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/*
23 * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
25 * Copyright 2019 Nexenta Systems, Inc. All rights reserved.
26 * Copyright 2020 Joyent, Inc.
27 */
28
29#ifndef _SYS_ZONE_H
30#define	_SYS_ZONE_H
31
32#include <sys/types.h>
33#include <sys/param.h>
34#include <sys/tsol/label.h>
35#include <sys/uadmin.h>
36#include <netinet/in.h>
37
38#ifdef _KERNEL
39/*
40 * Many includes are kernel-only to reduce namespace pollution of
41 * userland applications.
42 */
43#include <sys/mutex.h>
44#include <sys/rctl.h>
45#include <sys/ipc_rctl.h>
46#include <sys/pset.h>
47#include <sys/cred.h>
48#include <sys/netstack.h>
49#include <sys/ksynch.h>
50#include <sys/socket_impl.h>
51#include <sys/secflags.h>
52#include <sys/cpu_uarray.h>
53#include <sys/list.h>
54#include <sys/loadavg.h>
55#endif	/* _KERNEL */
56
57#ifdef	__cplusplus
58extern "C" {
59#endif
60
61/*
62 * NOTE
63 *
64 * The contents of this file are private to the implementation of
65 * Solaris and are subject to change at any time without notice.
66 * Applications and drivers using these interfaces may fail to
67 * run on future releases.
68 */
69
70/* Available both in kernel and for user space */
71
72/* zone id restrictions and special ids */
73#define	MAX_ZONEID	9999
74#define	MIN_USERZONEID	1	/* lowest user-creatable zone ID */
75#define	MIN_ZONEID	0	/* minimum zone ID on system */
76#define	GLOBAL_ZONEID	0
77#define	ZONEID_WIDTH	4	/* for printf */
78
79/*
80 * Special zoneid_t token to refer to all zones.
81 */
82#define	ALL_ZONES	(-1)
83
84/* system call subcodes */
85#define	ZONE_CREATE		0
86#define	ZONE_DESTROY		1
87#define	ZONE_GETATTR		2
88#define	ZONE_ENTER		3
89#define	ZONE_LIST		4
90#define	ZONE_SHUTDOWN		5
91#define	ZONE_LOOKUP		6
92#define	ZONE_BOOT		7
93#define	ZONE_VERSION		8
94#define	ZONE_SETATTR		9
95#define	ZONE_ADD_DATALINK	10
96#define	ZONE_DEL_DATALINK	11
97#define	ZONE_CHECK_DATALINK	12
98#define	ZONE_LIST_DATALINK	13
99
100/* zone attributes */
101#define	ZONE_ATTR_ROOT		1
102#define	ZONE_ATTR_NAME		2
103#define	ZONE_ATTR_STATUS	3
104#define	ZONE_ATTR_PRIVSET	4
105#define	ZONE_ATTR_UNIQID	5
106#define	ZONE_ATTR_POOLID	6
107#define	ZONE_ATTR_INITPID	7
108#define	ZONE_ATTR_SLBL		8
109#define	ZONE_ATTR_INITNAME	9
110#define	ZONE_ATTR_BOOTARGS	10
111#define	ZONE_ATTR_BRAND		11
112#define	ZONE_ATTR_PHYS_MCAP	12
113#define	ZONE_ATTR_SCHED_CLASS	13
114#define	ZONE_ATTR_FLAGS		14
115#define	ZONE_ATTR_HOSTID	15
116#define	ZONE_ATTR_FS_ALLOWED	16
117#define	ZONE_ATTR_NETWORK	17
118#define	ZONE_ATTR_INITNORESTART	20
119#define	ZONE_ATTR_SECFLAGS	21
120
121/* Start of the brand-specific attribute namespace */
122#define	ZONE_ATTR_BRAND_ATTRS	32768
123
124#define	ZONE_FS_ALLOWED_MAX	1024
125
126#define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
127#define	ZONE_EVENT_STATUS_CLASS	"status"
128#define	ZONE_EVENT_STATUS_SUBCLASS	"change"
129
130#define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
131#define	ZONE_EVENT_INITIALIZED		"initialized"
132#define	ZONE_EVENT_READY		"ready"
133#define	ZONE_EVENT_RUNNING		"running"
134#define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
135
136#define	ZONE_CB_NAME		"zonename"
137#define	ZONE_CB_NEWSTATE	"newstate"
138#define	ZONE_CB_OLDSTATE	"oldstate"
139#define	ZONE_CB_TIMESTAMP	"when"
140#define	ZONE_CB_ZONEID		"zoneid"
141
142/*
143 * Exit values that may be returned by scripts or programs invoked by various
144 * zone commands.
145 *
146 * These are defined as:
147 *
148 *	ZONE_SUBPROC_OK
149 *	===============
150 *	The subprocess completed successfully.
151 *
152 *	ZONE_SUBPROC_USAGE
153 *	==================
154 *	The subprocess failed with a usage message, or a usage message should
155 *	be output in its behalf.
156 *
157 *	ZONE_SUBPROC_NOTCOMPLETE
158 *	========================
159 *	The subprocess did not complete, but the actions performed by the
160 *	subprocess require no recovery actions by the user.
161 *
162 *	For example, if the subprocess were called by "zoneadm install," the
163 *	installation of the zone did not succeed but the user need not perform
164 *	a "zoneadm uninstall" before attempting another install.
165 *
166 *	ZONE_SUBPROC_FATAL
167 *	==================
168 *	The subprocess failed in a fatal manner, usually one that will require
169 *	some type of recovery action by the user.
170 *
171 *	For example, if the subprocess were called by "zoneadm install," the
172 *	installation of the zone did not succeed and the user will need to
173 *	perform a "zoneadm uninstall" before another install attempt is
174 *	possible.
175 *
176 *	The non-success exit values are large to avoid accidental collision
177 *	with values used internally by some commands (e.g. "Z_ERR" and
178 *	"Z_USAGE" as used by zoneadm.)
179 */
180#define	ZONE_SUBPROC_OK			0
181#define	ZONE_SUBPROC_USAGE		253
182#define	ZONE_SUBPROC_NOTCOMPLETE	254
183#define	ZONE_SUBPROC_FATAL		255
184
185#ifdef _SYSCALL32
186typedef struct {
187	caddr32_t zone_name;
188	caddr32_t zone_root;
189	caddr32_t zone_privs;
190	size32_t zone_privssz;
191	caddr32_t rctlbuf;
192	size32_t rctlbufsz;
193	caddr32_t extended_error;
194	caddr32_t zfsbuf;
195	size32_t  zfsbufsz;
196	int match;			/* match level */
197	uint32_t doi;			/* DOI for label */
198	caddr32_t label;		/* label associated with zone */
199	int flags;
200} zone_def32;
201#endif
202typedef struct {
203	const char *zone_name;
204	const char *zone_root;
205	const struct priv_set *zone_privs;
206	size_t zone_privssz;
207	const char *rctlbuf;
208	size_t rctlbufsz;
209	int *extended_error;
210	const char *zfsbuf;
211	size_t zfsbufsz;
212	int match;			/* match level */
213	uint32_t doi;			/* DOI for label */
214	const bslabel_t *label;		/* label associated with zone */
215	int flags;
216} zone_def;
217
218/* extended error information */
219#define	ZE_UNKNOWN	0	/* No extended error info */
220#define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
221#define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
222#define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
223
224/*
225 * zone_status values
226 *
227 * You must modify zone_status_names in mdb(1M)'s genunix module
228 * (genunix/zone.c) when you modify this enum.
229 */
230typedef enum {
231	ZONE_IS_UNINITIALIZED = 0,
232	ZONE_IS_INITIALIZED,
233	ZONE_IS_READY,
234	ZONE_IS_BOOTING,
235	ZONE_IS_RUNNING,
236	ZONE_IS_SHUTTING_DOWN,
237	ZONE_IS_EMPTY,
238	ZONE_IS_DOWN,
239	ZONE_IS_DYING,
240	ZONE_IS_DEAD
241} zone_status_t;
242#define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
243#define	ZONE_MAX_STATE		ZONE_IS_DEAD
244
245/*
246 * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
247 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
248 */
249typedef enum zone_cmd {
250	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
251	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
252} zone_cmd_t;
253
254/*
255 * The structure of a request to zoneadmd.
256 */
257typedef struct zone_cmd_arg {
258	uint64_t	uniqid;		/* unique "generation number" */
259	zone_cmd_t	cmd;		/* requested action */
260	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
261	char locale[MAXPATHLEN];	/* locale in which to render messages */
262	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
263} zone_cmd_arg_t;
264
265/*
266 * Structure of zoneadmd's response to a request.  A NULL return value means
267 * the caller should attempt to restart zoneadmd and retry.
268 */
269typedef struct zone_cmd_rval {
270	int rval;			/* return value of request */
271	char errbuf[1];	/* variable-sized buffer containing error messages */
272} zone_cmd_rval_t;
273
274/*
275 * The zone support infrastructure uses the zone name as a component
276 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
277 * in length, so ZONENAME_MAX is limited by that.
278 */
279#define	ZONENAME_MAX		64
280
281#define	GLOBAL_ZONENAME		"global"
282
283/*
284 * Extended Regular expression (see regex(5)) which matches all valid zone
285 * names.
286 */
287#define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
288
289/*
290 * Where the zones support infrastructure places temporary files.
291 */
292#define	ZONES_TMPDIR		"/var/run/zones"
293
294/*
295 * The path to the door used by clients to communicate with zoneadmd.
296 */
297#define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
298
299
300/* zone_flags */
301/*
302 * Threads that read or write the following flag must hold zone_lock.
303 */
304#define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
305
306/*
307 * The following threads are set when the zone is created and never changed.
308 * Threads that test for these flags don't have to hold zone_lock.
309 */
310#define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
311#define	ZF_IS_SCRATCH		0x4	/* scratch zone */
312#define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
313
314
315/* zone_create flags */
316#define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
317
318/* zone network properties */
319#define	ZONE_NETWORK_ADDRESS	1
320#define	ZONE_NETWORK_DEFROUTER	2
321
322#define	ZONE_NET_ADDRNAME	"address"
323#define	ZONE_NET_RTRNAME	"route"
324
325typedef struct zone_net_data {
326	int zn_type;
327	int zn_len;
328	datalink_id_t zn_linkid;
329	uint8_t zn_val[1];
330} zone_net_data_t;
331
332
333#ifdef _KERNEL
334
335#define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
336
337/*
338 * Indexes into ->zone_ustate array, summing the micro state of all threads in a
339 * particular zone.
340 */
341#define	ZONE_USTATE_STIME (0)
342#define	ZONE_USTATE_UTIME (1)
343#define	ZONE_USTATE_WTIME (2)
344#define	ZONE_USTATE_MAX (3)
345
346struct pool;
347struct brand;
348
349/*
350 * Each of these constants identifies a kernel subsystem that acquires and
351 * releases zone references.  Each subsystem that invokes
352 * zone_hold_ref() and zone_rele_ref() should specify the
353 * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
354 * help users and developers quickly identify subsystems that stall zone
355 * shutdowns indefinitely.
356 *
357 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
358 * when you modify this enumeration.
359 */
360typedef enum zone_ref_subsys {
361	ZONE_REF_NFS,			/* NFS */
362	ZONE_REF_NFSV4,			/* NFSv4 */
363	ZONE_REF_SMBFS,			/* SMBFS */
364	ZONE_REF_MNTFS,			/* MNTFS */
365	ZONE_REF_LOFI,			/* LOFI devices */
366	ZONE_REF_VFS,			/* VFS infrastructure */
367	ZONE_REF_IPC,			/* IPC infrastructure */
368	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
369} zone_ref_subsys_t;
370
371/*
372 * zone_ref represents a general-purpose references to a zone.  Each zone's
373 * references are linked into the zone's zone_t::zone_ref_list.  This allows
374 * debuggers to walk zones' references.
375 */
376typedef struct zone_ref {
377	struct zone	*zref_zone; /* the zone to which the reference refers */
378	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
379} zone_ref_t;
380
381/*
382 * Structure to record list of ZFS datasets exported to a zone.
383 */
384typedef struct zone_dataset {
385	char		*zd_dataset;
386	list_node_t	zd_linkage;
387} zone_dataset_t;
388
389/*
390 * structure for zone kstats
391 */
392typedef struct zone_kstat {
393	kstat_named_t zk_zonename;
394	kstat_named_t zk_usage;
395	kstat_named_t zk_value;
396} zone_kstat_t;
397
398struct cpucap;
399
400typedef struct {
401	kstat_named_t	zm_zonename;
402	kstat_named_t	zm_pgpgin;
403	kstat_named_t	zm_anonpgin;
404	kstat_named_t	zm_execpgin;
405	kstat_named_t	zm_fspgin;
406	kstat_named_t	zm_anon_alloc_fail;
407} zone_mcap_kstat_t;
408
409typedef struct {
410	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
411	kstat_named_t	zm_utime;
412	kstat_named_t	zm_stime;
413	kstat_named_t	zm_wtime;
414	kstat_named_t	zm_avenrun1;
415	kstat_named_t	zm_avenrun5;
416	kstat_named_t	zm_avenrun15;
417	kstat_named_t	zm_ffcap;
418	kstat_named_t	zm_ffnoproc;
419	kstat_named_t	zm_ffnomem;
420	kstat_named_t	zm_ffmisc;
421	kstat_named_t	zm_nested_intp;
422	kstat_named_t	zm_init_pid;
423	kstat_named_t	zm_boot_time;
424} zone_misc_kstat_t;
425
426typedef struct zone {
427	/*
428	 * zone_name is never modified once set.
429	 */
430	char		*zone_name;	/* zone's configuration name */
431	/*
432	 * zone_nodename and zone_domain are never freed once allocated.
433	 */
434	char		*zone_nodename;	/* utsname.nodename equivalent */
435	char		*zone_domain;	/* srpc_domain equivalent */
436	/*
437	 * zone_hostid is used for per-zone hostid emulation.
438	 * Currently it isn't modified after it's set (so no locks protect
439	 * accesses), but that might have to change when we allow
440	 * administrators to change running zones' properties.
441	 *
442	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
443	 * that zone_get_hostid() will function correctly.
444	 */
445	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
446					/* if not emulated */
447	/*
448	 * zone_lock protects the following fields of a zone_t:
449	 *	zone_ref
450	 *	zone_cred_ref
451	 *	zone_subsys_ref
452	 *	zone_ref_list
453	 *	zone_ntasks
454	 *	zone_flags
455	 *	zone_zsd
456	 *	zone_pfexecd
457	 */
458	kmutex_t	zone_lock;
459	/*
460	 * zone_linkage is the zone's linkage into the active or
461	 * death-row list.  The field is protected by zonehash_lock.
462	 */
463	list_node_t	zone_linkage;
464	zoneid_t	zone_id;	/* ID of zone */
465	uint_t		zone_ref;	/* count of zone_hold()s on zone */
466	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
467	/*
468	 * Fixed-sized array of subsystem-specific reference counts
469	 * The sum of all of the counts must be less than or equal to zone_ref.
470	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
471	 * constants.
472	 */
473	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
474	list_t		zone_ref_list;	/* list of zone_ref_t structs */
475	/*
476	 * zone_rootvp and zone_rootpath can never be modified once set.
477	 */
478	struct vnode	*zone_rootvp;	/* zone's root vnode */
479	char		*zone_rootpath;	/* Path to zone's root + '/' */
480	ushort_t	zone_flags;	/* misc flags */
481	zone_status_t	zone_status;	/* protected by zone_status_lock */
482	uint_t		zone_ntasks;	/* number of tasks executing in zone */
483	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
484					/* counters in projects and tasks */
485					/* that are within the zone */
486	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
487	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
488	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
489	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
490
491	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
492	uint32_t	zone_shares;	/* FSS shares allocated to zone */
493	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
494	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
495					/* kpd_locked_mem for all */
496					/* projects in zone. */
497					/* Also protects zone_max_swap */
498					/* grab after p_lock, before rcs_lock */
499	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
500						/* zone */
501	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
502						/* limit.  Protected by */
503						/* zone_rctls->rcs_lock */
504	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
505	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
506						/* Protected by */
507						/* zone_rctls->rcs_lock */
508	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
509	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
510	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
511						/* Protected by */
512						/* zone_rctls->rcs_lock */
513	list_t		zone_zsd;	/* list of Zone-Specific Data values */
514	kcondvar_t	zone_cv;	/* used to signal state changes */
515	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
516	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
517	char		*zone_initname;	/* fs path to 'init' */
518	int		zone_boot_err;  /* for zone_boot() if boot fails */
519	char		*zone_bootargs;	/* arguments passed via zone_boot() */
520	uint64_t	zone_phys_mcap;	/* physical memory cap */
521	/*
522	 * zone_kthreads is protected by zone_status_lock.
523	 */
524	kthread_t	*zone_kthreads;	/* kernel threads in zone */
525	struct priv_set	*zone_privset;	/* limit set for zone */
526	/*
527	 * zone_vfslist is protected by vfs_list_lock().
528	 */
529	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
530	uint64_t	zone_uniqid;	/* unique zone generation number */
531	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
532	/*
533	 * zone_pool is protected by pool_lock().
534	 */
535	struct pool	*zone_pool;	/* pool the zone is bound to */
536	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
537	/* zone_psetid is protected by cpu_lock */
538	psetid_t	zone_psetid;	/* pset the zone is bound to */
539
540	time_t		zone_boot_time;	/* Similar to boot_time */
541
542	/*
543	 * The following two can be read without holding any locks.  They are
544	 * updated under cpu_lock.
545	 */
546	int		zone_ncpus;  /* zone's idea of ncpus */
547	int		zone_ncpus_online; /* zone's idea of ncpus_online */
548	/*
549	 * List of ZFS datasets exported to this zone.
550	 */
551	list_t		zone_datasets;	/* list of datasets */
552
553	ts_label_t	*zone_slabel;	/* zone sensitivity label */
554	int		zone_match;	/* require label match for packets */
555	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
556
557	boolean_t	zone_restart_init;	/* Restart init if it dies? */
558	struct brand	*zone_brand;		/* zone's brand */
559	void		*zone_brand_data;	/* store brand specific data */
560	id_t		zone_defaultcid;	/* dflt scheduling class id */
561	kstat_t		*zone_swapresv_kstat;
562	kstat_t		*zone_lockedmem_kstat;
563	/*
564	 * zone_dl_list is protected by zone_lock
565	 */
566	list_t		zone_dl_list;
567	netstack_t	*zone_netstack;
568	struct cpucap	*zone_cpucap;	/* CPU caps data */
569	/*
570	 * Solaris Auditing per-zone audit context
571	 */
572	struct au_kcontext	*zone_audit_kctxt;
573	/*
574	 * For private use by mntfs.
575	 */
576	struct mntelem	*zone_mntfs_db;
577	krwlock_t	zone_mntfs_db_lock;
578
579	struct klpd_reg		*zone_pfexecd;
580
581	char		*zone_fs_allowed;
582	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
583	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
584						/* zone_rctls->rcs_lock */
585	kstat_t		*zone_nprocs_kstat;
586
587	kmutex_t	zone_mcap_lock;	/* protects mcap statistics */
588	kstat_t		*zone_mcap_ksp;
589	zone_mcap_kstat_t *zone_mcap_stats;
590	uint64_t	zone_pgpgin;		/* pages paged in */
591	uint64_t	zone_anonpgin;		/* anon pages paged in */
592	uint64_t	zone_execpgin;		/* exec pages paged in */
593	uint64_t	zone_fspgin;		/* fs pages paged in */
594	uint64_t	zone_anon_alloc_fail;	/* cnt of anon alloc fails */
595
596	psecflags_t	zone_secflags; /* default zone security-flags */
597
598	/*
599	 * Misc. kstats and counters for zone cpu-usage aggregation.
600	 */
601	kmutex_t	zone_misc_lock;		/* protects misc statistics */
602	kstat_t		*zone_misc_ksp;
603	zone_misc_kstat_t *zone_misc_stats;
604	/* Accumulated microstate for all threads in this zone. */
605	cpu_uarray_t	*zone_ustate;
606	/* fork-fail kstat tracking */
607	uint32_t	zone_ffcap;		/* hit an rctl cap */
608	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
609	uint32_t	zone_ffnomem;		/* as_dup/memory error */
610	uint32_t	zone_ffmisc;		/* misc. other error */
611
612	uint32_t	zone_nested_intp;	/* nested interp. kstat */
613
614	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
615	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
616	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
617
618	/*
619	 * FSS stats updated once per second by fss_decay_usage.
620	 */
621	uint32_t	zone_fss_gen;		/* FSS generation cntr */
622	uint64_t	zone_run_ticks;		/* tot # of ticks running */
623
624	/*
625	 * DTrace-private per-zone state
626	 */
627	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
628
629	/*
630	 * Synchronization primitives used to synchronize between mounts and
631	 * zone creation/destruction.
632	 */
633	int		zone_mounts_in_progress;
634	kcondvar_t	zone_mount_cv;
635	kmutex_t	zone_mount_lock;
636} zone_t;
637
638/*
639 * Special value of zone_psetid to indicate that pools are disabled.
640 */
641#define	ZONE_PS_INVAL	PS_MYID
642
643extern zone_t zone0;
644extern zone_t *global_zone;
645extern uint_t maxzones;
646extern rctl_hndl_t rc_zone_nlwps;
647extern rctl_hndl_t rc_zone_nprocs;
648
649extern long zone(int, void *, void *, void *, void *);
650extern void zone_zsd_init(void);
651extern void zone_init(void);
652extern void zone_hold(zone_t *);
653extern void zone_rele(zone_t *);
654extern void zone_init_ref(zone_ref_t *);
655extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
656extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
657extern void zone_cred_hold(zone_t *);
658extern void zone_cred_rele(zone_t *);
659extern void zone_task_hold(zone_t *);
660extern void zone_task_rele(zone_t *);
661extern zone_t *zone_find_by_id(zoneid_t);
662extern zone_t *zone_find_by_label(const ts_label_t *);
663extern zone_t *zone_find_by_name(char *);
664extern zone_t *zone_find_by_any_path(const char *, boolean_t);
665extern zone_t *zone_find_by_path(const char *);
666extern zoneid_t getzoneid(void);
667extern zone_t *zone_find_by_id_nolock(zoneid_t);
668extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
669extern int zone_check_datalink(zoneid_t *, datalink_id_t);
670extern 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
682typedef uint_t zone_key_t;
683
684extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
685    void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
686extern int	zone_key_delete(zone_key_t);
687extern void	*zone_getspecific(zone_key_t, zone_t *);
688extern 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 */
701struct 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 * A root vnode of the current zone.
785 *
786 * NOTE: It may be necessary (initialization time for file sharing where an
787 * NGZ loads a file-sharing kernel module that does zsd initialization) to NOT
788 * use this macro. One should ASSERT() that curzone == active ZSD (an
789 * ASSERTion that's not always true at ZSD initialization time) during regular
790 * use of this macro.
791 */
792#define	ZONE_ROOTVP()	(curzone->zone_rootvp)
793
794/*
795 * Since a zone's root isn't necessarily an actual filesystem boundary
796 * (i.e. VROOT may not be set on zone->zone_rootvp) we need to not assume it.
797 * This macro helps in checking if a vnode is the current zone's rootvp.
798 * NOTE:  Using the VN_ prefix, even though it's defined here in zone.h.
799 * NOTE2: See above warning about ZONE_ROOTVP().
800 */
801#define	VN_IS_CURZONEROOT(vp)   (VN_CMP(vp, ZONE_ROOTVP()))
802
803/*
804 * Zone-safe version of thread_create() to be used when the caller wants to
805 * create a kernel thread to run within the current zone's context.
806 */
807extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
808    pri_t);
809extern void zthread_exit(void);
810
811/*
812 * Functions for an external observer to register interest in a zone's status
813 * change.  Observers will be woken up when the zone status equals the status
814 * argument passed in (in the case of zone_status_timedwait, the function may
815 * also return because of a timeout; zone_status_wait_sig may return early due
816 * to a signal being delivered; zone_status_timedwait_sig may return for any of
817 * the above reasons).
818 *
819 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
820 * cv_wait_sig() respectively.
821 */
822extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
823extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
824extern void zone_status_wait(zone_t *, zone_status_t);
825extern int zone_status_wait_sig(zone_t *, zone_status_t);
826
827/*
828 * Get the status  of the zone (at the time it was called).  The state may
829 * have progressed by the time it is returned.
830 */
831extern zone_status_t zone_status_get(zone_t *);
832
833/*
834 * Safely get the hostid of the specified zone (defaults to machine's hostid
835 * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
836 * the global zone's (i.e., physical system's) hostid.
837 */
838extern uint32_t zone_get_hostid(zone_t *);
839
840/*
841 * Get the "kcred" credentials corresponding to the given zone.
842 */
843extern struct cred *zone_get_kcred(zoneid_t);
844
845/*
846 * Get/set the pool the zone is currently bound to.
847 */
848extern struct pool *zone_pool_get(zone_t *);
849extern void zone_pool_set(zone_t *, struct pool *);
850
851/*
852 * Get/set the pset the zone is currently using.
853 */
854extern psetid_t zone_pset_get(zone_t *);
855extern void zone_pset_set(zone_t *, psetid_t);
856
857/*
858 * Get the number of cpus/online-cpus visible from the given zone.
859 */
860extern int zone_ncpus_get(zone_t *);
861extern int zone_ncpus_online_get(zone_t *);
862
863/*
864 * Returns true if the named pool/dataset is visible in the current zone.
865 */
866extern int zone_dataset_visible(const char *, int *);
867
868/*
869 * zone version of kadmin()
870 */
871extern int zone_kadmin(int, int, const char *, cred_t *);
872extern void zone_shutdown_global(void);
873
874extern void mount_in_progress(zone_t *);
875extern void mount_completed(zone_t *);
876
877extern int zone_walk(int (*)(zone_t *, void *), void *);
878
879extern rctl_hndl_t rc_zone_locked_mem;
880extern rctl_hndl_t rc_zone_max_swap;
881extern rctl_hndl_t rc_zone_max_lofi;
882
883#endif	/* _KERNEL */
884
885#ifdef	__cplusplus
886}
887#endif
888
889#endif	/* _SYS_ZONE_H */
890