xref: /illumos-gate/usr/src/uts/common/sys/zone.h (revision f3682895b2a97c009685f16e8a4e5d3dc80e11f2)
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
58 extern "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 #define	ZONE_ATTR_INITRESTART0	22
121 #define	ZONE_ATTR_INITREBOOT	23
122 
123 /* Start of the brand-specific attribute namespace */
124 #define	ZONE_ATTR_BRAND_ATTRS	32768
125 
126 #define	ZONE_FS_ALLOWED_MAX	1024
127 
128 #define	ZONE_EVENT_CHANNEL	"com.sun:zones:status"
129 #define	ZONE_EVENT_STATUS_CLASS	"status"
130 #define	ZONE_EVENT_STATUS_SUBCLASS	"change"
131 
132 #define	ZONE_EVENT_UNINITIALIZED	"uninitialized"
133 #define	ZONE_EVENT_INITIALIZED		"initialized"
134 #define	ZONE_EVENT_READY		"ready"
135 #define	ZONE_EVENT_RUNNING		"running"
136 #define	ZONE_EVENT_SHUTTING_DOWN	"shutting_down"
137 
138 #define	ZONE_CB_NAME		"zonename"
139 #define	ZONE_CB_NEWSTATE	"newstate"
140 #define	ZONE_CB_OLDSTATE	"oldstate"
141 #define	ZONE_CB_TIMESTAMP	"when"
142 #define	ZONE_CB_ZONEID		"zoneid"
143 
144 /*
145  * Exit values that may be returned by scripts or programs invoked by various
146  * zone commands.
147  *
148  * These are defined as:
149  *
150  *	ZONE_SUBPROC_OK
151  *	===============
152  *	The subprocess completed successfully.
153  *
154  *	ZONE_SUBPROC_USAGE
155  *	==================
156  *	The subprocess failed with a usage message, or a usage message should
157  *	be output in its behalf.
158  *
159  *	ZONE_SUBPROC_NOTCOMPLETE
160  *	========================
161  *	The subprocess did not complete, but the actions performed by the
162  *	subprocess require no recovery actions by the user.
163  *
164  *	For example, if the subprocess were called by "zoneadm install," the
165  *	installation of the zone did not succeed but the user need not perform
166  *	a "zoneadm uninstall" before attempting another install.
167  *
168  *	ZONE_SUBPROC_FATAL
169  *	==================
170  *	The subprocess failed in a fatal manner, usually one that will require
171  *	some type of recovery action by the user.
172  *
173  *	For example, if the subprocess were called by "zoneadm install," the
174  *	installation of the zone did not succeed and the user will need to
175  *	perform a "zoneadm uninstall" before another install attempt is
176  *	possible.
177  *
178  *	The non-success exit values are large to avoid accidental collision
179  *	with values used internally by some commands (e.g. "Z_ERR" and
180  *	"Z_USAGE" as used by zoneadm.)
181  */
182 #define	ZONE_SUBPROC_OK			0
183 #define	ZONE_SUBPROC_USAGE		253
184 #define	ZONE_SUBPROC_NOTCOMPLETE	254
185 #define	ZONE_SUBPROC_FATAL		255
186 
187 #ifdef _SYSCALL32
188 typedef struct {
189 	caddr32_t zone_name;
190 	caddr32_t zone_root;
191 	caddr32_t zone_privs;
192 	size32_t zone_privssz;
193 	caddr32_t rctlbuf;
194 	size32_t rctlbufsz;
195 	caddr32_t extended_error;
196 	caddr32_t zfsbuf;
197 	size32_t  zfsbufsz;
198 	int match;			/* match level */
199 	uint32_t doi;			/* DOI for label */
200 	caddr32_t label;		/* label associated with zone */
201 	int flags;
202 } zone_def32;
203 #endif
204 typedef struct {
205 	const char *zone_name;
206 	const char *zone_root;
207 	const struct priv_set *zone_privs;
208 	size_t zone_privssz;
209 	const char *rctlbuf;
210 	size_t rctlbufsz;
211 	int *extended_error;
212 	const char *zfsbuf;
213 	size_t zfsbufsz;
214 	int match;			/* match level */
215 	uint32_t doi;			/* DOI for label */
216 	const bslabel_t *label;		/* label associated with zone */
217 	int flags;
218 } zone_def;
219 
220 /* extended error information */
221 #define	ZE_UNKNOWN	0	/* No extended error info */
222 #define	ZE_CHROOTED	1	/* tried to zone_create from chroot */
223 #define	ZE_AREMOUNTS	2	/* there are mounts within the zone */
224 #define	ZE_LABELINUSE	3	/* label is already in use by some other zone */
225 
226 /*
227  * zone_status values
228  *
229  * You must modify zone_status_names in mdb(1M)'s genunix module
230  * (genunix/zone.c) when you modify this enum.
231  */
232 typedef enum {
233 	ZONE_IS_UNINITIALIZED = 0,
234 	ZONE_IS_INITIALIZED,
235 	ZONE_IS_READY,
236 	ZONE_IS_BOOTING,
237 	ZONE_IS_RUNNING,
238 	ZONE_IS_SHUTTING_DOWN,
239 	ZONE_IS_EMPTY,
240 	ZONE_IS_DOWN,
241 	ZONE_IS_DYING,
242 	ZONE_IS_DEAD
243 } zone_status_t;
244 #define	ZONE_MIN_STATE		ZONE_IS_UNINITIALIZED
245 #define	ZONE_MAX_STATE		ZONE_IS_DEAD
246 
247 /*
248  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
249  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
250  */
251 typedef enum zone_cmd {
252 	Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
253 	Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
254 } zone_cmd_t;
255 
256 /*
257  * The structure of a request to zoneadmd.
258  */
259 typedef struct zone_cmd_arg {
260 	uint64_t	uniqid;		/* unique "generation number" */
261 	zone_cmd_t	cmd;		/* requested action */
262 	uint32_t	_pad;		/* need consistent 32/64 bit alignmt */
263 	char locale[MAXPATHLEN];	/* locale in which to render messages */
264 	char bootbuf[BOOTARGS_MAX];	/* arguments passed to zone_boot() */
265 } zone_cmd_arg_t;
266 
267 /*
268  * Structure of zoneadmd's response to a request.  A NULL return value means
269  * the caller should attempt to restart zoneadmd and retry.
270  */
271 typedef struct zone_cmd_rval {
272 	int rval;			/* return value of request */
273 	char errbuf[1];	/* variable-sized buffer containing error messages */
274 } zone_cmd_rval_t;
275 
276 /*
277  * The zone support infrastructure uses the zone name as a component
278  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
279  * in length, so ZONENAME_MAX is limited by that.
280  */
281 #define	ZONENAME_MAX		64
282 
283 #define	GLOBAL_ZONENAME		"global"
284 
285 /*
286  * Extended Regular expression (see regex(5)) which matches all valid zone
287  * names.
288  */
289 #define	ZONENAME_REGEXP		"[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
290 
291 /*
292  * Where the zones support infrastructure places temporary files.
293  */
294 #define	ZONES_TMPDIR		"/var/run/zones"
295 
296 /*
297  * The path to the door used by clients to communicate with zoneadmd.
298  */
299 #define	ZONE_DOOR_PATH		ZONES_TMPDIR "/%s.zoneadmd_door"
300 
301 
302 /* zone_flags */
303 /*
304  * Threads that read or write the following flag must hold zone_lock.
305  */
306 #define	ZF_REFCOUNTS_LOGGED	0x1	/* a thread logged the zone's refs */
307 
308 /*
309  * The following threads are set when the zone is created and never changed.
310  * Threads that test for these flags don't have to hold zone_lock.
311  */
312 #define	ZF_HASHED_LABEL		0x2	/* zone has a unique label */
313 #define	ZF_IS_SCRATCH		0x4	/* scratch zone */
314 #define	ZF_NET_EXCL		0x8	/* Zone has an exclusive IP stack */
315 
316 
317 /* zone_create flags */
318 #define	ZCF_NET_EXCL		0x1	/* Create a zone with exclusive IP */
319 
320 /* zone network properties */
321 #define	ZONE_NETWORK_ADDRESS	1
322 #define	ZONE_NETWORK_DEFROUTER	2
323 
324 #define	ZONE_NET_ADDRNAME	"address"
325 #define	ZONE_NET_RTRNAME	"route"
326 
327 typedef struct zone_net_data {
328 	int zn_type;
329 	int zn_len;
330 	datalink_id_t zn_linkid;
331 	uint8_t zn_val[1];
332 } zone_net_data_t;
333 
334 
335 #ifdef _KERNEL
336 
337 #define	GLOBAL_ZONEUNIQID	0	/* uniqid of the global zone */
338 
339 /*
340  * Indexes into ->zone_ustate array, summing the micro state of all threads in a
341  * particular zone.
342  */
343 #define	ZONE_USTATE_STIME (0)
344 #define	ZONE_USTATE_UTIME (1)
345 #define	ZONE_USTATE_WTIME (2)
346 #define	ZONE_USTATE_MAX (3)
347 
348 struct pool;
349 struct brand;
350 
351 /*
352  * Each of these constants identifies a kernel subsystem that acquires and
353  * releases zone references.  Each subsystem that invokes
354  * zone_hold_ref() and zone_rele_ref() should specify the
355  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
356  * help users and developers quickly identify subsystems that stall zone
357  * shutdowns indefinitely.
358  *
359  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
360  * when you modify this enumeration.
361  */
362 typedef enum zone_ref_subsys {
363 	ZONE_REF_NFS,			/* NFS */
364 	ZONE_REF_NFSV4,			/* NFSv4 */
365 	ZONE_REF_SMBFS,			/* SMBFS */
366 	ZONE_REF_MNTFS,			/* MNTFS */
367 	ZONE_REF_LOFI,			/* LOFI devices */
368 	ZONE_REF_VFS,			/* VFS infrastructure */
369 	ZONE_REF_IPC,			/* IPC infrastructure */
370 	ZONE_REF_NUM_SUBSYS		/* This must be the last entry. */
371 } zone_ref_subsys_t;
372 
373 /*
374  * zone_ref represents a general-purpose references to a zone.  Each zone's
375  * references are linked into the zone's zone_t::zone_ref_list.  This allows
376  * debuggers to walk zones' references.
377  */
378 typedef struct zone_ref {
379 	struct zone	*zref_zone; /* the zone to which the reference refers */
380 	list_node_t	zref_linkage; /* linkage for zone_t::zone_ref_list */
381 } zone_ref_t;
382 
383 /*
384  * Structure to record list of ZFS datasets exported to a zone.
385  */
386 typedef struct zone_dataset {
387 	char		*zd_dataset;
388 	list_node_t	zd_linkage;
389 } zone_dataset_t;
390 
391 /*
392  * structure for zone kstats
393  */
394 typedef struct zone_kstat {
395 	kstat_named_t zk_zonename;
396 	kstat_named_t zk_usage;
397 	kstat_named_t zk_value;
398 } zone_kstat_t;
399 
400 struct cpucap;
401 
402 typedef struct {
403 	kstat_named_t	zm_zonename;
404 	kstat_named_t	zm_pgpgin;
405 	kstat_named_t	zm_anonpgin;
406 	kstat_named_t	zm_execpgin;
407 	kstat_named_t	zm_fspgin;
408 	kstat_named_t	zm_anon_alloc_fail;
409 } zone_mcap_kstat_t;
410 
411 typedef struct {
412 	kstat_named_t	zm_zonename;	/* full name, kstat truncates name */
413 	kstat_named_t	zm_utime;
414 	kstat_named_t	zm_stime;
415 	kstat_named_t	zm_wtime;
416 	kstat_named_t	zm_avenrun1;
417 	kstat_named_t	zm_avenrun5;
418 	kstat_named_t	zm_avenrun15;
419 	kstat_named_t	zm_ffcap;
420 	kstat_named_t	zm_ffnoproc;
421 	kstat_named_t	zm_ffnomem;
422 	kstat_named_t	zm_ffmisc;
423 	kstat_named_t	zm_nested_intp;
424 	kstat_named_t	zm_init_pid;
425 	kstat_named_t	zm_boot_time;
426 } zone_misc_kstat_t;
427 
428 typedef struct zone {
429 	/*
430 	 * zone_name is never modified once set.
431 	 */
432 	char		*zone_name;	/* zone's configuration name */
433 	/*
434 	 * zone_nodename and zone_domain are never freed once allocated.
435 	 */
436 	char		*zone_nodename;	/* utsname.nodename equivalent */
437 	char		*zone_domain;	/* srpc_domain equivalent */
438 	/*
439 	 * zone_hostid is used for per-zone hostid emulation.
440 	 * Currently it isn't modified after it's set (so no locks protect
441 	 * accesses), but that might have to change when we allow
442 	 * administrators to change running zones' properties.
443 	 *
444 	 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
445 	 * that zone_get_hostid() will function correctly.
446 	 */
447 	uint32_t	zone_hostid;	/* zone's hostid, HW_INVALID_HOSTID */
448 					/* if not emulated */
449 	/*
450 	 * zone_lock protects the following fields of a zone_t:
451 	 *	zone_ref
452 	 *	zone_cred_ref
453 	 *	zone_subsys_ref
454 	 *	zone_ref_list
455 	 *	zone_ntasks
456 	 *	zone_flags
457 	 *	zone_zsd
458 	 *	zone_pfexecd
459 	 */
460 	kmutex_t	zone_lock;
461 	/*
462 	 * zone_linkage is the zone's linkage into the active or
463 	 * death-row list.  The field is protected by zonehash_lock.
464 	 */
465 	list_node_t	zone_linkage;
466 	zoneid_t	zone_id;	/* ID of zone */
467 	uint_t		zone_ref;	/* count of zone_hold()s on zone */
468 	uint_t		zone_cred_ref;	/* count of zone_hold_cred()s on zone */
469 	/*
470 	 * Fixed-sized array of subsystem-specific reference counts
471 	 * The sum of all of the counts must be less than or equal to zone_ref.
472 	 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
473 	 * constants.
474 	 */
475 	uint_t		zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
476 	list_t		zone_ref_list;	/* list of zone_ref_t structs */
477 	/*
478 	 * zone_rootvp and zone_rootpath can never be modified once set.
479 	 */
480 	struct vnode	*zone_rootvp;	/* zone's root vnode */
481 	char		*zone_rootpath;	/* Path to zone's root + '/' */
482 	ushort_t	zone_flags;	/* misc flags */
483 	zone_status_t	zone_status;	/* protected by zone_status_lock */
484 	uint_t		zone_ntasks;	/* number of tasks executing in zone */
485 	kmutex_t	zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
486 					/* counters in projects and tasks */
487 					/* that are within the zone */
488 	rctl_qty_t	zone_nlwps;	/* number of lwps in zone */
489 	rctl_qty_t	zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
490 	rctl_qty_t	zone_shmmax;	/* System V shared memory usage */
491 	ipc_rqty_t	zone_ipc;	/* System V IPC id resource usage */
492 
493 	uint_t		zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
494 	uint32_t	zone_shares;	/* FSS shares allocated to zone */
495 	rctl_set_t	*zone_rctls;	/* zone-wide (zone.*) rctls */
496 	kmutex_t	zone_mem_lock;	/* protects zone_locked_mem and */
497 					/* kpd_locked_mem for all */
498 					/* projects in zone. */
499 					/* Also protects zone_max_swap */
500 					/* grab after p_lock, before rcs_lock */
501 	rctl_qty_t	zone_locked_mem;	/* bytes of locked memory in */
502 						/* zone */
503 	rctl_qty_t	zone_locked_mem_ctl;	/* Current locked memory */
504 						/* limit.  Protected by */
505 						/* zone_rctls->rcs_lock */
506 	rctl_qty_t	zone_max_swap; /* bytes of swap reserved by zone */
507 	rctl_qty_t	zone_max_swap_ctl;	/* current swap limit. */
508 						/* Protected by */
509 						/* zone_rctls->rcs_lock */
510 	kmutex_t	zone_rctl_lock;	/* protects zone_max_lofi */
511 	rctl_qty_t	zone_max_lofi; /* lofi devs for zone */
512 	rctl_qty_t	zone_max_lofi_ctl;	/* current lofi limit. */
513 						/* Protected by */
514 						/* zone_rctls->rcs_lock */
515 	list_t		zone_zsd;	/* list of Zone-Specific Data values */
516 	kcondvar_t	zone_cv;	/* used to signal state changes */
517 	struct proc	*zone_zsched;	/* Dummy kernel "zsched" process */
518 	pid_t		zone_proc_initpid; /* pid of "init" for this zone */
519 	char		*zone_initname;	/* fs path to 'init' */
520 	int		zone_boot_err;  /* for zone_boot() if boot fails */
521 	char		*zone_bootargs;	/* arguments passed via zone_boot() */
522 	uint64_t	zone_phys_mcap;	/* physical memory cap */
523 	/*
524 	 * zone_kthreads is protected by zone_status_lock.
525 	 */
526 	kthread_t	*zone_kthreads;	/* kernel threads in zone */
527 	struct priv_set	*zone_privset;	/* limit set for zone */
528 	/*
529 	 * zone_vfslist is protected by vfs_list_lock().
530 	 */
531 	struct vfs	*zone_vfslist;	/* list of FS's mounted in zone */
532 	uint64_t	zone_uniqid;	/* unique zone generation number */
533 	struct cred	*zone_kcred;	/* kcred-like, zone-limited cred */
534 	/*
535 	 * zone_pool is protected by pool_lock().
536 	 */
537 	struct pool	*zone_pool;	/* pool the zone is bound to */
538 	hrtime_t	zone_pool_mod;	/* last pool bind modification time */
539 	/* zone_psetid is protected by cpu_lock */
540 	psetid_t	zone_psetid;	/* pset the zone is bound to */
541 
542 	time_t		zone_boot_time;	/* Similar to boot_time */
543 
544 	/*
545 	 * The following two can be read without holding any locks.  They are
546 	 * updated under cpu_lock.
547 	 */
548 	int		zone_ncpus;  /* zone's idea of ncpus */
549 	int		zone_ncpus_online; /* zone's idea of ncpus_online */
550 	/*
551 	 * List of ZFS datasets exported to this zone.
552 	 */
553 	list_t		zone_datasets;	/* list of datasets */
554 
555 	ts_label_t	*zone_slabel;	/* zone sensitivity label */
556 	int		zone_match;	/* require label match for packets */
557 	tsol_mlp_list_t zone_mlps;	/* MLPs on zone-private addresses */
558 
559 	boolean_t	zone_restart_init;	/* Restart init if it dies? */
560 	boolean_t	zone_reboot_on_init_exit; /* Reboot if init dies? */
561 	boolean_t	zone_restart_init_0;	/* Restart only if it exits 0 */
562 	struct brand	*zone_brand;		/* zone's brand */
563 	void		*zone_brand_data;	/* store brand specific data */
564 	id_t		zone_defaultcid;	/* dflt scheduling class id */
565 	kstat_t		*zone_swapresv_kstat;
566 	kstat_t		*zone_lockedmem_kstat;
567 	/*
568 	 * zone_dl_list is protected by zone_lock
569 	 */
570 	list_t		zone_dl_list;
571 	netstack_t	*zone_netstack;
572 	struct cpucap	*zone_cpucap;	/* CPU caps data */
573 	/*
574 	 * Solaris Auditing per-zone audit context
575 	 */
576 	struct au_kcontext	*zone_audit_kctxt;
577 	/*
578 	 * For private use by mntfs.
579 	 */
580 	struct mntelem	*zone_mntfs_db;
581 	krwlock_t	zone_mntfs_db_lock;
582 
583 	struct klpd_reg		*zone_pfexecd;
584 
585 	char		*zone_fs_allowed;
586 	rctl_qty_t	zone_nprocs;	/* number of processes in the zone */
587 	rctl_qty_t	zone_nprocs_ctl;	/* current limit protected by */
588 						/* zone_rctls->rcs_lock */
589 	kstat_t		*zone_nprocs_kstat;
590 
591 	kmutex_t	zone_mcap_lock;	/* protects mcap statistics */
592 	kstat_t		*zone_mcap_ksp;
593 	zone_mcap_kstat_t *zone_mcap_stats;
594 	uint64_t	zone_pgpgin;		/* pages paged in */
595 	uint64_t	zone_anonpgin;		/* anon pages paged in */
596 	uint64_t	zone_execpgin;		/* exec pages paged in */
597 	uint64_t	zone_fspgin;		/* fs pages paged in */
598 	uint64_t	zone_anon_alloc_fail;	/* cnt of anon alloc fails */
599 
600 	psecflags_t	zone_secflags; /* default zone security-flags */
601 
602 	/*
603 	 * Misc. kstats and counters for zone cpu-usage aggregation.
604 	 */
605 	kmutex_t	zone_misc_lock;		/* protects misc statistics */
606 	kstat_t		*zone_misc_ksp;
607 	zone_misc_kstat_t *zone_misc_stats;
608 	/* Accumulated microstate for all threads in this zone. */
609 	cpu_uarray_t	*zone_ustate;
610 	/* fork-fail kstat tracking */
611 	uint32_t	zone_ffcap;		/* hit an rctl cap */
612 	uint32_t	zone_ffnoproc;		/* get proc/lwp error */
613 	uint32_t	zone_ffnomem;		/* as_dup/memory error */
614 	uint32_t	zone_ffmisc;		/* misc. other error */
615 
616 	uint32_t	zone_nested_intp;	/* nested interp. kstat */
617 
618 	struct loadavg_s zone_loadavg;		/* loadavg for this zone */
619 	uint64_t	zone_hp_avenrun[3];	/* high-precision avenrun */
620 	int		zone_avenrun[3];	/* FSCALED avg. run queue len */
621 
622 	/*
623 	 * FSS stats updated once per second by fss_decay_usage.
624 	 */
625 	uint32_t	zone_fss_gen;		/* FSS generation cntr */
626 	uint64_t	zone_run_ticks;		/* tot # of ticks running */
627 
628 	/*
629 	 * DTrace-private per-zone state
630 	 */
631 	int		zone_dtrace_getf;	/* # of unprivileged getf()s */
632 
633 	/*
634 	 * Synchronization primitives used to synchronize between mounts and
635 	 * zone creation/destruction.
636 	 */
637 	int		zone_mounts_in_progress;
638 	kcondvar_t	zone_mount_cv;
639 	kmutex_t	zone_mount_lock;
640 } zone_t;
641 
642 /*
643  * Special value of zone_psetid to indicate that pools are disabled.
644  */
645 #define	ZONE_PS_INVAL	PS_MYID
646 
647 extern zone_t zone0;
648 extern zone_t *global_zone;
649 extern uint_t maxzones;
650 extern rctl_hndl_t rc_zone_nlwps;
651 extern rctl_hndl_t rc_zone_nprocs;
652 
653 extern long zone(int, void *, void *, void *, void *);
654 extern void zone_zsd_init(void);
655 extern void zone_init(void);
656 extern void zone_hold(zone_t *);
657 extern void zone_rele(zone_t *);
658 extern void zone_init_ref(zone_ref_t *);
659 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
660 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
661 extern void zone_cred_hold(zone_t *);
662 extern void zone_cred_rele(zone_t *);
663 extern void zone_task_hold(zone_t *);
664 extern void zone_task_rele(zone_t *);
665 extern zone_t *zone_find_by_id(zoneid_t);
666 extern zone_t *zone_find_by_label(const ts_label_t *);
667 extern zone_t *zone_find_by_name(char *);
668 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
669 extern zone_t *zone_find_by_path(const char *);
670 extern zoneid_t getzoneid(void);
671 extern zone_t *zone_find_by_id_nolock(zoneid_t);
672 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
673 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
674 extern void zone_loadavg_update();
675 
676 /*
677  * Zone-specific data (ZSD) APIs
678  */
679 /*
680  * The following is what code should be initializing its zone_key_t to if it
681  * calls zone_getspecific() without necessarily knowing that zone_key_create()
682  * has been called on the key.
683  */
684 #define	ZONE_KEY_UNINITIALIZED	0
685 
686 typedef uint_t zone_key_t;
687 
688 extern void	zone_key_create(zone_key_t *, void *(*)(zoneid_t),
689     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
690 extern int	zone_key_delete(zone_key_t);
691 extern void	*zone_getspecific(zone_key_t, zone_t *);
692 extern int	zone_setspecific(zone_key_t, zone_t *, const void *);
693 
694 /*
695  * The definition of a zsd_entry is truly private to zone.c and is only
696  * placed here so it can be shared with mdb.
697  *
698  * State maintained for each zone times each registered key, which tracks
699  * the state of the create, shutdown and destroy callbacks.
700  *
701  * zsd_flags is used to keep track of pending actions to avoid holding locks
702  * when calling the create/shutdown/destroy callbacks, since doing so
703  * could lead to deadlocks.
704  */
705 struct zsd_entry {
706 	zone_key_t		zsd_key;	/* Key used to lookup value */
707 	void			*zsd_data;	/* Caller-managed value */
708 	/*
709 	 * Callbacks to be executed when a zone is created, shutdown, and
710 	 * destroyed, respectively.
711 	 */
712 	void			*(*zsd_create)(zoneid_t);
713 	void			(*zsd_shutdown)(zoneid_t, void *);
714 	void			(*zsd_destroy)(zoneid_t, void *);
715 	list_node_t		zsd_linkage;
716 	uint16_t		zsd_flags;	/* See below */
717 	kcondvar_t		zsd_cv;
718 };
719 
720 /*
721  * zsd_flags
722  */
723 #define	ZSD_CREATE_NEEDED	0x0001
724 #define	ZSD_CREATE_INPROGRESS	0x0002
725 #define	ZSD_CREATE_COMPLETED	0x0004
726 #define	ZSD_SHUTDOWN_NEEDED	0x0010
727 #define	ZSD_SHUTDOWN_INPROGRESS	0x0020
728 #define	ZSD_SHUTDOWN_COMPLETED	0x0040
729 #define	ZSD_DESTROY_NEEDED	0x0100
730 #define	ZSD_DESTROY_INPROGRESS	0x0200
731 #define	ZSD_DESTROY_COMPLETED	0x0400
732 
733 #define	ZSD_CREATE_ALL	\
734 	(ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
735 #define	ZSD_SHUTDOWN_ALL	\
736 	(ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
737 #define	ZSD_DESTROY_ALL	\
738 	(ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
739 
740 #define	ZSD_ALL_INPROGRESS \
741 	(ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
742 
743 /*
744  * Macros to help with zone visibility restrictions.
745  */
746 
747 /*
748  * Is process in the global zone?
749  */
750 #define	INGLOBALZONE(p) \
751 	((p)->p_zone == global_zone)
752 
753 /*
754  * Can process view objects in given zone?
755  */
756 #define	HASZONEACCESS(p, zoneid) \
757 	((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
758 
759 /*
760  * Convenience macro to see if a resolved path is visible from within a
761  * given zone.
762  *
763  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
764  * two strings must be equal.  Since the rootpathlen has a trailing '/',
765  * we want to skip everything in the path up to (but not including) the
766  * trailing '/'.
767  */
768 #define	ZONE_PATH_VISIBLE(path, zone) \
769 	(strncmp((path), (zone)->zone_rootpath,		\
770 	    (zone)->zone_rootpathlen - 1) == 0)
771 
772 /*
773  * Convenience macro to go from the global view of a path to that seen
774  * from within said zone.  It is the responsibility of the caller to
775  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
776  * in fact visible from within the zone.
777  */
778 #define	ZONE_PATH_TRANSLATE(path, zone)	\
779 	(ASSERT(ZONE_PATH_VISIBLE(path, zone)),	\
780 	(path) + (zone)->zone_rootpathlen - 2)
781 
782 /*
783  * Special processes visible in all zones.
784  */
785 #define	ZONE_SPECIALPID(x)	 ((x) == 0 || (x) == 1)
786 
787 /*
788  * A root vnode of the current zone.
789  *
790  * NOTE: It may be necessary (initialization time for file sharing where an
791  * NGZ loads a file-sharing kernel module that does zsd initialization) to NOT
792  * use this macro. One should ASSERT() that curzone == active ZSD (an
793  * ASSERTion that's not always true at ZSD initialization time) during regular
794  * use of this macro.
795  */
796 #define	ZONE_ROOTVP()	(curzone->zone_rootvp)
797 
798 /*
799  * Since a zone's root isn't necessarily an actual filesystem boundary
800  * (i.e. VROOT may not be set on zone->zone_rootvp) we need to not assume it.
801  * This macro helps in checking if a vnode is the current zone's rootvp.
802  * NOTE:  Using the VN_ prefix, even though it's defined here in zone.h.
803  * NOTE2: See above warning about ZONE_ROOTVP().
804  */
805 #define	VN_IS_CURZONEROOT(vp)   (VN_CMP(vp, ZONE_ROOTVP()))
806 
807 /*
808  * Zone-safe version of thread_create() to be used when the caller wants to
809  * create a kernel thread to run within the current zone's context.
810  */
811 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
812     pri_t);
813 extern void zthread_exit(void);
814 
815 /*
816  * Functions for an external observer to register interest in a zone's status
817  * change.  Observers will be woken up when the zone status equals the status
818  * argument passed in (in the case of zone_status_timedwait, the function may
819  * also return because of a timeout; zone_status_wait_sig may return early due
820  * to a signal being delivered; zone_status_timedwait_sig may return for any of
821  * the above reasons).
822  *
823  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
824  * cv_wait_sig() respectively.
825  */
826 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
827 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
828 extern void zone_status_wait(zone_t *, zone_status_t);
829 extern int zone_status_wait_sig(zone_t *, zone_status_t);
830 
831 /*
832  * Get the status  of the zone (at the time it was called).  The state may
833  * have progressed by the time it is returned.
834  */
835 extern zone_status_t zone_status_get(zone_t *);
836 
837 /*
838  * Safely get the hostid of the specified zone (defaults to machine's hostid
839  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
840  * the global zone's (i.e., physical system's) hostid.
841  */
842 extern uint32_t zone_get_hostid(zone_t *);
843 
844 /*
845  * Get the "kcred" credentials corresponding to the given zone.
846  */
847 extern struct cred *zone_get_kcred(zoneid_t);
848 
849 /*
850  * Get/set the pool the zone is currently bound to.
851  */
852 extern struct pool *zone_pool_get(zone_t *);
853 extern void zone_pool_set(zone_t *, struct pool *);
854 
855 /*
856  * Get/set the pset the zone is currently using.
857  */
858 extern psetid_t zone_pset_get(zone_t *);
859 extern void zone_pset_set(zone_t *, psetid_t);
860 
861 /*
862  * Get the number of cpus/online-cpus visible from the given zone.
863  */
864 extern int zone_ncpus_get(zone_t *);
865 extern int zone_ncpus_online_get(zone_t *);
866 
867 /*
868  * Returns true if the named pool/dataset is visible in the current zone.
869  */
870 extern int zone_dataset_visible(const char *, int *);
871 
872 /*
873  * zone version of kadmin()
874  */
875 extern int zone_kadmin(int, int, const char *, cred_t *);
876 extern void zone_shutdown_global(void);
877 
878 extern void mount_in_progress(zone_t *);
879 extern void mount_completed(zone_t *);
880 
881 extern int zone_walk(int (*)(zone_t *, void *), void *);
882 
883 extern rctl_hndl_t rc_zone_locked_mem;
884 extern rctl_hndl_t rc_zone_max_swap;
885 extern rctl_hndl_t rc_zone_max_lofi;
886 
887 #endif	/* _KERNEL */
888 
889 #ifdef	__cplusplus
890 }
891 #endif
892 
893 #endif	/* _SYS_ZONE_H */
894