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 2014 Nexenta Systems, Inc. All rights reserved.
24 * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
25 * Copyright 2016, Joyent, Inc.
26 */
27
28 #ifndef _SYS_ZONE_H
29 #define _SYS_ZONE_H
30
31 #include <sys/types.h>
32 #include <sys/mutex.h>
33 #include <sys/param.h>
34 #include <sys/rctl.h>
35 #include <sys/ipc_rctl.h>
36 #include <sys/pset.h>
37 #include <sys/tsol/label.h>
38 #include <sys/cred.h>
39 #include <sys/netstack.h>
40 #include <sys/uadmin.h>
41 #include <sys/ksynch.h>
42 #include <sys/socket_impl.h>
43 #include <netinet/in.h>
44
45 #ifdef __cplusplus
46 extern "C" {
47 #endif
48
49 /*
50 * NOTE
51 *
52 * The contents of this file are private to the implementation of
53 * Solaris and are subject to change at any time without notice.
54 * Applications and drivers using these interfaces may fail to
55 * run on future releases.
56 */
57
58 /* Available both in kernel and for user space */
59
60 /* zone id restrictions and special ids */
61 #define MAX_ZONEID 9999
62 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */
63 #define MIN_ZONEID 0 /* minimum zone ID on system */
64 #define GLOBAL_ZONEID 0
65 #define ZONEID_WIDTH 4 /* for printf */
66
67 /*
68 * Special zoneid_t token to refer to all zones.
69 */
70 #define ALL_ZONES (-1)
71
72 /* system call subcodes */
73 #define ZONE_CREATE 0
74 #define ZONE_DESTROY 1
75 #define ZONE_GETATTR 2
76 #define ZONE_ENTER 3
77 #define ZONE_LIST 4
78 #define ZONE_SHUTDOWN 5
79 #define ZONE_LOOKUP 6
80 #define ZONE_BOOT 7
81 #define ZONE_VERSION 8
82 #define ZONE_SETATTR 9
83 #define ZONE_ADD_DATALINK 10
84 #define ZONE_DEL_DATALINK 11
85 #define ZONE_CHECK_DATALINK 12
86 #define ZONE_LIST_DATALINK 13
87
88 /* zone attributes */
89 #define ZONE_ATTR_ROOT 1
90 #define ZONE_ATTR_NAME 2
91 #define ZONE_ATTR_STATUS 3
92 #define ZONE_ATTR_PRIVSET 4
93 #define ZONE_ATTR_UNIQID 5
94 #define ZONE_ATTR_POOLID 6
95 #define ZONE_ATTR_INITPID 7
96 #define ZONE_ATTR_SLBL 8
97 #define ZONE_ATTR_INITNAME 9
98 #define ZONE_ATTR_BOOTARGS 10
99 #define ZONE_ATTR_BRAND 11
100 #define ZONE_ATTR_PMCAP_NOVER 12
101 #define ZONE_ATTR_SCHED_CLASS 13
102 #define ZONE_ATTR_FLAGS 14
103 #define ZONE_ATTR_HOSTID 15
104 #define ZONE_ATTR_FS_ALLOWED 16
105 #define ZONE_ATTR_NETWORK 17
106 #define ZONE_ATTR_DID 18
107 #define ZONE_ATTR_PMCAP_PAGEOUT 19
108 #define ZONE_ATTR_INITNORESTART 20
109 #define ZONE_ATTR_PG_FLT_DELAY 21
110 #define ZONE_ATTR_RSS 22
111 #define ZONE_ATTR_APP_SVC_CT 23
112 #define ZONE_ATTR_SCHED_FIXEDHI 24
113
114 /* Start of the brand-specific attribute namespace */
115 #define ZONE_ATTR_BRAND_ATTRS 32768
116
117 #define ZONE_FS_ALLOWED_MAX 1024
118
119 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
120 #define ZONE_EVENT_STATUS_CLASS "status"
121 #define ZONE_EVENT_STATUS_SUBCLASS "change"
122
123 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
124 #define ZONE_EVENT_INITIALIZED "initialized"
125 #define ZONE_EVENT_READY "ready"
126 #define ZONE_EVENT_RUNNING "running"
127 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
128
129 #define ZONE_CB_NAME "zonename"
130 #define ZONE_CB_NEWSTATE "newstate"
131 #define ZONE_CB_OLDSTATE "oldstate"
132 #define ZONE_CB_TIMESTAMP "when"
133 #define ZONE_CB_ZONEID "zoneid"
134
135 /*
136 * Exit values that may be returned by scripts or programs invoked by various
137 * zone commands.
138 *
139 * These are defined as:
140 *
141 * ZONE_SUBPROC_OK
142 * ===============
143 * The subprocess completed successfully.
144 *
145 * ZONE_SUBPROC_USAGE
146 * ==================
147 * The subprocess failed with a usage message, or a usage message should
148 * be output in its behalf.
149 *
150 * ZONE_SUBPROC_NOTCOMPLETE
151 * ========================
152 * The subprocess did not complete, but the actions performed by the
153 * subprocess require no recovery actions by the user.
154 *
155 * For example, if the subprocess were called by "zoneadm install," the
156 * installation of the zone did not succeed but the user need not perform
157 * a "zoneadm uninstall" before attempting another install.
158 *
159 * ZONE_SUBPROC_FATAL
160 * ==================
161 * The subprocess failed in a fatal manner, usually one that will require
162 * some type of recovery action by the user.
163 *
164 * For example, if the subprocess were called by "zoneadm install," the
165 * installation of the zone did not succeed and the user will need to
166 * perform a "zoneadm uninstall" before another install attempt is
167 * possible.
168 *
169 * The non-success exit values are large to avoid accidental collision
170 * with values used internally by some commands (e.g. "Z_ERR" and
171 * "Z_USAGE" as used by zoneadm.)
172 */
173 #define ZONE_SUBPROC_OK 0
174 #define ZONE_SUBPROC_USAGE 253
175 #define ZONE_SUBPROC_NOTCOMPLETE 254
176 #define ZONE_SUBPROC_FATAL 255
177
178 #ifdef _SYSCALL32
179 typedef struct {
180 caddr32_t zone_name;
181 caddr32_t zone_root;
182 caddr32_t zone_privs;
183 size32_t zone_privssz;
184 caddr32_t rctlbuf;
185 size32_t rctlbufsz;
186 caddr32_t extended_error;
187 caddr32_t zfsbuf;
188 size32_t zfsbufsz;
189 int match; /* match level */
190 uint32_t doi; /* DOI for label */
191 caddr32_t label; /* label associated with zone */
192 int flags;
193 zoneid_t zoneid; /* requested zoneid */
194 } zone_def32;
195 #endif
196 typedef struct {
197 const char *zone_name;
198 const char *zone_root;
199 const struct priv_set *zone_privs;
200 size_t zone_privssz;
201 const char *rctlbuf;
202 size_t rctlbufsz;
203 int *extended_error;
204 const char *zfsbuf;
205 size_t zfsbufsz;
206 int match; /* match level */
207 uint32_t doi; /* DOI for label */
208 const bslabel_t *label; /* label associated with zone */
209 int flags;
210 zoneid_t zoneid; /* requested zoneid */
211 } zone_def;
212
213 /* extended error information */
214 #define ZE_UNKNOWN 0 /* No extended error info */
215 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
216 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
217 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
218
219 /*
220 * zone_status values
221 *
222 * You must modify zone_status_names in mdb(1M)'s genunix module
223 * (genunix/zone.c) when you modify this enum.
224 */
225 typedef enum {
226 ZONE_IS_UNINITIALIZED = 0,
227 ZONE_IS_INITIALIZED,
228 ZONE_IS_READY,
229 ZONE_IS_BOOTING,
230 ZONE_IS_RUNNING,
231 ZONE_IS_SHUTTING_DOWN,
232 ZONE_IS_EMPTY,
233 ZONE_IS_DOWN,
234 ZONE_IS_DYING,
235 ZONE_IS_DEAD
236 } zone_status_t;
237 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
238 #define ZONE_MAX_STATE ZONE_IS_DEAD
239
240 /*
241 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
242 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
243 */
244 typedef enum zone_cmd {
245 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
246 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
247 } zone_cmd_t;
248
249 /*
250 * The structure of a request to zoneadmd.
251 */
252 typedef struct zone_cmd_arg {
253 uint64_t uniqid; /* unique "generation number" */
254 zone_cmd_t cmd; /* requested action */
255 int status; /* init status on shutdown */
256 uint32_t debug; /* enable brand hook debug */
257 char locale[MAXPATHLEN]; /* locale in which to render messages */
258 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
259 /* Needed for 32/64 zoneadm -> zoneadmd door arg size check. */
260 int pad;
261 } zone_cmd_arg_t;
262
263 /*
264 * Structure of zoneadmd's response to a request. A NULL return value means
265 * the caller should attempt to restart zoneadmd and retry.
266 */
267 typedef struct zone_cmd_rval {
268 int rval; /* return value of request */
269 char errbuf[1]; /* variable-sized buffer containing error messages */
270 } zone_cmd_rval_t;
271
272 /*
273 * The zone support infrastructure uses the zone name as a component
274 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
275 * in length, so ZONENAME_MAX is limited by that.
276 */
277 #define ZONENAME_MAX 64
278
279 #define GLOBAL_ZONENAME "global"
280
281 /*
282 * Extended Regular expression (see regex(5)) which matches all valid zone
283 * names.
284 */
285 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
286
287 /*
288 * Where the zones support infrastructure places temporary files.
289 */
290 #define ZONES_TMPDIR "/var/run/zones"
291
292 /*
293 * The path to the door used by clients to communicate with zoneadmd.
294 */
295 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
296
297
298 /* zone_flags */
299 /*
300 * Threads that read or write the following flag must hold zone_lock.
301 */
302 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
303
304 /*
305 * The following threads are set when the zone is created and never changed.
306 * Threads that test for these flags don't have to hold zone_lock.
307 */
308 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
309 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
310 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
311
312
313 /* zone_create flags */
314 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
315
316 /* zone network properties */
317 #define ZONE_NETWORK_ADDRESS 1
318 #define ZONE_NETWORK_DEFROUTER 2
319
320 #define ZONE_NET_ADDRNAME "address"
321 #define ZONE_NET_RTRNAME "route"
322
323 typedef struct zone_net_data {
324 int zn_type;
325 int zn_len;
326 datalink_id_t zn_linkid;
327 uint8_t zn_val[1];
328 } zone_net_data_t;
329
330
331 #ifdef _KERNEL
332
333 /*
334 * We need to protect the definition of 'list_t' from userland applications and
335 * libraries which may be defining ther own versions.
336 */
337 #include <sys/list.h>
338 #include <sys/loadavg.h>
339
340 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
341
342 struct pool;
343 struct brand;
344
345 /*
346 * Each of these constants identifies a kernel subsystem that acquires and
347 * releases zone references. Each subsystem that invokes
348 * zone_hold_ref() and zone_rele_ref() should specify the
349 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
350 * help users and developers quickly identify subsystems that stall zone
351 * shutdowns indefinitely.
352 *
353 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
354 * when you modify this enumeration.
355 */
356 typedef enum zone_ref_subsys {
357 ZONE_REF_NFS, /* NFS */
358 ZONE_REF_NFSV4, /* NFSv4 */
359 ZONE_REF_SMBFS, /* SMBFS */
360 ZONE_REF_MNTFS, /* MNTFS */
361 ZONE_REF_LOFI, /* LOFI devices */
362 ZONE_REF_VFS, /* VFS infrastructure */
363 ZONE_REF_IPC, /* IPC infrastructure */
364 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
365 } zone_ref_subsys_t;
366
367 /*
368 * zone_ref represents a general-purpose references to a zone. Each zone's
369 * references are linked into the zone's zone_t::zone_ref_list. This allows
370 * debuggers to walk zones' references.
371 */
372 typedef struct zone_ref {
373 struct zone *zref_zone; /* the zone to which the reference refers */
374 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
375 } zone_ref_t;
376
377 /*
378 * Structure to record list of ZFS datasets exported to a zone.
379 */
380 typedef struct zone_dataset {
381 char *zd_dataset;
382 list_node_t zd_linkage;
383 } zone_dataset_t;
384
385 /*
386 * structure for rctl zone kstats
387 */
388 typedef struct zone_kstat {
389 kstat_named_t zk_zonename;
390 kstat_named_t zk_usage;
391 kstat_named_t zk_value;
392 } zone_kstat_t;
393
394 struct cpucap;
395
396 typedef struct {
397 hrtime_t cycle_start;
398 uint_t cycle_cnt;
399 hrtime_t zone_avg_cnt;
400 } sys_zio_cntr_t;
401
402 typedef struct {
403 kstat_named_t zv_zonename;
404 kstat_named_t zv_nread;
405 kstat_named_t zv_reads;
406 kstat_named_t zv_rtime;
407 kstat_named_t zv_rlentime;
408 kstat_named_t zv_rcnt;
409 kstat_named_t zv_nwritten;
410 kstat_named_t zv_writes;
411 kstat_named_t zv_wtime;
412 kstat_named_t zv_wlentime;
413 kstat_named_t zv_wcnt;
414 kstat_named_t zv_10ms_ops;
415 kstat_named_t zv_100ms_ops;
416 kstat_named_t zv_1s_ops;
417 kstat_named_t zv_10s_ops;
418 kstat_named_t zv_delay_cnt;
419 kstat_named_t zv_delay_time;
420 } zone_vfs_kstat_t;
421
422 typedef struct {
423 kstat_named_t zz_zonename;
424 kstat_named_t zz_nread;
425 kstat_named_t zz_reads;
426 kstat_named_t zz_rtime;
427 kstat_named_t zz_rlentime;
428 kstat_named_t zz_nwritten;
429 kstat_named_t zz_writes;
430 kstat_named_t zz_waittime;
431 } zone_zfs_kstat_t;
432
433 typedef struct {
434 kstat_named_t zm_zonename;
435 kstat_named_t zm_rss;
436 kstat_named_t zm_phys_cap;
437 kstat_named_t zm_swap;
438 kstat_named_t zm_swap_cap;
439 kstat_named_t zm_nover;
440 kstat_named_t zm_pagedout;
441 kstat_named_t zm_pgpgin;
442 kstat_named_t zm_anonpgin;
443 kstat_named_t zm_execpgin;
444 kstat_named_t zm_fspgin;
445 kstat_named_t zm_anon_alloc_fail;
446 kstat_named_t zm_pf_throttle;
447 kstat_named_t zm_pf_throttle_usec;
448 } zone_mcap_kstat_t;
449
450 typedef struct {
451 kstat_named_t zm_zonename; /* full name, kstat truncates name */
452 kstat_named_t zm_utime;
453 kstat_named_t zm_stime;
454 kstat_named_t zm_wtime;
455 kstat_named_t zm_avenrun1;
456 kstat_named_t zm_avenrun5;
457 kstat_named_t zm_avenrun15;
458 kstat_named_t zm_ffcap;
459 kstat_named_t zm_ffnoproc;
460 kstat_named_t zm_ffnomem;
461 kstat_named_t zm_ffmisc;
462 kstat_named_t zm_mfseglim;
463 kstat_named_t zm_nested_intp;
464 kstat_named_t zm_init_pid;
465 kstat_named_t zm_boot_time;
466 } zone_misc_kstat_t;
467
468 typedef struct zone {
469 /*
470 * zone_name is never modified once set.
471 */
472 char *zone_name; /* zone's configuration name */
473 /*
474 * zone_nodename and zone_domain are never freed once allocated.
475 */
476 char *zone_nodename; /* utsname.nodename equivalent */
477 char *zone_domain; /* srpc_domain equivalent */
478 /*
479 * zone_hostid is used for per-zone hostid emulation.
480 * Currently it isn't modified after it's set (so no locks protect
481 * accesses), but that might have to change when we allow
482 * administrators to change running zones' properties.
483 *
484 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
485 * that zone_get_hostid() will function correctly.
486 */
487 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
488 /* if not emulated */
489 /*
490 * zone_lock protects the following fields of a zone_t:
491 * zone_ref
492 * zone_cred_ref
493 * zone_subsys_ref
494 * zone_ref_list
495 * zone_ntasks
496 * zone_flags
497 * zone_zsd
498 * zone_pfexecd
499 */
500 kmutex_t zone_lock;
501 /*
502 * zone_linkage is the zone's linkage into the active or
503 * death-row list. The field is protected by zonehash_lock.
504 */
505 list_node_t zone_linkage;
506 zoneid_t zone_id; /* ID of zone */
507 zoneid_t zone_did; /* persistent debug ID of zone */
508 uint_t zone_ref; /* count of zone_hold()s on zone */
509 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
510 /*
511 * Fixed-sized array of subsystem-specific reference counts
512 * The sum of all of the counts must be less than or equal to zone_ref.
513 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
514 * constants.
515 */
516 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
517 list_t zone_ref_list; /* list of zone_ref_t structs */
518 /*
519 * zone_rootvp and zone_rootpath can never be modified once set.
520 */
521 struct vnode *zone_rootvp; /* zone's root vnode */
522 char *zone_rootpath; /* Path to zone's root + '/' */
523 ushort_t zone_flags; /* misc flags */
524 zone_status_t zone_status; /* protected by zone_status_lock */
525 uint_t zone_ntasks; /* number of tasks executing in zone */
526 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
527 /* counters in projects and tasks */
528 /* that are within the zone */
529 rctl_qty_t zone_nlwps; /* number of lwps in zone */
530 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
531 rctl_qty_t zone_shmmax; /* System V shared memory usage */
532 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
533
534 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
535 uint32_t zone_shares; /* FSS shares allocated to zone */
536 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
537 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
538 /* kpd_locked_mem for all */
539 /* projects in zone. */
540 /* Also protects zone_max_swap */
541 /* grab after p_lock, before rcs_lock */
542 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
543 /* zone */
544 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
545 /* limit. Protected by */
546 /* zone_rctls->rcs_lock */
547 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
548 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
549 /* Protected by */
550 /* zone_rctls->rcs_lock */
551 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
552 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
553 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
554 /* Protected by */
555 /* zone_rctls->rcs_lock */
556 list_t zone_zsd; /* list of Zone-Specific Data values */
557 kcondvar_t zone_cv; /* used to signal state changes */
558 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
559 pid_t zone_proc_initpid; /* pid of "init" for this zone */
560 char *zone_initname; /* fs path to 'init' */
561 int zone_init_status; /* init's exit status */
562 int zone_boot_err; /* for zone_boot() if boot fails */
563 char *zone_bootargs; /* arguments passed via zone_boot() */
564 rctl_qty_t zone_phys_mem_ctl; /* current phys. memory limit */
565 /*
566 * zone_kthreads is protected by zone_status_lock.
567 */
568 kthread_t *zone_kthreads; /* kernel threads in zone */
569 struct priv_set *zone_privset; /* limit set for zone */
570 /*
571 * zone_vfslist is protected by vfs_list_lock().
572 */
573 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
574 uint64_t zone_uniqid; /* unique zone generation number */
575 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
576 /*
577 * zone_pool is protected by pool_lock().
578 */
579 struct pool *zone_pool; /* pool the zone is bound to */
580 hrtime_t zone_pool_mod; /* last pool bind modification time */
581 /* zone_psetid is protected by cpu_lock */
582 psetid_t zone_psetid; /* pset the zone is bound to */
583
584 time_t zone_boot_time; /* Similar to boot_time */
585
586 /*
587 * The following two can be read without holding any locks. They are
588 * updated under cpu_lock.
589 */
590 int zone_ncpus; /* zone's idea of ncpus */
591 int zone_ncpus_online; /* zone's idea of ncpus_online */
592 /*
593 * List of ZFS datasets exported to this zone.
594 */
595 list_t zone_datasets; /* list of datasets */
596
597 ts_label_t *zone_slabel; /* zone sensitivity label */
598 int zone_match; /* require label match for packets */
599 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
600
601 boolean_t zone_restart_init; /* Restart init if it dies? */
602 boolean_t zone_reboot_on_init_exit; /* Reboot if init dies? */
603 boolean_t zone_setup_app_contract; /* setup contract? */
604 struct brand *zone_brand; /* zone's brand */
605 void *zone_brand_data; /* store brand specific data */
606 id_t zone_defaultcid; /* dflt scheduling class id */
607 boolean_t zone_fixed_hipri; /* fixed sched. hi prio */
608 kstat_t *zone_swapresv_kstat;
609 kstat_t *zone_lockedmem_kstat;
610 /*
611 * zone_dl_list is protected by zone_lock
612 */
613 list_t zone_dl_list;
614 netstack_t *zone_netstack;
615 struct cpucap *zone_cpucap; /* CPU caps data */
616
617 /*
618 * Data and counters used for ZFS fair-share disk IO.
619 */
620 rctl_qty_t zone_zfs_io_pri; /* ZFS IO priority */
621 uint_t zone_zfs_queued[2]; /* sync I/O enqueued count */
622 uint64_t zone_zfs_weight; /* used to prevent starvation */
623 uint64_t zone_io_util; /* IO utilization metric */
624 boolean_t zone_io_util_above_avg; /* IO util percent > avg. */
625 uint16_t zone_io_delay; /* IO delay on logical r/w */
626 kmutex_t zone_stg_io_lock; /* protects IO window data */
627 sys_zio_cntr_t zone_rd_ops; /* Counters for ZFS reads, */
628 sys_zio_cntr_t zone_wr_ops; /* writes and */
629 sys_zio_cntr_t zone_lwr_ops; /* logical writes. */
630
631 /*
632 * kstats and counters for VFS ops and bytes.
633 */
634 kmutex_t zone_vfs_lock; /* protects VFS statistics */
635 kstat_t *zone_vfs_ksp;
636 kstat_io_t zone_vfs_rwstats;
637 zone_vfs_kstat_t *zone_vfs_stats;
638
639 /*
640 * kstats for ZFS I/O ops and bytes.
641 */
642 kmutex_t zone_zfs_lock; /* protects ZFS statistics */
643 kstat_t *zone_zfs_ksp;
644 kstat_io_t zone_zfs_rwstats;
645 zone_zfs_kstat_t *zone_zfs_stats;
646
647 /*
648 * Solaris Auditing per-zone audit context
649 */
650 struct au_kcontext *zone_audit_kctxt;
651 /*
652 * For private use by mntfs.
653 */
654 struct mntelem *zone_mntfs_db;
655 krwlock_t zone_mntfs_db_lock;
656
657 struct klpd_reg *zone_pfexecd;
658
659 char *zone_fs_allowed;
660 rctl_qty_t zone_nprocs; /* number of processes in the zone */
661 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
662 /* zone_rctls->rcs_lock */
663 kstat_t *zone_nprocs_kstat;
664
665 /*
666 * kstats and counters for physical memory capping.
667 */
668 rctl_qty_t zone_phys_mem; /* current bytes of phys. mem. (RSS) */
669 kstat_t *zone_physmem_kstat;
670 uint64_t zone_mcap_nover; /* # of times over phys. cap */
671 uint64_t zone_mcap_pagedout; /* bytes of mem. paged out */
672 kmutex_t zone_mcap_lock; /* protects mcap statistics */
673 kstat_t *zone_mcap_ksp;
674 zone_mcap_kstat_t *zone_mcap_stats;
675 uint64_t zone_pgpgin; /* pages paged in */
676 uint64_t zone_anonpgin; /* anon pages paged in */
677 uint64_t zone_execpgin; /* exec pages paged in */
678 uint64_t zone_fspgin; /* fs pages paged in */
679 uint64_t zone_anon_alloc_fail; /* cnt of anon alloc fails */
680 uint64_t zone_pf_throttle; /* cnt of page flt throttles */
681 uint64_t zone_pf_throttle_usec; /* time of page flt throttles */
682
683 /* Num usecs to throttle page fault when zone is over phys. mem cap */
684 uint32_t zone_pg_flt_delay;
685
686 /*
687 * Misc. kstats and counters for zone cpu-usage aggregation.
688 * The zone_Xtime values are the sum of the micro-state accounting
689 * values for all threads that are running or have run in the zone.
690 * This is tracked in msacct.c as threads change state.
691 * The zone_stime is the sum of the LMS_SYSTEM times.
692 * The zone_utime is the sum of the LMS_USER times.
693 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
694 * As with per-thread micro-state accounting values, these values are
695 * not scaled to nanosecs. The scaling is done by the
696 * zone_misc_kstat_update function when kstats are requested.
697 */
698 kmutex_t zone_misc_lock; /* protects misc statistics */
699 kstat_t *zone_misc_ksp;
700 zone_misc_kstat_t *zone_misc_stats;
701 uint64_t zone_stime; /* total system time */
702 uint64_t zone_utime; /* total user time */
703 uint64_t zone_wtime; /* total time waiting in runq */
704 /* fork-fail kstat tracking */
705 uint32_t zone_ffcap; /* hit an rctl cap */
706 uint32_t zone_ffnoproc; /* get proc/lwp error */
707 uint32_t zone_ffnomem; /* as_dup/memory error */
708 uint32_t zone_ffmisc; /* misc. other error */
709
710 uint32_t zone_mfseglim; /* map failure (# segs limit) */
711
712 uint32_t zone_nested_intp; /* nested interp. kstat */
713
714 struct loadavg_s zone_loadavg; /* loadavg for this zone */
715 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
716 int zone_avenrun[3]; /* FSCALED avg. run queue len */
717
718 /*
719 * FSS stats updated once per second by fss_decay_usage.
720 */
721 uint32_t zone_fss_gen; /* FSS generation cntr */
722 uint64_t zone_run_ticks; /* tot # of ticks running */
723
724 /*
725 * DTrace-private per-zone state
726 */
727 int zone_dtrace_getf; /* # of unprivileged getf()s */
728
729 /*
730 * Synchronization primitives used to synchronize between mounts and
731 * zone creation/destruction.
732 */
733 int zone_mounts_in_progress;
734 kcondvar_t zone_mount_cv;
735 kmutex_t zone_mount_lock;
736 } zone_t;
737
738 /*
739 * Special value of zone_psetid to indicate that pools are disabled.
740 */
741 #define ZONE_PS_INVAL PS_MYID
742
743
744 extern zone_t zone0;
745 extern zone_t *global_zone;
746 extern uint_t maxzones;
747 extern rctl_hndl_t rc_zone_nlwps;
748 extern rctl_hndl_t rc_zone_nprocs;
749
750 extern long zone(int, void *, void *, void *, void *);
751 extern void zone_zsd_init(void);
752 extern void zone_init(void);
753 extern void zone_hold(zone_t *);
754 extern void zone_rele(zone_t *);
755 extern void zone_init_ref(zone_ref_t *);
756 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
757 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
758 extern void zone_cred_hold(zone_t *);
759 extern void zone_cred_rele(zone_t *);
760 extern void zone_task_hold(zone_t *);
761 extern void zone_task_rele(zone_t *);
762 extern zone_t *zone_find_by_id(zoneid_t);
763 extern zone_t *zone_find_by_label(const ts_label_t *);
764 extern zone_t *zone_find_by_name(char *);
765 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
766 extern zone_t *zone_find_by_path(const char *);
767 extern zoneid_t getzoneid(void);
768 extern zoneid_t getzonedid(void);
769 extern zone_t *zone_find_by_id_nolock(zoneid_t);
770 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
771 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
772 extern void zone_loadavg_update();
773
774 /*
775 * Zone-specific data (ZSD) APIs
776 */
777 /*
778 * The following is what code should be initializing its zone_key_t to if it
779 * calls zone_getspecific() without necessarily knowing that zone_key_create()
780 * has been called on the key.
781 */
782 #define ZONE_KEY_UNINITIALIZED 0
783
784 typedef uint_t zone_key_t;
785
786 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
787 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
788 extern int zone_key_delete(zone_key_t);
789 extern void *zone_getspecific(zone_key_t, zone_t *);
790 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
791
792 /*
793 * The definition of a zsd_entry is truly private to zone.c and is only
794 * placed here so it can be shared with mdb.
795 *
796 * State maintained for each zone times each registered key, which tracks
797 * the state of the create, shutdown and destroy callbacks.
798 *
799 * zsd_flags is used to keep track of pending actions to avoid holding locks
800 * when calling the create/shutdown/destroy callbacks, since doing so
801 * could lead to deadlocks.
802 */
803 struct zsd_entry {
804 zone_key_t zsd_key; /* Key used to lookup value */
805 void *zsd_data; /* Caller-managed value */
806 /*
807 * Callbacks to be executed when a zone is created, shutdown, and
808 * destroyed, respectively.
809 */
810 void *(*zsd_create)(zoneid_t);
811 void (*zsd_shutdown)(zoneid_t, void *);
812 void (*zsd_destroy)(zoneid_t, void *);
813 list_node_t zsd_linkage;
814 uint16_t zsd_flags; /* See below */
815 kcondvar_t zsd_cv;
816 };
817
818 /*
819 * zsd_flags
820 */
821 #define ZSD_CREATE_NEEDED 0x0001
822 #define ZSD_CREATE_INPROGRESS 0x0002
823 #define ZSD_CREATE_COMPLETED 0x0004
824 #define ZSD_SHUTDOWN_NEEDED 0x0010
825 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
826 #define ZSD_SHUTDOWN_COMPLETED 0x0040
827 #define ZSD_DESTROY_NEEDED 0x0100
828 #define ZSD_DESTROY_INPROGRESS 0x0200
829 #define ZSD_DESTROY_COMPLETED 0x0400
830
831 #define ZSD_CREATE_ALL \
832 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
833 #define ZSD_SHUTDOWN_ALL \
834 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
835 #define ZSD_DESTROY_ALL \
836 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
837
838 #define ZSD_ALL_INPROGRESS \
839 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
840
841 /*
842 * Macros to help with zone visibility restrictions.
843 */
844
845 /*
846 * Is process in the global zone?
847 */
848 #define INGLOBALZONE(p) \
849 ((p)->p_zone == global_zone)
850
851 /*
852 * Can process view objects in given zone?
853 */
854 #define HASZONEACCESS(p, zoneid) \
855 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
856
857 /*
858 * Convenience macro to see if a resolved path is visible from within a
859 * given zone.
860 *
861 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
862 * two strings must be equal. Since the rootpathlen has a trailing '/',
863 * we want to skip everything in the path up to (but not including) the
864 * trailing '/'.
865 */
866 #define ZONE_PATH_VISIBLE(path, zone) \
867 (strncmp((path), (zone)->zone_rootpath, \
868 (zone)->zone_rootpathlen - 1) == 0)
869
870 /*
871 * Convenience macro to go from the global view of a path to that seen
872 * from within said zone. It is the responsibility of the caller to
873 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
874 * in fact visible from within the zone.
875 */
876 #define ZONE_PATH_TRANSLATE(path, zone) \
877 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
878 (path) + (zone)->zone_rootpathlen - 2)
879
880 /*
881 * Special processes visible in all zones.
882 */
883 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
884
885 /*
886 * Zone-safe version of thread_create() to be used when the caller wants to
887 * create a kernel thread to run within the current zone's context.
888 */
889 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
890 pri_t);
891 extern void zthread_exit(void);
892
893 /*
894 * Functions for an external observer to register interest in a zone's status
895 * change. Observers will be woken up when the zone status equals the status
896 * argument passed in (in the case of zone_status_timedwait, the function may
897 * also return because of a timeout; zone_status_wait_sig may return early due
898 * to a signal being delivered; zone_status_timedwait_sig may return for any of
899 * the above reasons).
900 *
901 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
902 * cv_wait_sig() respectively.
903 */
904 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
905 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
906 extern void zone_status_wait(zone_t *, zone_status_t);
907 extern int zone_status_wait_sig(zone_t *, zone_status_t);
908
909 /*
910 * Get the status of the zone (at the time it was called). The state may
911 * have progressed by the time it is returned.
912 */
913 extern zone_status_t zone_status_get(zone_t *);
914
915 /*
916 * Safely get the hostid of the specified zone (defaults to machine's hostid
917 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
918 * the global zone's (i.e., physical system's) hostid.
919 */
920 extern uint32_t zone_get_hostid(zone_t *);
921
922 /*
923 * Get the "kcred" credentials corresponding to the given zone.
924 */
925 extern struct cred *zone_get_kcred(zoneid_t);
926
927 /*
928 * Get/set the pool the zone is currently bound to.
929 */
930 extern struct pool *zone_pool_get(zone_t *);
931 extern void zone_pool_set(zone_t *, struct pool *);
932
933 /*
934 * Get/set the pset the zone is currently using.
935 */
936 extern psetid_t zone_pset_get(zone_t *);
937 extern void zone_pset_set(zone_t *, psetid_t);
938
939 /*
940 * Get the number of cpus/online-cpus visible from the given zone.
941 */
942 extern int zone_ncpus_get(zone_t *);
943 extern int zone_ncpus_online_get(zone_t *);
944
945 /*
946 * Returns true if the named pool/dataset is visible in the current zone.
947 */
948 extern int zone_dataset_visible(const char *, int *);
949 extern int zone_dataset_visible_inzone(zone_t *, const char *, int *);
950
951 /*
952 * zone version of kadmin()
953 */
954 extern int zone_kadmin(int, int, const char *, cred_t *);
955 extern void zone_shutdown_global(void);
956
957 extern void mount_in_progress(zone_t *);
958 extern void mount_completed(zone_t *);
959
960 extern int zone_walk(int (*)(zone_t *, void *), void *);
961
962 extern rctl_hndl_t rc_zone_locked_mem;
963 extern rctl_hndl_t rc_zone_max_swap;
964 extern rctl_hndl_t rc_zone_phys_mem;
965 extern rctl_hndl_t rc_zone_max_lofi;
966
967 #endif /* _KERNEL */
968
969 #ifdef __cplusplus
970 }
971 #endif
972
973 #endif /* _SYS_ZONE_H */