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 2013, Joyent, Inc. All rights reserved.
24 * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
25 * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
26 * Copyright 2016 Garrett D'Amore
27 */
28
29 #ifndef _SYS_ZONE_H
30 #define _SYS_ZONE_H
31
32 #include <sys/types.h>
33 #include <sys/mutex.h>
34 #include <sys/param.h>
35 #include <sys/rctl.h>
36 #include <sys/ipc_rctl.h>
37 #include <sys/pset.h>
38 #include <sys/tsol/label.h>
39 #include <sys/cred.h>
40 #include <sys/netstack.h>
41 #include <sys/uadmin.h>
42 #include <sys/ksynch.h>
43 #include <sys/socket_impl.h>
44 #include <netinet/in.h>
45
46 #ifdef __cplusplus
47 extern "C" {
48 #endif
49
50 /*
51 * NOTE
52 *
53 * The contents of this file are private to the implementation of
54 * Solaris and are subject to change at any time without notice.
55 * Applications and drivers using these interfaces may fail to
56 * run on future releases.
57 */
58
59 /* Available both in kernel and for user space */
60
61 /* zone id restrictions and special ids */
62 #define MAX_ZONEID 9999
63 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */
64 #define MIN_ZONEID 0 /* minimum zone ID on system */
65 #define GLOBAL_ZONEID 0
66 #define ZONEID_WIDTH 4 /* for printf */
67
68 /*
69 * Special zoneid_t token to refer to all zones.
70 */
71 #define ALL_ZONES (-1)
72
73 /* system call subcodes */
74 #define ZONE_CREATE 0
75 #define ZONE_DESTROY 1
76 #define ZONE_GETATTR 2
77 #define ZONE_ENTER 3
78 #define ZONE_LIST 4
79 #define ZONE_SHUTDOWN 5
80 #define ZONE_LOOKUP 6
81 #define ZONE_BOOT 7
82 #define ZONE_VERSION 8
83 #define ZONE_SETATTR 9
84 #define ZONE_ADD_DATALINK 10
85 #define ZONE_DEL_DATALINK 11
86 #define ZONE_CHECK_DATALINK 12
87 #define ZONE_LIST_DATALINK 13
88
89 /* zone attributes */
90 #define ZONE_ATTR_ROOT 1
91 #define ZONE_ATTR_NAME 2
92 #define ZONE_ATTR_STATUS 3
93 #define ZONE_ATTR_PRIVSET 4
94 #define ZONE_ATTR_UNIQID 5
95 #define ZONE_ATTR_POOLID 6
96 #define ZONE_ATTR_INITPID 7
97 #define ZONE_ATTR_SLBL 8
98 #define ZONE_ATTR_INITNAME 9
99 #define ZONE_ATTR_BOOTARGS 10
100 #define ZONE_ATTR_BRAND 11
101 #define ZONE_ATTR_PHYS_MCAP 12
102 #define ZONE_ATTR_SCHED_CLASS 13
103 #define ZONE_ATTR_FLAGS 14
104 #define ZONE_ATTR_HOSTID 15
105 #define ZONE_ATTR_FS_ALLOWED 16
106 #define ZONE_ATTR_NETWORK 17
107 #define ZONE_ATTR_INITNORESTART 20
108
109 /* Start of the brand-specific attribute namespace */
110 #define ZONE_ATTR_BRAND_ATTRS 32768
111
112 #define ZONE_FS_ALLOWED_MAX 1024
113
114 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
115 #define ZONE_EVENT_STATUS_CLASS "status"
116 #define ZONE_EVENT_STATUS_SUBCLASS "change"
117
118 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
119 #define ZONE_EVENT_INITIALIZED "initialized"
120 #define ZONE_EVENT_READY "ready"
121 #define ZONE_EVENT_RUNNING "running"
122 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
123
124 #define ZONE_CB_NAME "zonename"
125 #define ZONE_CB_NEWSTATE "newstate"
126 #define ZONE_CB_OLDSTATE "oldstate"
127 #define ZONE_CB_TIMESTAMP "when"
128 #define ZONE_CB_ZONEID "zoneid"
129
130 /*
131 * Exit values that may be returned by scripts or programs invoked by various
132 * zone commands.
133 *
134 * These are defined as:
135 *
136 * ZONE_SUBPROC_OK
137 * ===============
138 * The subprocess completed successfully.
139 *
140 * ZONE_SUBPROC_USAGE
141 * ==================
142 * The subprocess failed with a usage message, or a usage message should
143 * be output in its behalf.
144 *
145 * ZONE_SUBPROC_NOTCOMPLETE
146 * ========================
147 * The subprocess did not complete, but the actions performed by the
148 * subprocess require no recovery actions by the user.
149 *
150 * For example, if the subprocess were called by "zoneadm install," the
151 * installation of the zone did not succeed but the user need not perform
152 * a "zoneadm uninstall" before attempting another install.
153 *
154 * ZONE_SUBPROC_FATAL
155 * ==================
156 * The subprocess failed in a fatal manner, usually one that will require
157 * some type of recovery action by the user.
158 *
159 * For example, if the subprocess were called by "zoneadm install," the
160 * installation of the zone did not succeed and the user will need to
161 * perform a "zoneadm uninstall" before another install attempt is
162 * possible.
163 *
164 * The non-success exit values are large to avoid accidental collision
165 * with values used internally by some commands (e.g. "Z_ERR" and
166 * "Z_USAGE" as used by zoneadm.)
167 */
168 #define ZONE_SUBPROC_OK 0
169 #define ZONE_SUBPROC_USAGE 253
170 #define ZONE_SUBPROC_NOTCOMPLETE 254
171 #define ZONE_SUBPROC_FATAL 255
172
173 #ifdef _SYSCALL32
174 typedef struct {
175 caddr32_t zone_name;
176 caddr32_t zone_root;
177 caddr32_t zone_privs;
178 size32_t zone_privssz;
179 caddr32_t rctlbuf;
180 size32_t rctlbufsz;
181 caddr32_t extended_error;
182 caddr32_t zfsbuf;
183 size32_t zfsbufsz;
184 int match; /* match level */
185 uint32_t doi; /* DOI for label */
186 caddr32_t label; /* label associated with zone */
187 int flags;
188 } zone_def32;
189 #endif
190 typedef struct {
191 const char *zone_name;
192 const char *zone_root;
193 const struct priv_set *zone_privs;
194 size_t zone_privssz;
195 const char *rctlbuf;
196 size_t rctlbufsz;
197 int *extended_error;
198 const char *zfsbuf;
199 size_t zfsbufsz;
200 int match; /* match level */
201 uint32_t doi; /* DOI for label */
202 const bslabel_t *label; /* label associated with zone */
203 int flags;
204 } zone_def;
205
206 /* extended error information */
207 #define ZE_UNKNOWN 0 /* No extended error info */
208 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
209 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
210 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
211
212 /*
213 * zone_status values
214 *
215 * You must modify zone_status_names in mdb(1M)'s genunix module
216 * (genunix/zone.c) when you modify this enum.
217 */
218 typedef enum {
219 ZONE_IS_UNINITIALIZED = 0,
220 ZONE_IS_INITIALIZED,
221 ZONE_IS_READY,
222 ZONE_IS_BOOTING,
223 ZONE_IS_RUNNING,
224 ZONE_IS_SHUTTING_DOWN,
225 ZONE_IS_EMPTY,
226 ZONE_IS_DOWN,
227 ZONE_IS_DYING,
228 ZONE_IS_DEAD
229 } zone_status_t;
230 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
231 #define ZONE_MAX_STATE ZONE_IS_DEAD
232
233 /*
234 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
235 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
236 */
237 typedef enum zone_cmd {
238 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
239 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
240 } zone_cmd_t;
241
242 /*
243 * The structure of a request to zoneadmd.
244 */
245 typedef struct zone_cmd_arg {
246 uint64_t uniqid; /* unique "generation number" */
247 zone_cmd_t cmd; /* requested action */
248 uint32_t _pad; /* need consistent 32/64 bit alignmt */
249 char locale[MAXPATHLEN]; /* locale in which to render messages */
250 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
251 } zone_cmd_arg_t;
252
253 /*
254 * Structure of zoneadmd's response to a request. A NULL return value means
255 * the caller should attempt to restart zoneadmd and retry.
256 */
257 typedef struct zone_cmd_rval {
258 int rval; /* return value of request */
259 char errbuf[1]; /* variable-sized buffer containing error messages */
260 } zone_cmd_rval_t;
261
262 /*
263 * The zone support infrastructure uses the zone name as a component
264 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
265 * in length, so ZONENAME_MAX is limited by that.
266 */
267 #define ZONENAME_MAX 64
268
269 #define GLOBAL_ZONENAME "global"
270
271 /*
272 * Extended Regular expression (see regex(5)) which matches all valid zone
273 * names.
274 */
275 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
276
277 /*
278 * Where the zones support infrastructure places temporary files.
279 */
280 #define ZONES_TMPDIR "/var/run/zones"
281
282 /*
283 * The path to the door used by clients to communicate with zoneadmd.
284 */
285 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
286
287
288 /* zone_flags */
289 /*
290 * Threads that read or write the following flag must hold zone_lock.
291 */
292 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
293
294 /*
295 * The following threads are set when the zone is created and never changed.
296 * Threads that test for these flags don't have to hold zone_lock.
297 */
298 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
299 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
300 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
301
302
303 /* zone_create flags */
304 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
305
306 /* zone network properties */
307 #define ZONE_NETWORK_ADDRESS 1
308 #define ZONE_NETWORK_DEFROUTER 2
309
310 #define ZONE_NET_ADDRNAME "address"
311 #define ZONE_NET_RTRNAME "route"
312
313 typedef struct zone_net_data {
314 int zn_type;
315 int zn_len;
316 datalink_id_t zn_linkid;
317 uint8_t zn_val[1];
318 } zone_net_data_t;
319
320
321 #ifdef _KERNEL
322
323 /*
324 * We need to protect the definition of 'list_t' from userland applications and
325 * libraries which may be defining ther own versions.
326 */
327 #include <sys/list.h>
328 #include <sys/loadavg.h>
329
330 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
331
332 struct pool;
333 struct brand;
334
335 /*
336 * Each of these constants identifies a kernel subsystem that acquires and
337 * releases zone references. Each subsystem that invokes
338 * zone_hold_ref() and zone_rele_ref() should specify the
339 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
340 * help users and developers quickly identify subsystems that stall zone
341 * shutdowns indefinitely.
342 *
343 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
344 * when you modify this enumeration.
345 */
346 typedef enum zone_ref_subsys {
347 ZONE_REF_NFS, /* NFS */
348 ZONE_REF_NFSV4, /* NFSv4 */
349 ZONE_REF_SMBFS, /* SMBFS */
350 ZONE_REF_MNTFS, /* MNTFS */
351 ZONE_REF_LOFI, /* LOFI devices */
352 ZONE_REF_VFS, /* VFS infrastructure */
353 ZONE_REF_IPC, /* IPC infrastructure */
354 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
355 } zone_ref_subsys_t;
356
357 /*
358 * zone_ref represents a general-purpose references to a zone. Each zone's
359 * references are linked into the zone's zone_t::zone_ref_list. This allows
360 * debuggers to walk zones' references.
361 */
362 typedef struct zone_ref {
363 struct zone *zref_zone; /* the zone to which the reference refers */
364 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
365 } zone_ref_t;
366
367 /*
368 * Structure to record list of ZFS datasets exported to a zone.
369 */
370 typedef struct zone_dataset {
371 char *zd_dataset;
372 list_node_t zd_linkage;
373 } zone_dataset_t;
374
375 /*
376 * structure for zone kstats
377 */
378 typedef struct zone_kstat {
379 kstat_named_t zk_zonename;
380 kstat_named_t zk_usage;
381 kstat_named_t zk_value;
382 } zone_kstat_t;
383
384 struct cpucap;
385
386 typedef struct {
387 kstat_named_t zm_zonename; /* full name, kstat truncates name */
388 kstat_named_t zm_utime;
389 kstat_named_t zm_stime;
390 kstat_named_t zm_wtime;
391 kstat_named_t zm_avenrun1;
392 kstat_named_t zm_avenrun5;
393 kstat_named_t zm_avenrun15;
394 kstat_named_t zm_run_ticks;
395 kstat_named_t zm_run_wait;
396 kstat_named_t zm_fss_shr_pct;
397 kstat_named_t zm_fss_pri_hi;
398 kstat_named_t zm_fss_pri_avg;
399 kstat_named_t zm_ffcap;
400 kstat_named_t zm_ffnoproc;
401 kstat_named_t zm_ffnomem;
402 kstat_named_t zm_ffmisc;
403 kstat_named_t zm_boot_hrtime;
404 } zone_misc_kstat_t;
405
406 typedef struct zone {
407 /*
408 * zone_name is never modified once set.
409 */
410 char *zone_name; /* zone's configuration name */
411 /*
412 * zone_nodename and zone_domain are never freed once allocated.
413 */
414 char *zone_nodename; /* utsname.nodename equivalent */
415 char *zone_domain; /* srpc_domain equivalent */
416 /*
417 * zone_hostid is used for per-zone hostid emulation.
418 * Currently it isn't modified after it's set (so no locks protect
419 * accesses), but that might have to change when we allow
420 * administrators to change running zones' properties.
421 *
422 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
423 * that zone_get_hostid() will function correctly.
424 */
425 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
426 /* if not emulated */
427 /*
428 * zone_lock protects the following fields of a zone_t:
429 * zone_ref
430 * zone_cred_ref
431 * zone_subsys_ref
432 * zone_ref_list
433 * zone_ntasks
434 * zone_flags
435 * zone_zsd
436 * zone_pfexecd
437 */
438 kmutex_t zone_lock;
439 /*
440 * zone_linkage is the zone's linkage into the active or
441 * death-row list. The field is protected by zonehash_lock.
442 */
443 list_node_t zone_linkage;
444 zoneid_t zone_id; /* ID of zone */
445 uint_t zone_ref; /* count of zone_hold()s on zone */
446 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
447 /*
448 * Fixed-sized array of subsystem-specific reference counts
449 * The sum of all of the counts must be less than or equal to zone_ref.
450 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
451 * constants.
452 */
453 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
454 list_t zone_ref_list; /* list of zone_ref_t structs */
455 /*
456 * zone_rootvp and zone_rootpath can never be modified once set.
457 */
458 struct vnode *zone_rootvp; /* zone's root vnode */
459 char *zone_rootpath; /* Path to zone's root + '/' */
460 ushort_t zone_flags; /* misc flags */
461 zone_status_t zone_status; /* protected by zone_status_lock */
462 uint_t zone_ntasks; /* number of tasks executing in zone */
463 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
464 /* counters in projects and tasks */
465 /* that are within the zone */
466 rctl_qty_t zone_nlwps; /* number of lwps in zone */
467 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
468 rctl_qty_t zone_shmmax; /* System V shared memory usage */
469 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
470
471 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
472 uint32_t zone_shares; /* FSS shares allocated to zone */
473 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
474 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
475 /* kpd_locked_mem for all */
476 /* projects in zone. */
477 /* Also protects zone_max_swap */
478 /* grab after p_lock, before rcs_lock */
479 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
480 /* zone */
481 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
482 /* limit. Protected by */
483 /* zone_rctls->rcs_lock */
484 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
485 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
486 /* Protected by */
487 /* zone_rctls->rcs_lock */
488 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
489 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
490 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
491 /* Protected by */
492 /* zone_rctls->rcs_lock */
493 list_t zone_zsd; /* list of Zone-Specific Data values */
494 kcondvar_t zone_cv; /* used to signal state changes */
495 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
496 pid_t zone_proc_initpid; /* pid of "init" for this zone */
497 char *zone_initname; /* fs path to 'init' */
498 int zone_boot_err; /* for zone_boot() if boot fails */
499 char *zone_bootargs; /* arguments passed via zone_boot() */
500 uint64_t zone_phys_mcap; /* physical memory cap */
501 /*
502 * zone_kthreads is protected by zone_status_lock.
503 */
504 kthread_t *zone_kthreads; /* kernel threads in zone */
505 struct priv_set *zone_privset; /* limit set for zone */
506 /*
507 * zone_vfslist is protected by vfs_list_lock().
508 */
509 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
510 uint64_t zone_uniqid; /* unique zone generation number */
511 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
512 /*
513 * zone_pool is protected by pool_lock().
514 */
515 struct pool *zone_pool; /* pool the zone is bound to */
516 hrtime_t zone_pool_mod; /* last pool bind modification time */
517 /* zone_psetid is protected by cpu_lock */
518 psetid_t zone_psetid; /* pset the zone is bound to */
519
520 time_t zone_boot_time; /* Similar to boot_time */
521 hrtime_t zone_boot_hrtime; /* gethrtime() at zone boot */
522
523 /*
524 * The following two can be read without holding any locks. They are
525 * updated under cpu_lock.
526 */
527 int zone_ncpus; /* zone's idea of ncpus */
528 int zone_ncpus_online; /* zone's idea of ncpus_online */
529 /*
530 * List of ZFS datasets exported to this zone.
531 */
532 list_t zone_datasets; /* list of datasets */
533
534 ts_label_t *zone_slabel; /* zone sensitivity label */
535 int zone_match; /* require label match for packets */
536 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
537
538 boolean_t zone_restart_init; /* Restart init if it dies? */
539 struct brand *zone_brand; /* zone's brand */
540 void *zone_brand_data; /* store brand specific data */
541 id_t zone_defaultcid; /* dflt scheduling class id */
542 kstat_t *zone_swapresv_kstat;
543 kstat_t *zone_lockedmem_kstat;
544 /*
545 * zone_dl_list is protected by zone_lock
546 */
547 list_t zone_dl_list;
548 netstack_t *zone_netstack;
549 struct cpucap *zone_cpucap; /* CPU caps data */
550 /*
551 * Solaris Auditing per-zone audit context
552 */
553 struct au_kcontext *zone_audit_kctxt;
554 /*
555 * For private use by mntfs.
556 */
557 struct mntelem *zone_mntfs_db;
558 krwlock_t zone_mntfs_db_lock;
559
560 struct klpd_reg *zone_pfexecd;
561
562 char *zone_fs_allowed;
563 rctl_qty_t zone_nprocs; /* number of processes in the zone */
564 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
565 /* zone_rctls->rcs_lock */
566 kstat_t *zone_nprocs_kstat;
567
568 /*
569 * Misc. kstats and counters for zone cpu-usage aggregation.
570 * The zone_Xtime values are the sum of the micro-state accounting
571 * values for all threads that are running or have run in the zone.
572 * This is tracked in msacct.c as threads change state.
573 * The zone_stime is the sum of the LMS_SYSTEM times.
574 * The zone_utime is the sum of the LMS_USER times.
575 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
576 * As with per-thread micro-state accounting values, these values are
577 * not scaled to nanosecs. The scaling is done by the
578 * zone_misc_kstat_update function when kstats are requested.
579 */
580 kmutex_t zone_misc_lock; /* protects misc statistics */
581 kstat_t *zone_misc_ksp;
582 zone_misc_kstat_t *zone_misc_stats;
583 uint64_t zone_stime; /* total system time */
584 uint64_t zone_utime; /* total user time */
585 uint64_t zone_wtime; /* total time waiting in runq */
586 /* fork-fail kstat tracking */
587 uint32_t zone_ffcap; /* hit an rctl cap */
588 uint32_t zone_ffnoproc; /* get proc/lwp error */
589 uint32_t zone_ffnomem; /* as_dup/memory error */
590 uint32_t zone_ffmisc; /* misc. other error */
591
592 struct loadavg_s zone_loadavg; /* loadavg for this zone */
593 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
594 int zone_avenrun[3]; /* FSCALED avg. run queue len */
595
596 /*
597 * FSS stats updated once per second by fss_decay_usage.
598 */
599 uint32_t zone_fss_gen; /* FSS generation cntr */
600 uint64_t zone_run_ticks; /* tot # of ticks running */
601
602 /*
603 * DTrace-private per-zone state
604 */
605 int zone_dtrace_getf; /* # of unprivileged getf()s */
606
607 /*
608 * Synchronization primitives used to synchronize between mounts and
609 * zone creation/destruction.
610 */
611 int zone_mounts_in_progress;
612 kcondvar_t zone_mount_cv;
613 kmutex_t zone_mount_lock;
614 } zone_t;
615
616 /*
617 * Special value of zone_psetid to indicate that pools are disabled.
618 */
619 #define ZONE_PS_INVAL PS_MYID
620
621
622 extern zone_t zone0;
623 extern zone_t *global_zone;
624 extern uint_t maxzones;
625 extern rctl_hndl_t rc_zone_nlwps;
626 extern rctl_hndl_t rc_zone_nprocs;
627
628 extern long zone(int, void *, void *, void *, void *);
629 extern void zone_zsd_init(void);
630 extern void zone_init(void);
631 extern void zone_hold(zone_t *);
632 extern void zone_rele(zone_t *);
633 extern void zone_init_ref(zone_ref_t *);
634 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
635 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
636 extern void zone_cred_hold(zone_t *);
637 extern void zone_cred_rele(zone_t *);
638 extern void zone_task_hold(zone_t *);
639 extern void zone_task_rele(zone_t *);
640 extern zone_t *zone_find_by_id(zoneid_t);
641 extern zone_t *zone_find_by_label(const ts_label_t *);
642 extern zone_t *zone_find_by_name(char *);
643 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
644 extern zone_t *zone_find_by_path(const char *);
645 extern zoneid_t getzoneid(void);
646 extern zone_t *zone_find_by_id_nolock(zoneid_t);
647 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
648 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
649 extern void zone_loadavg_update();
650
651 /*
652 * Zone-specific data (ZSD) APIs
653 */
654 /*
655 * The following is what code should be initializing its zone_key_t to if it
656 * calls zone_getspecific() without necessarily knowing that zone_key_create()
657 * has been called on the key.
658 */
659 #define ZONE_KEY_UNINITIALIZED 0
660
661 typedef uint_t zone_key_t;
662
663 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
664 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
665 extern int zone_key_delete(zone_key_t);
666 extern void *zone_getspecific(zone_key_t, zone_t *);
667 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
668
669 /*
670 * The definition of a zsd_entry is truly private to zone.c and is only
671 * placed here so it can be shared with mdb.
672 *
673 * State maintained for each zone times each registered key, which tracks
674 * the state of the create, shutdown and destroy callbacks.
675 *
676 * zsd_flags is used to keep track of pending actions to avoid holding locks
677 * when calling the create/shutdown/destroy callbacks, since doing so
678 * could lead to deadlocks.
679 */
680 struct zsd_entry {
681 zone_key_t zsd_key; /* Key used to lookup value */
682 void *zsd_data; /* Caller-managed value */
683 /*
684 * Callbacks to be executed when a zone is created, shutdown, and
685 * destroyed, respectively.
686 */
687 void *(*zsd_create)(zoneid_t);
688 void (*zsd_shutdown)(zoneid_t, void *);
689 void (*zsd_destroy)(zoneid_t, void *);
690 list_node_t zsd_linkage;
691 uint16_t zsd_flags; /* See below */
692 kcondvar_t zsd_cv;
693 };
694
695 /*
696 * zsd_flags
697 */
698 #define ZSD_CREATE_NEEDED 0x0001
699 #define ZSD_CREATE_INPROGRESS 0x0002
700 #define ZSD_CREATE_COMPLETED 0x0004
701 #define ZSD_SHUTDOWN_NEEDED 0x0010
702 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
703 #define ZSD_SHUTDOWN_COMPLETED 0x0040
704 #define ZSD_DESTROY_NEEDED 0x0100
705 #define ZSD_DESTROY_INPROGRESS 0x0200
706 #define ZSD_DESTROY_COMPLETED 0x0400
707
708 #define ZSD_CREATE_ALL \
709 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
710 #define ZSD_SHUTDOWN_ALL \
711 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
712 #define ZSD_DESTROY_ALL \
713 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
714
715 #define ZSD_ALL_INPROGRESS \
716 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
717
718 /*
719 * Macros to help with zone visibility restrictions.
720 */
721
722 /*
723 * Is process in the global zone?
724 */
725 #define INGLOBALZONE(p) \
726 ((p)->p_zone == global_zone)
727
728 /*
729 * Can process view objects in given zone?
730 */
731 #define HASZONEACCESS(p, zoneid) \
732 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
733
734 /*
735 * Convenience macro to see if a resolved path is visible from within a
736 * given zone.
737 *
738 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
739 * two strings must be equal. Since the rootpathlen has a trailing '/',
740 * we want to skip everything in the path up to (but not including) the
741 * trailing '/'.
742 */
743 #define ZONE_PATH_VISIBLE(path, zone) \
744 (strncmp((path), (zone)->zone_rootpath, \
745 (zone)->zone_rootpathlen - 1) == 0)
746
747 /*
748 * Convenience macro to go from the global view of a path to that seen
749 * from within said zone. It is the responsibility of the caller to
750 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
751 * in fact visible from within the zone.
752 */
753 #define ZONE_PATH_TRANSLATE(path, zone) \
754 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
755 (path) + (zone)->zone_rootpathlen - 2)
756
757 /*
758 * Special processes visible in all zones.
759 */
760 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
761
762 /*
763 * Zone-safe version of thread_create() to be used when the caller wants to
764 * create a kernel thread to run within the current zone's context.
765 */
766 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
767 pri_t);
768 extern void zthread_exit(void);
769
770 /*
771 * Functions for an external observer to register interest in a zone's status
772 * change. Observers will be woken up when the zone status equals the status
773 * argument passed in (in the case of zone_status_timedwait, the function may
774 * also return because of a timeout; zone_status_wait_sig may return early due
775 * to a signal being delivered; zone_status_timedwait_sig may return for any of
776 * the above reasons).
777 *
778 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
779 * cv_wait_sig() respectively.
780 */
781 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
782 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
783 extern void zone_status_wait(zone_t *, zone_status_t);
784 extern int zone_status_wait_sig(zone_t *, zone_status_t);
785
786 /*
787 * Get the status of the zone (at the time it was called). The state may
788 * have progressed by the time it is returned.
789 */
790 extern zone_status_t zone_status_get(zone_t *);
791
792 /*
793 * Safely get the hostid of the specified zone (defaults to machine's hostid
794 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
795 * the global zone's (i.e., physical system's) hostid.
796 */
797 extern uint32_t zone_get_hostid(zone_t *);
798
799 /*
800 * Get the "kcred" credentials corresponding to the given zone.
801 */
802 extern struct cred *zone_get_kcred(zoneid_t);
803
804 /*
805 * Get/set the pool the zone is currently bound to.
806 */
807 extern struct pool *zone_pool_get(zone_t *);
808 extern void zone_pool_set(zone_t *, struct pool *);
809
810 /*
811 * Get/set the pset the zone is currently using.
812 */
813 extern psetid_t zone_pset_get(zone_t *);
814 extern void zone_pset_set(zone_t *, psetid_t);
815
816 /*
817 * Get the number of cpus/online-cpus visible from the given zone.
818 */
819 extern int zone_ncpus_get(zone_t *);
820 extern int zone_ncpus_online_get(zone_t *);
821
822 /*
823 * Returns true if the named pool/dataset is visible in the current zone.
824 */
825 extern int zone_dataset_visible(const char *, int *);
826
827 /*
828 * zone version of kadmin()
829 */
830 extern int zone_kadmin(int, int, const char *, cred_t *);
831 extern void zone_shutdown_global(void);
832
833 extern void mount_in_progress(zone_t *);
834 extern void mount_completed(zone_t *);
835
836 extern int zone_walk(int (*)(zone_t *, void *), void *);
837
838 extern rctl_hndl_t rc_zone_locked_mem;
839 extern rctl_hndl_t rc_zone_max_swap;
840 extern rctl_hndl_t rc_zone_max_lofi;
841
842 #endif /* _KERNEL */
843
844 #ifdef __cplusplus
845 }
846 #endif
847
848 #endif /* _SYS_ZONE_H */