_online */
545 	/*
546 	 * List of ZFS datasets exported to this zone.
547 	 */
548 	list_t		zone_datasets;	/* list of datasets */
549 
550 	ts_label_t	*zone_slabel;	/* zone sensitivity label */
551 	int		zone_match;	/* require label match for packets */
552 	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
553 
554 	boolean_t	zone_restart_init;	/* Restart init if it dies? */
555 	struct brand	*zone_brand;		/* zone's brand */
556 	void		*zone_brand_data;	/* store brand specific data */
557 	id_t		zone_defaultcid;	/* dflt scheduling class id */
558 	kstat_t		*zone_swapresv_kstat;
559 	kstat_t		*zone_lockedmem_kstat;
560 	/*
561 	 * zone_dl_list is protected by zone_lock
562 	 */
563 	list_t		zone_dl_list;
564 	netstack_t	*zone_netstack;
565 	struct cpucap	*zone_cpucap;	/* CPU caps data */
566 	/*
567 	 * Solaris Auditing per-zone audit context
568 	 */
569 	struct au_kcontext	*zone_audit_kctxt;
570 	/*
571 	 * For private use by mntfs.
572 	 */
573 	struct mntelem	*zone_mntfs_db;
574 	krwlock_t	zone_mntfs_db_lock;
575 
576 	struct klpd_reg		*zone_pfexecd;
577 
578 	char		*zone_fs_allowed;
579 	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
580 	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
581 						/* zone_rctls->rcs_lock */
582 	kstat_t		*zone_nprocs_kstat;
583 
584 	kmutex_t	zone_mcap_lock;	/* protects mcap statistics */
585 	kstat_t		*zone_mcap_ksp;
586 	zone_mcap_kstat_t *zone_mcap_stats;
587 	uint64_t	zone_pgpgin;		/* pages paged in */
588 	uint64_t	zone_anonpgin;		/* anon pages paged in */
589 	uint64_t	zone_execpgin;		/* exec pages paged in */
590 	uint64_t	zone_fspgin;		/* fs pages paged in */
591 	uint64_t	zone_anon_alloc_fail;	/* cnt of anon alloc fails */
592 
593 	psecflags_t	zone_secflags; /* default zone security-flags */
594 
595 	/*
596 	 * Misc. kstats and counters for zone cpu-usage aggregation.
597 	 */
598 	kmutex_t	zone_misc_lock;		/* protects misc statistics */
599 	kstat_t		*zone_misc_ksp;
600 	zone_misc_kstat_t *zone_misc_stats;
601 	/* Accumulated microstate for all threads in this zone. */
602 	cpu_uarray_t	*zone_ustate;
603 	/* fork-fail kstat tracking */
604 	uint32_t	zone_ffcap;		/* hit an rctl cap */
605 	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
606 	uint32_t	zone_ffnomem;		/* as_dup/memory error */
607 	uint32_t	zone_ffmisc;		/* misc. other error */
608 
609 	uint32_t	zone_nested_intp;	/* nested interp. kstat */
610 
611 	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
612 	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
613 	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
614 
615 	/*
616 	 * FSS stats updated once per second by fss_decay_usage.
617 	 */
618 	uint32_t	zone_fss_gen;		/* FSS generation cntr */
619 	uint64_t	zone_run_ticks;		/* tot # of ticks running */
620 
621 	/*
622 	 * DTrace-private per-zone state
623 	 */
624 	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
625 
626 	/*
627 	 * Synchronization primitives used to synchronize between mounts and
628 	 * zone creation/destruction.
629 	 */
630 	int		zone_mounts_in_progress;
631 	kcondvar_t	zone_mount_cv;
632 	kmutex_t	zone_mount_lock;
633 } zone_t;
634 
635 /*
636  * Special value of zone_psetid to indicate that pools are disabled.
637  */
638 #define	ZONE_PS_INVAL	PS_MYID
639 
640 
641 extern zone_t zone0;
642 extern zone_t *global_zone;
643 extern uint_t maxzones;
644 extern rctl_hndl_t rc_zone_nlwps;
645 extern rctl_hndl_t rc_zone_nprocs;
646 
647 extern long zone(int, void *, void *, void *, void *);
648 extern void zone_zsd_init(void);
649 extern void zone_init(void);
650 extern void zone_hold(zone_t *);
651 extern void zone_rele(zone_t *);
652 extern void zone_init_ref(zone_ref_t *);
653 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
654 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
655 extern void zone_cred_hold(zone_t *);
656 extern void zone_cred_rele(zone_t *);
657 extern void zone_task_hold(zone_t *);
658 extern void zone_task_rele(zone_t *);
659 extern zone_t *zone_find_by_id(zoneid_t);
660 extern zone_t *zone_find_by_label(const ts_label_t *);
661 extern zone_t *zone_find_by_name(char *);
662 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
663 extern zone_t *zone_find_by_path(const char *);
664 extern zoneid_t getzoneid(void);
665 extern zone_t *zone_find_by_id_nolock(zoneid_t);
666 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
667 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
668 extern void zone_loadavg_update();
669 
670 /*
671  * Zone-specific data (ZSD) APIs
672  */
673 /*
674  * The following is what code should be initializing its zone_key_t to if it
675  * calls zone_getspecific() without necessarily knowing that zone_key_create()
676  * has been called on the key.
677  */
678 #define	ZONE_KEY_UNINITIALIZED	0
679 
680 typedef uint_t zone_key_t;
681 
682 extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
683     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
684 extern int	zone_key_delete(zone_key_t);
685 extern void	*zone_getspecific(zone_key_t, zone_t *);
686 extern int	zone_setspecific(zone_key_t, zone_t *, const void *);
687 
688 /*
689  * The definition of a zsd_entry is truly private to zone.c and is only
690  * placed here so it can be shared with mdb.
691  *
692  * State maintained for each zone times each registered key, which tracks
693  * the state of the create, shutdown and destroy callbacks.
694  *
695  * zsd_flags is used to keep track of pending actions to avoid holding locks
696  * when calling the create/shutdown/destroy callbacks, since doing so
697  * could lead to deadlocks.
698  */
699 struct zsd_entry {
700 	zone_key_t		zsd_key;	/* Key used to lookup value */
701 	void			*zsd_data;	/* Caller-managed value */
702 	/*
703 	 * Callbacks to be executed when a zone is created, shutdown, and
704 	 * destroyed, respectively.
705 	 */
706 	void			*(*zsd_create)(zoneid_t);
707 	void			(*zsd_shutdown)(zoneid_t, void *);
708 	void			(*zsd_destroy)(zoneid_t, void *);
709 	list_node_t		zsd_linkage;
710 	uint16_t		zsd_flags;	/* See below */
711 	kcondvar_t		zsd_cv;
712 };
713 
714 /*
715  * zsd_flags
716  */
717 #define	ZSD_CREATE_NEEDED	0x0001
718 #define	ZSD_CREATE_INPROGRESS	0x0002
719 #define	ZSD_CREATE_COMPLETED	0x0004
720 #define	ZSD_SHUTDOWN_NEEDED	0x0010
721 #define	ZSD_SHUTDOWN_INPROGRESS	0x0020
722 #define	ZSD_SHUTDOWN_COMPLETED	0x0040
723 #define	ZSD_DESTROY_NEEDED	0x0100
724 #define	ZSD_DESTROY_INPROGRESS	0x0200
725 #define	ZSD_DESTROY_COMPLETED	0x0400
726 
727 #define	ZSD_CREATE_ALL	\
728 	(ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
729 #define	ZSD_SHUTDOWN_ALL	\
730 	(ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
731 #define	ZSD_DESTROY_ALL	\
732 	(ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
733 
734 #define	ZSD_ALL_INPROGRESS \
735 	(ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
736 
737 /*
738  * Macros to help with zone visibility restrictions.
739  */
740 
741 /*
742  * Is process in the global zone?
743  */
744 #define	INGLOBALZONE(p) \
745 	((p)->p_zone == global_zone)
746 
747 /*
748  * Can process view objects in given zone?
749  */
750 #define	HASZONEACCESS(p, zoneid) \
751 	((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
752 
753 /*
754  * Convenience macro to see if a resolved path is visible from within a
755  * given zone.
756  *
757  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
758  * two strings must be equal.  Since the rootpathlen has a trailing '/',
759  * we want to skip everything in the path up to (but not including) the
760  * trailing '/'.
761  */
762 #define	ZONE_PATH_VISIBLE(path, zone) \
763 	(strncmp((path), (zone)->zone_rootpath,		\
764 	    (zone)->zone_rootpathlen - 1) == 0)
765 
766 /*
767  * Convenience macro to go from the global view of a path to that seen
768  * from within said zone.  It is the responsibility of the caller to
769  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
770  * in fact visible from within the zone.
771  */
772 #define	ZONE_PATH_TRANSLATE(path, zone)	\
773 	(ASSERT(ZONE_PATH_VISIBLE(path, zone)),	\
774 	(path) + (zone)->zone_rootpathlen - 2)
775 
776 /*
777  * Special processes visible in all zones.
778  */
779 #define	ZONE_SPECIALPID(x)	 ((x) == 0 || (x) == 1)
780 
781 /*
782  * Zone-safe version of thread_create() to be used when the caller wants to
783  * create a kernel thread to run within the current zone's context.
784  */
785 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
786     pri_t);
787 extern void zthread_exit(void);
788 
789 /*
790  * Functions for an external observer to register interest in a zone's status
791  * change.  Observers will be woken up when the zone status equals the status
792  * argument passed in (in the case of zone_status_timedwait, the function may
793  * also return because of a timeout; zone_status_wait_sig may return early due
794  * to a signal being delivered; zone_status_timedwait_sig may return for any of
795  * the above reasons).
796  *
797  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
798  * cv_wait_sig() respectively.
799  */
800 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
801 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
802 extern void zone_status_wait(zone_t *, zone_status_t);
803 extern int zone_status_wait_sig(zone_t *, zone_status_t);
804 
805 /*
806  * Get the status  of the zone (at the time it was called).  The state may
807  * have progressed by the time it is returned.
808  */
809 extern zone_status_t zone_status_get(zone_t *);
810 
811 /*
812  * Safely get the hostid of the specified zone (defaults to machine's hostid
813  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
814  * the global zone's (i.e., physical system's) hostid.
815  */
816 extern uint32_t zone_get_hostid(zone_t *);
817 
818 /*
819  * Get the "kcred" credentials corresponding to the given zone.
820  */
821 extern struct cred *zone_get_kcred(zoneid_t);
822 
823 /*
824  * Get/set the pool the zone is currently bound to.
825  */
826 extern struct pool *zone_pool_get(zone_t *);
827 extern void zone_pool_set(zone_t *, struct pool *);
828 
829 /*
830  * Get/set the pset the zone is currently using.
831  */
832 extern psetid_t zone_pset_get(zone_t *);
833 extern void zone_pset_set(zone_t *, psetid_t);
834 
835 /*
836  * Get the number of cpus/online-cpus visible from the given zone.
837  */
838 extern int zone_ncpus_get(zone_t *);
839 extern int zone_ncpus_online_get(zone_t *);
840 
841 /*
842  * Returns true if the named pool/dataset is visible in the current zone.
843  */
844 extern int zone_dataset_visible(const char *, int *);
845 
846 /*
847  * zone version of kadmin()
848  */
849 extern int zone_kadmin(int, int, const char *, cred_t *);
850 extern void zone_shutdown_global(void);
851 
852 extern void mount_in_progress(zone_t *);
853 extern void mount_completed(zone_t *);
854 
855 extern int zone_walk(int (*)(zone_t *, void *), void *);
856 
857 extern rctl_hndl_t rc_zone_locked_mem;
858 extern rctl_hndl_t rc_zone_max_swap;
859 extern rctl_hndl_t rc_zone_max_lofi;
860 
861 #endif	/* _KERNEL */
862 
863 #ifdef	__cplusplus
864 }
865 #endif
866 
867 #endif	/* _SYS_ZONE_H */
868