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/nvpair.h>
54 #include <sys/list.h>
55 #include <sys/loadavg.h>
56 #endif /* _KERNEL */
57
58 #ifdef __cplusplus
59 extern "C" {
60 #endif
61
62 /*
63 * NOTE
64 *
65 * The contents of this file are private to the implementation of
66 * illumos and are subject to change at any time without notice.
67 * Applications and drivers using these interfaces may fail to
68 * run on future releases.
69 */
70
71 /* Available both in kernel and for user space */
72
73 /*
74 * zone id restrictions and special ids.
75 * See 'maxzones' for run-time zone limit.
76 *
77 * The current 8k value for MAX_ZONES was originally derived from the virtual
78 * interface limit in IP when "shared-stack" was the only supported networking
79 * for zones. The virtual interface limit is the number of addresses allowed
80 * on an interface (see MAX_ADDRS_PER_IF). Even with exclusive stacks, an 8k
81 * zone limit is still a reasonable choice at this time, given other limits
82 * within the kernel. Since we only support 8192 zones (which includes GZ),
83 * there is no point in allowing MAX_ZONEID > 8k.
84 */
85 #define MAX_ZONES 8192
86 #define MAX_ZONEID (MAX_ZONES - 1)
87 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */
88 #define MIN_ZONEID 0 /* minimum zone ID on system */
89 #define GLOBAL_ZONEID 0
90 #define ZONEID_WIDTH 4 /* for printf */
91
92 /*
93 * Special zoneid_t token to refer to all zones.
94 */
95 #define ALL_ZONES (-1)
96
97 /* system call subcodes */
98 #define ZONE_CREATE 0
99 #define ZONE_DESTROY 1
100 #define ZONE_GETATTR 2
101 #define ZONE_ENTER 3
102 #define ZONE_LIST 4
103 #define ZONE_SHUTDOWN 5
104 #define ZONE_LOOKUP 6
105 #define ZONE_BOOT 7
106 #define ZONE_VERSION 8
107 #define ZONE_SETATTR 9
108 #define ZONE_ADD_DATALINK 10
109 #define ZONE_DEL_DATALINK 11
110 #define ZONE_CHECK_DATALINK 12
111 #define ZONE_LIST_DATALINK 13
112
113 /* zone attributes */
114 #define ZONE_ATTR_ROOT 1
115 #define ZONE_ATTR_NAME 2
116 #define ZONE_ATTR_STATUS 3
117 #define ZONE_ATTR_PRIVSET 4
118 #define ZONE_ATTR_UNIQID 5
119 #define ZONE_ATTR_POOLID 6
120 #define ZONE_ATTR_INITPID 7
121 #define ZONE_ATTR_SLBL 8
122 #define ZONE_ATTR_INITNAME 9
123 #define ZONE_ATTR_BOOTARGS 10
124 #define ZONE_ATTR_BRAND 11
125 #define ZONE_ATTR_SCHED_CLASS 12
126 #define ZONE_ATTR_FLAGS 13
127 #define ZONE_ATTR_HOSTID 14
128 #define ZONE_ATTR_FS_ALLOWED 15
129 #define ZONE_ATTR_NETWORK 16
130 #define ZONE_ATTR_DID 17
131 #define ZONE_ATTR_INITNORESTART 18
132 #define ZONE_ATTR_APP_SVC_CT 19
133 #define ZONE_ATTR_SCHED_FIXEDHI 20
134 #define ZONE_ATTR_SECFLAGS 21
135 #define ZONE_ATTR_INITRESTART0 22
136 #define ZONE_ATTR_INITREBOOT 23
137
138 /* Start of the brand-specific attribute namespace */
139 #define ZONE_ATTR_BRAND_ATTRS 32768
140
141 #define ZONE_FS_ALLOWED_MAX 1024
142
143 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
144 #define ZONE_EVENT_STATUS_CLASS "status"
145 #define ZONE_EVENT_STATUS_SUBCLASS "change"
146
147 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
148 #define ZONE_EVENT_INITIALIZED "initialized"
149 #define ZONE_EVENT_READY "ready"
150 #define ZONE_EVENT_RUNNING "running"
151 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
152 #define ZONE_EVENT_FREE "free"
153
154 #define ZONE_CB_NAME "zonename"
155 #define ZONE_CB_NEWSTATE "newstate"
156 #define ZONE_CB_OLDSTATE "oldstate"
157 #define ZONE_CB_RESTARTS "restarts"
158 #define ZONE_CB_TIMESTAMP "when"
159 #define ZONE_CB_ZONEID "zoneid"
160
161 #define ZONE_EVENT_INIT_CLASS "init"
162 #define ZONE_EVENT_INIT_RESTART_SC "restart"
163
164 /*
165 * Exit values that may be returned by scripts or programs invoked by various
166 * zone commands.
167 *
168 * These are defined as:
169 *
170 * ZONE_SUBPROC_OK
171 * ===============
172 * The subprocess completed successfully.
173 *
174 * ZONE_SUBPROC_USAGE
175 * ==================
176 * The subprocess failed with a usage message, or a usage message should
177 * be output in its behalf.
178 *
179 * ZONE_SUBPROC_NOTCOMPLETE
180 * ========================
181 * The subprocess did not complete, but the actions performed by the
182 * subprocess require no recovery actions by the user.
183 *
184 * For example, if the subprocess were called by "zoneadm install," the
185 * installation of the zone did not succeed but the user need not perform
186 * a "zoneadm uninstall" before attempting another install.
187 *
188 * ZONE_SUBPROC_FATAL
189 * ==================
190 * The subprocess failed in a fatal manner, usually one that will require
191 * some type of recovery action by the user.
192 *
193 * For example, if the subprocess were called by "zoneadm install," the
194 * installation of the zone did not succeed and the user will need to
195 * perform a "zoneadm uninstall" before another install attempt is
196 * possible.
197 *
198 * The non-success exit values are large to avoid accidental collision
199 * with values used internally by some commands (e.g. "Z_ERR" and
200 * "Z_USAGE" as used by zoneadm.)
201 */
202 #define ZONE_SUBPROC_OK 0
203 #define ZONE_SUBPROC_USAGE 253
204 #define ZONE_SUBPROC_NOTCOMPLETE 254
205 #define ZONE_SUBPROC_FATAL 255
206
207 #ifdef _SYSCALL32
208 typedef struct {
209 caddr32_t zone_name;
210 caddr32_t zone_root;
211 caddr32_t zone_privs;
212 size32_t zone_privssz;
213 caddr32_t rctlbuf;
214 size32_t rctlbufsz;
215 caddr32_t extended_error;
216 caddr32_t zfsbuf;
217 size32_t zfsbufsz;
218 int match; /* match level */
219 uint32_t doi; /* DOI for label */
220 caddr32_t label; /* label associated with zone */
221 int flags;
222 zoneid_t zoneid; /* requested zoneid */
223 } zone_def32;
224 #endif
225 typedef struct {
226 const char *zone_name;
227 const char *zone_root;
228 const struct priv_set *zone_privs;
229 size_t zone_privssz;
230 const char *rctlbuf;
231 size_t rctlbufsz;
232 int *extended_error;
233 const char *zfsbuf;
234 size_t zfsbufsz;
235 int match; /* match level */
236 uint32_t doi; /* DOI for label */
237 const bslabel_t *label; /* label associated with zone */
238 int flags;
239 zoneid_t zoneid; /* requested zoneid */
240 } zone_def;
241
242 /* extended error information */
243 #define ZE_UNKNOWN 0 /* No extended error info */
244 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
245 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
246 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
247
248 /*
249 * zone_status values
250 *
251 * You must modify zone_status_names in mdb(1M)'s genunix module
252 * (genunix/zone.c) when you modify this enum.
253 */
254 typedef enum {
255 ZONE_IS_UNINITIALIZED = 0,
256 ZONE_IS_INITIALIZED,
257 ZONE_IS_READY,
258 ZONE_IS_BOOTING,
259 ZONE_IS_RUNNING,
260 ZONE_IS_SHUTTING_DOWN,
261 ZONE_IS_EMPTY,
262 ZONE_IS_DOWN,
263 ZONE_IS_DYING,
264 ZONE_IS_DEAD,
265 ZONE_IS_FREE /* transient state for zone sysevent */
266 } zone_status_t;
267 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
268 #define ZONE_MAX_STATE ZONE_IS_DEAD
269
270 /*
271 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
272 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
273 */
274 typedef enum zone_cmd {
275 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
276 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
277 } zone_cmd_t;
278
279 /*
280 * The structure of a request to zoneadmd.
281 */
282 typedef struct zone_cmd_arg {
283 uint64_t uniqid; /* unique "generation number" */
284 zone_cmd_t cmd; /* requested action */
285 int status; /* init status on shutdown */
286 uint32_t debug; /* enable brand hook debug */
287 char locale[MAXPATHLEN]; /* locale in which to render messages */
288 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
289 /* Needed for 32/64 zoneadm -> zoneadmd door arg size check. */
290 int pad;
291 } zone_cmd_arg_t;
292
293 /*
294 * Structure of zoneadmd's response to a request. A NULL return value means
295 * the caller should attempt to restart zoneadmd and retry.
296 */
297 typedef struct zone_cmd_rval {
298 int rval; /* return value of request */
299 char errbuf[1]; /* variable-sized buffer containing error messages */
300 } zone_cmd_rval_t;
301
302 /*
303 * The zone support infrastructure uses the zone name as a component
304 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
305 * in length, so ZONENAME_MAX is limited by that.
306 */
307 #define ZONENAME_MAX 64
308
309 #define GLOBAL_ZONENAME "global"
310
311 /*
312 * Extended Regular expression (see regex(5)) which matches all valid zone
313 * names.
314 */
315 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
316
317 /*
318 * Where the zones support infrastructure places temporary files.
319 */
320 #define ZONES_TMPDIR "/var/run/zones"
321
322 /*
323 * The path to the door used by clients to communicate with zoneadmd.
324 */
325 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
326
327
328 /* zone_flags */
329 /*
330 * Threads that read or write the following flag must hold zone_lock.
331 */
332 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
333
334 /*
335 * The following threads are set when the zone is created and never changed.
336 * Threads that test for these flags don't have to hold zone_lock.
337 */
338 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
339 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
340 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
341
342
343 /* zone_create flags */
344 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
345
346 /* zone network properties */
347 #define ZONE_NETWORK_ADDRESS 1
348 #define ZONE_NETWORK_DEFROUTER 2
349
350 #define ZONE_NET_ADDRNAME "address"
351 #define ZONE_NET_RTRNAME "route"
352
353 typedef struct zone_net_data {
354 int zn_type;
355 int zn_len;
356 datalink_id_t zn_linkid;
357 uint8_t zn_val[1];
358 } zone_net_data_t;
359
360
361 #ifdef _KERNEL
362
363 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
364
365 /*
366 * Indexes into ->zone_ustate array, summing the micro state of all threads in a
367 * particular zone.
368 */
369 #define ZONE_USTATE_STIME (0)
370 #define ZONE_USTATE_UTIME (1)
371 #define ZONE_USTATE_WTIME (2)
372 #define ZONE_USTATE_MAX (3)
373
374 struct pool;
375 struct brand;
376
377 /*
378 * Each of these constants identifies a kernel subsystem that acquires and
379 * releases zone references. Each subsystem that invokes
380 * zone_hold_ref() and zone_rele_ref() should specify the
381 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
382 * help users and developers quickly identify subsystems that stall zone
383 * shutdowns indefinitely.
384 *
385 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
386 * when you modify this enumeration.
387 */
388 typedef enum zone_ref_subsys {
389 ZONE_REF_NFS, /* NFS */
390 ZONE_REF_NFSV4, /* NFSv4 */
391 ZONE_REF_SMBFS, /* SMBFS */
392 ZONE_REF_MNTFS, /* MNTFS */
393 ZONE_REF_LOFI, /* LOFI devices */
394 ZONE_REF_VFS, /* VFS infrastructure */
395 ZONE_REF_IPC, /* IPC infrastructure */
396 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
397 } zone_ref_subsys_t;
398
399 /*
400 * zone_ref represents a general-purpose references to a zone. Each zone's
401 * references are linked into the zone's zone_t::zone_ref_list. This allows
402 * debuggers to walk zones' references.
403 */
404 typedef struct zone_ref {
405 struct zone *zref_zone; /* the zone to which the reference refers */
406 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
407 } zone_ref_t;
408
409 /*
410 * Structure to record list of ZFS datasets exported to a zone.
411 */
412 typedef struct zone_dataset {
413 char *zd_dataset;
414 list_node_t zd_linkage;
415 } zone_dataset_t;
416
417 /*
418 * structure for rctl zone kstats
419 */
420 typedef struct zone_kstat {
421 kstat_named_t zk_zonename;
422 kstat_named_t zk_usage;
423 kstat_named_t zk_value;
424 } zone_kstat_t;
425
426 struct cpucap;
427
428 typedef struct {
429 hrtime_t cycle_start;
430 uint_t cycle_cnt;
431 hrtime_t zone_avg_cnt;
432 } sys_zio_cntr_t;
433
434 typedef struct {
435 kstat_named_t zv_zonename;
436 kstat_named_t zv_nread;
437 kstat_named_t zv_reads;
438 kstat_named_t zv_rtime;
439 kstat_named_t zv_rlentime;
440 kstat_named_t zv_rcnt;
441 kstat_named_t zv_nwritten;
442 kstat_named_t zv_writes;
443 kstat_named_t zv_wtime;
444 kstat_named_t zv_wlentime;
445 kstat_named_t zv_wcnt;
446 kstat_named_t zv_10ms_ops;
447 kstat_named_t zv_100ms_ops;
448 kstat_named_t zv_1s_ops;
449 kstat_named_t zv_10s_ops;
450 kstat_named_t zv_delay_cnt;
451 kstat_named_t zv_delay_time;
452 } zone_vfs_kstat_t;
453
454 typedef struct {
455 kstat_named_t zz_zonename;
456 kstat_named_t zz_nread;
457 kstat_named_t zz_reads;
458 kstat_named_t zz_rtime;
459 kstat_named_t zz_rlentime;
460 kstat_named_t zz_nwritten;
461 kstat_named_t zz_writes;
462 kstat_named_t zz_waittime;
463 } zone_zfs_kstat_t;
464
465 typedef struct {
466 kstat_named_t zm_zonename;
467 kstat_named_t zm_rss;
468 kstat_named_t zm_phys_cap;
469 kstat_named_t zm_swap;
470 kstat_named_t zm_swap_cap;
471 kstat_named_t zm_nover;
472 kstat_named_t zm_pagedout;
473 kstat_named_t zm_pgpgin;
474 kstat_named_t zm_anonpgin;
475 kstat_named_t zm_execpgin;
476 kstat_named_t zm_fspgin;
477 kstat_named_t zm_anon_alloc_fail;
478 kstat_named_t zm_pf_throttle;
479 kstat_named_t zm_pf_throttle_usec;
480 } zone_mcap_kstat_t;
481
482 typedef struct {
483 kstat_named_t zm_zonename; /* full name, kstat truncates name */
484 kstat_named_t zm_utime;
485 kstat_named_t zm_stime;
486 kstat_named_t zm_wtime;
487 kstat_named_t zm_avenrun1;
488 kstat_named_t zm_avenrun5;
489 kstat_named_t zm_avenrun15;
490 kstat_named_t zm_ffcap;
491 kstat_named_t zm_ffnoproc;
492 kstat_named_t zm_ffnomem;
493 kstat_named_t zm_ffmisc;
494 kstat_named_t zm_mfseglim;
495 kstat_named_t zm_nested_intp;
496 kstat_named_t zm_init_pid;
497 kstat_named_t zm_init_restarts;
498 kstat_named_t zm_boot_time;
499 } zone_misc_kstat_t;
500
501 typedef struct zone {
502 /*
503 * zone_name is never modified once set.
504 */
505 char *zone_name; /* zone's configuration name */
506 /*
507 * zone_nodename and zone_domain are never freed once allocated.
508 */
509 char *zone_nodename; /* utsname.nodename equivalent */
510 char *zone_domain; /* srpc_domain equivalent */
511 /*
512 * zone_hostid is used for per-zone hostid emulation.
513 * Currently it isn't modified after it's set (so no locks protect
514 * accesses), but that might have to change when we allow
515 * administrators to change running zones' properties.
516 *
517 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
518 * that zone_get_hostid() will function correctly.
519 */
520 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
521 /* if not emulated */
522 /*
523 * zone_lock protects the following fields of a zone_t:
524 * zone_ref
525 * zone_cred_ref
526 * zone_subsys_ref
527 * zone_ref_list
528 * zone_ntasks
529 * zone_flags
530 * zone_zsd
531 * zone_pfexecd
532 */
533 kmutex_t zone_lock;
534 /*
535 * zone_linkage is the zone's linkage into the active or
536 * death-row list. The field is protected by zonehash_lock.
537 */
538 list_node_t zone_linkage;
539 zoneid_t zone_id; /* ID of zone */
540 zoneid_t zone_did; /* persistent debug ID of zone */
541 uint_t zone_ref; /* count of zone_hold()s on zone */
542 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
543 /*
544 * Fixed-sized array of subsystem-specific reference counts
545 * The sum of all of the counts must be less than or equal to zone_ref.
546 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
547 * constants.
548 */
549 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
550 list_t zone_ref_list; /* list of zone_ref_t structs */
551 /*
552 * zone_rootvp and zone_rootpath can never be modified once set.
553 */
554 struct vnode *zone_rootvp; /* zone's root vnode */
555 char *zone_rootpath; /* Path to zone's root + '/' */
556 ushort_t zone_flags; /* misc flags */
557 zone_status_t zone_status; /* protected by zone_status_lock */
558 uint_t zone_ntasks; /* number of tasks executing in zone */
559 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
560 /* counters in projects and tasks */
561 /* that are within the zone */
562 rctl_qty_t zone_nlwps; /* number of lwps in zone */
563 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
564 rctl_qty_t zone_shmmax; /* System V shared memory usage */
565 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
566
567 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
568 uint32_t zone_shares; /* FSS shares allocated to zone */
569 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
570 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
571 /* kpd_locked_mem for all */
572 /* projects in zone. */
573 /* Also protects zone_max_swap */
574 /* grab after p_lock, before rcs_lock */
575 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
576 /* zone */
577 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
578 /* limit. Protected by */
579 /* zone_rctls->rcs_lock */
580 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
581 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
582 /* Protected by */
583 /* zone_rctls->rcs_lock */
584 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
585 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
586 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
587 /* Protected by */
588 /* zone_rctls->rcs_lock */
589 list_t zone_zsd; /* list of Zone-Specific Data values */
590 kcondvar_t zone_cv; /* used to signal state changes */
591 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
592 pid_t zone_proc_initpid; /* pid of "init" for this zone */
593 uint_t zone_proc_init_restarts; /* times init restarted */
594 char *zone_initname; /* fs path to 'init' */
595 int zone_init_status; /* init's exit status */
596 int zone_boot_err; /* for zone_boot() if boot fails */
597 char *zone_bootargs; /* arguments passed via zone_boot() */
598 /*
599 * zone_kthreads is protected by zone_status_lock.
600 */
601 kthread_t *zone_kthreads; /* kernel threads in zone */
602 struct priv_set *zone_privset; /* limit set for zone */
603 /*
604 * zone_vfslist is protected by vfs_list_lock().
605 */
606 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
607 uint64_t zone_uniqid; /* unique zone generation number */
608 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
609 /*
610 * zone_pool is protected by pool_lock().
611 */
612 struct pool *zone_pool; /* pool the zone is bound to */
613 hrtime_t zone_pool_mod; /* last pool bind modification time */
614 /* zone_psetid is protected by cpu_lock */
615 psetid_t zone_psetid; /* pset the zone is bound to */
616
617 time_t zone_boot_time; /* Similar to boot_time */
618
619 /*
620 * The following two can be read without holding any locks. They are
621 * updated under cpu_lock.
622 */
623 int zone_ncpus; /* zone's idea of ncpus */
624 int zone_ncpus_online; /* zone's idea of ncpus_online */
625 /*
626 * List of ZFS datasets exported to this zone.
627 */
628 list_t zone_datasets; /* list of datasets */
629
630 ts_label_t *zone_slabel; /* zone sensitivity label */
631 int zone_match; /* require label match for packets */
632 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
633
634 boolean_t zone_restart_init; /* Restart init if it dies? */
635 boolean_t zone_reboot_on_init_exit; /* Reboot if init dies? */
636 boolean_t zone_restart_init_0; /* Restart only if it exits 0 */
637 boolean_t zone_setup_app_contract; /* setup contract? */
638 struct brand *zone_brand; /* zone's brand */
639 void *zone_brand_data; /* store brand specific data */
640 id_t zone_defaultcid; /* dflt scheduling class id */
641 boolean_t zone_fixed_hipri; /* fixed sched. hi prio */
642 kstat_t *zone_swapresv_kstat;
643 kstat_t *zone_lockedmem_kstat;
644 /*
645 * zone_dl_list is protected by zone_lock
646 */
647 list_t zone_dl_list;
648 netstack_t *zone_netstack;
649 struct cpucap *zone_cpucap; /* CPU caps data */
650
651 /*
652 * kstats and counters for VFS ops and bytes.
653 */
654 kmutex_t zone_vfs_lock; /* protects VFS statistics */
655 kstat_t *zone_vfs_ksp;
656 kstat_io_t zone_vfs_rwstats;
657 zone_vfs_kstat_t *zone_vfs_stats;
658
659 /*
660 * kstats for ZFS I/O ops and bytes.
661 */
662 kmutex_t zone_zfs_lock; /* protects ZFS statistics */
663 kstat_t *zone_zfs_ksp;
664 zone_zfs_kstat_t *zone_zfs_stats;
665
666 /*
667 * illumos Auditing per-zone audit context
668 */
669 struct au_kcontext *zone_audit_kctxt;
670 /*
671 * For private use by mntfs.
672 */
673 struct mntelem *zone_mntfs_db;
674 krwlock_t zone_mntfs_db_lock;
675
676 struct klpd_reg *zone_pfexecd;
677
678 char *zone_fs_allowed;
679 rctl_qty_t zone_nprocs; /* number of processes in the zone */
680 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
681 /* zone_rctls->rcs_lock */
682 kstat_t *zone_nprocs_kstat;
683
684 /*
685 * kstats and counters for physical memory capping.
686 */
687 kstat_t *zone_physmem_kstat;
688 kmutex_t zone_mcap_lock; /* protects mcap statistics */
689 kstat_t *zone_mcap_ksp;
690 zone_mcap_kstat_t *zone_mcap_stats;
691 uint64_t zone_pgpgin; /* pages paged in */
692 uint64_t zone_anonpgin; /* anon pages paged in */
693 uint64_t zone_execpgin; /* exec pages paged in */
694 uint64_t zone_fspgin; /* fs pages paged in */
695 uint64_t zone_anon_alloc_fail; /* cnt of anon alloc fails */
696
697 psecflags_t zone_secflags; /* default zone security-flags */
698
699 /*
700 * Misc. kstats and counters for zone cpu-usage aggregation.
701 */
702 kmutex_t zone_misc_lock; /* protects misc statistics */
703 kstat_t *zone_misc_ksp;
704 zone_misc_kstat_t *zone_misc_stats;
705 /* Accumulated microstate for all threads in this zone. */
706 cpu_uarray_t *zone_ustate;
707 /* fork-fail kstat tracking */
708 uint32_t zone_ffcap; /* hit an rctl cap */
709 uint32_t zone_ffnoproc; /* get proc/lwp error */
710 uint32_t zone_ffnomem; /* as_dup/memory error */
711 uint32_t zone_ffmisc; /* misc. other error */
712
713 uint32_t zone_mfseglim; /* map failure (# segs limit) */
714
715 uint32_t zone_nested_intp; /* nested interp. kstat */
716
717 struct loadavg_s zone_loadavg; /* loadavg for this zone */
718 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
719 int zone_avenrun[3]; /* FSCALED avg. run queue len */
720
721 /*
722 * FSS stats updated once per second by fss_decay_usage.
723 */
724 uint32_t zone_fss_gen; /* FSS generation cntr */
725 uint64_t zone_run_ticks; /* tot # of ticks running */
726
727 /*
728 * DTrace-private per-zone state
729 */
730 int zone_dtrace_getf; /* # of unprivileged getf()s */
731
732 /*
733 * Synchronization primitives used to synchronize between mounts and
734 * zone creation/destruction.
735 */
736 int zone_mounts_in_progress;
737 kcondvar_t zone_mount_cv;
738 kmutex_t zone_mount_lock;
739 } zone_t;
740
741 /*
742 * Data and counters used for ZFS fair-share disk IO.
743 */
744 typedef struct zone_zfs_io {
745 uint16_t zpers_zfs_io_pri; /* ZFS IO priority - 16k max */
746 uint_t zpers_zfs_queued[2]; /* sync I/O enqueued count */
747 sys_zio_cntr_t zpers_rd_ops; /* Counters for ZFS reads, */
748 sys_zio_cntr_t zpers_wr_ops; /* writes, and */
749 sys_zio_cntr_t zpers_lwr_ops; /* logical writes. */
750 kstat_io_t zpers_zfs_rwstats;
751 uint64_t zpers_io_util; /* IO utilization metric */
752 uint64_t zpers_zfs_rd_waittime;
753 uint8_t zpers_io_delay; /* IO delay on logical r/w */
754 uint8_t zpers_zfs_weight; /* used to prevent starvation */
755 uint8_t zpers_io_util_above_avg; /* IO util percent > avg. */
756 } zone_zfs_io_t;
757
758 /*
759 * "Persistent" zone data which can be accessed idependently of the zone_t.
760 */
761 typedef struct zone_persist {
762 kmutex_t zpers_zfs_lock; /* Protects zpers_zfsp references */
763 zone_zfs_io_t *zpers_zfsp; /* ZFS fair-share IO data */
764 uint8_t zpers_over; /* currently over cap */
765 uint32_t zpers_pg_cnt; /* current RSS in pages */
766 uint32_t zpers_pg_limit; /* current RRS limit in pages */
767 uint32_t zpers_nover; /* # of times over phys. cap */
768 #ifndef DEBUG
769 uint64_t zpers_pg_out; /* # pages flushed */
770 #else
771 /*
772 * To conserve memory, some detailed kstats are only kept for DEBUG
773 * builds.
774 */
775 uint64_t zpers_zfs_rd_waittime;
776
777 uint64_t zpers_pg_anon; /* # clean anon pages flushed */
778 uint64_t zpers_pg_anondirty; /* # dirty anon pages flushed */
779 uint64_t zpers_pg_fs; /* # clean fs pages flushed */
780 uint64_t zpers_pg_fsdirty; /* # dirty fs pages flushed */
781 #endif
782 } zone_persist_t;
783
784 typedef enum zone_pageout_op {
785 ZPO_DIRTY, ZPO_FS, ZPO_ANON, ZPO_ANONDIRTY
786 } zone_pageout_op_t;
787
788 /*
789 * Special value of zone_psetid to indicate that pools are disabled.
790 */
791 #define ZONE_PS_INVAL PS_MYID
792
793 extern zone_t zone0;
794 extern zone_t *global_zone;
795 extern uint_t maxzones;
796 extern rctl_hndl_t rc_zone_nlwps;
797 extern rctl_hndl_t rc_zone_nprocs;
798
799 extern long zone(int, void *, void *, void *, void *);
800 extern void zone_zsd_init(void);
801 extern void zone_init(void);
802 extern void zone_hold(zone_t *);
803 extern void zone_rele(zone_t *);
804 extern void zone_init_ref(zone_ref_t *);
805 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
806 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
807 extern void zone_cred_hold(zone_t *);
808 extern void zone_cred_rele(zone_t *);
809 extern void zone_task_hold(zone_t *);
810 extern void zone_task_rele(zone_t *);
811 extern zone_t *zone_find_by_id(zoneid_t);
812 extern zone_t *zone_find_by_label(const ts_label_t *);
813 extern zone_t *zone_find_by_name(char *);
814 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
815 extern zone_t *zone_find_by_path(const char *);
816 extern zoneid_t getzoneid(void);
817 extern zoneid_t getzonedid(void);
818 extern zone_t *zone_find_by_id_nolock(zoneid_t);
819 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
820 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
821 extern void zone_loadavg_update();
822
823 /*
824 * Zone-specific data (ZSD) APIs
825 */
826 /*
827 * The following is what code should be initializing its zone_key_t to if it
828 * calls zone_getspecific() without necessarily knowing that zone_key_create()
829 * has been called on the key.
830 */
831 #define ZONE_KEY_UNINITIALIZED 0
832
833 typedef uint_t zone_key_t;
834
835 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
836 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
837 extern int zone_key_delete(zone_key_t);
838 extern void *zone_getspecific(zone_key_t, zone_t *);
839 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
840
841 /*
842 * The definition of a zsd_entry is truly private to zone.c and is only
843 * placed here so it can be shared with mdb.
844 *
845 * State maintained for each zone times each registered key, which tracks
846 * the state of the create, shutdown and destroy callbacks.
847 *
848 * zsd_flags is used to keep track of pending actions to avoid holding locks
849 * when calling the create/shutdown/destroy callbacks, since doing so
850 * could lead to deadlocks.
851 */
852 struct zsd_entry {
853 zone_key_t zsd_key; /* Key used to lookup value */
854 void *zsd_data; /* Caller-managed value */
855 /*
856 * Callbacks to be executed when a zone is created, shutdown, and
857 * destroyed, respectively.
858 */
859 void *(*zsd_create)(zoneid_t);
860 void (*zsd_shutdown)(zoneid_t, void *);
861 void (*zsd_destroy)(zoneid_t, void *);
862 list_node_t zsd_linkage;
863 uint16_t zsd_flags; /* See below */
864 kcondvar_t zsd_cv;
865 };
866
867 /*
868 * zsd_flags
869 */
870 #define ZSD_CREATE_NEEDED 0x0001
871 #define ZSD_CREATE_INPROGRESS 0x0002
872 #define ZSD_CREATE_COMPLETED 0x0004
873 #define ZSD_SHUTDOWN_NEEDED 0x0010
874 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
875 #define ZSD_SHUTDOWN_COMPLETED 0x0040
876 #define ZSD_DESTROY_NEEDED 0x0100
877 #define ZSD_DESTROY_INPROGRESS 0x0200
878 #define ZSD_DESTROY_COMPLETED 0x0400
879
880 #define ZSD_CREATE_ALL \
881 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
882 #define ZSD_SHUTDOWN_ALL \
883 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
884 #define ZSD_DESTROY_ALL \
885 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
886
887 #define ZSD_ALL_INPROGRESS \
888 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
889
890 /*
891 * Macros to help with zone visibility restrictions.
892 */
893
894 /*
895 * Is process in the global zone?
896 */
897 #define INGLOBALZONE(p) \
898 ((p)->p_zone == global_zone)
899
900 /*
901 * Can process view objects in given zone?
902 */
903 #define HASZONEACCESS(p, zoneid) \
904 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
905
906 /*
907 * Convenience macro to see if a resolved path is visible from within a
908 * given zone.
909 *
910 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
911 * two strings must be equal. Since the rootpathlen has a trailing '/',
912 * we want to skip everything in the path up to (but not including) the
913 * trailing '/'.
914 */
915 #define ZONE_PATH_VISIBLE(path, zone) \
916 (strncmp((path), (zone)->zone_rootpath, \
917 (zone)->zone_rootpathlen - 1) == 0)
918
919 /*
920 * Convenience macro to go from the global view of a path to that seen
921 * from within said zone. It is the responsibility of the caller to
922 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
923 * in fact visible from within the zone.
924 */
925 #define ZONE_PATH_TRANSLATE(path, zone) \
926 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
927 (path) + (zone)->zone_rootpathlen - 2)
928
929 /*
930 * Special processes visible in all zones.
931 */
932 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
933
934 /*
935 * A root vnode of the current zone.
936 *
937 * NOTE: It may be necessary (initialization time for file sharing where an
938 * NGZ loads a file-sharing kernel module that does zsd initialization) to NOT
939 * use this macro. One should ASSERT() that curzone == active ZSD (an
940 * ASSERTion that's not always true at ZSD initialization time) during regular
941 * use of this macro.
942 */
943 #define ZONE_ROOTVP() (curzone->zone_rootvp)
944
945 /*
946 * Since a zone's root isn't necessarily an actual filesystem boundary
947 * (i.e. VROOT may not be set on zone->zone_rootvp) we need to not assume it.
948 * This macro helps in checking if a vnode is the current zone's rootvp.
949 * NOTE: Using the VN_ prefix, even though it's defined here in zone.h.
950 * NOTE2: See above warning about ZONE_ROOTVP().
951 */
952 #define VN_IS_CURZONEROOT(vp) (VN_CMP(vp, ZONE_ROOTVP()))
953
954 /*
955 * Zone-safe version of thread_create() to be used when the caller wants to
956 * create a kernel thread to run within the current zone's context.
957 */
958 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
959 pri_t);
960 extern void zthread_exit(void);
961
962 /*
963 * Functions for an external observer to register interest in a zone's status
964 * change. Observers will be woken up when the zone status equals the status
965 * argument passed in (in the case of zone_status_timedwait, the function may
966 * also return because of a timeout; zone_status_wait_sig may return early due
967 * to a signal being delivered; zone_status_timedwait_sig may return for any of
968 * the above reasons).
969 *
970 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
971 * cv_wait_sig() respectively.
972 */
973 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
974 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
975 extern void zone_status_wait(zone_t *, zone_status_t);
976 extern int zone_status_wait_sig(zone_t *, zone_status_t);
977
978 /*
979 * Get the status of the zone (at the time it was called). The state may
980 * have progressed by the time it is returned.
981 */
982 extern zone_status_t zone_status_get(zone_t *);
983
984 /*
985 * Safely get the hostid of the specified zone (defaults to machine's hostid
986 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
987 * the global zone's (i.e., physical system's) hostid.
988 */
989 extern uint32_t zone_get_hostid(zone_t *);
990
991 /*
992 * Get the "kcred" credentials corresponding to the given zone.
993 */
994 extern struct cred *zone_get_kcred(zoneid_t);
995
996 /*
997 * Get/set the pool the zone is currently bound to.
998 */
999 extern struct pool *zone_pool_get(zone_t *);
1000 extern void zone_pool_set(zone_t *, struct pool *);
1001
1002 /*
1003 * Get/set the pset the zone is currently using.
1004 */
1005 extern psetid_t zone_pset_get(zone_t *);
1006 extern void zone_pset_set(zone_t *, psetid_t);
1007
1008 /*
1009 * Get the number of cpus/online-cpus visible from the given zone.
1010 */
1011 extern int zone_ncpus_get(zone_t *);
1012 extern int zone_ncpus_online_get(zone_t *);
1013
1014 /*
1015 * Returns true if the named pool/dataset is visible in the current zone.
1016 */
1017 extern int zone_dataset_visible(const char *, int *);
1018 extern int zone_dataset_visible_inzone(zone_t *, const char *, int *);
1019
1020 /*
1021 * zone version of kadmin()
1022 */
1023 extern int zone_kadmin(int, int, const char *, cred_t *);
1024 extern void zone_shutdown_global(void);
1025
1026 extern void mount_in_progress(zone_t *);
1027 extern void mount_completed(zone_t *);
1028
1029 extern int zone_walk(int (*)(zone_t *, void *), void *);
1030
1031 struct page;
1032 extern void zone_add_page(struct page *);
1033 extern void zone_rm_page(struct page *);
1034 extern void zone_pageout_stat(int, zone_pageout_op_t);
1035 extern void zone_get_physmem_data(int, pgcnt_t *, pgcnt_t *);
1036
1037 /* Interfaces for page scanning */
1038 extern uint_t zone_num_over_cap;
1039 extern zone_persist_t zone_pdata[MAX_ZONES];
1040
1041 extern rctl_hndl_t rc_zone_locked_mem;
1042 extern rctl_hndl_t rc_zone_max_swap;
1043 extern rctl_hndl_t rc_zone_phys_mem;
1044 extern rctl_hndl_t rc_zone_max_lofi;
1045
1046 /* For publishing sysevents related to a particular zone */
1047 extern void zone_sysevent_publish(zone_t *, const char *, const char *,
1048 nvlist_t *);
1049
1050 #endif /* _KERNEL */
1051
1052 #ifdef __cplusplus
1053 }
1054 #endif
1055
1056 #endif /* _SYS_ZONE_H */