Print this page
2594 implement graceful shutdown for local zones in zoneadm
| Split |
Close |
| Expand all |
| Collapse all |
--- old/usr/src/uts/common/sys/zone.h
+++ new/usr/src/uts/common/sys/zone.h
1 1 /*
2 2 * CDDL HEADER START
3 3 *
4 4 * The contents of this file are subject to the terms of the
5 5 * Common Development and Distribution License (the "License").
6 6 * You may not use this file except in compliance with the License.
7 7 *
8 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 9 * or http://www.opensolaris.org/os/licensing.
10 10 * See the License for the specific language governing permissions
11 11 * and limitations under the License.
12 12 *
|
↓ open down ↓ |
12 lines elided |
↑ open up ↑ |
13 13 * When distributing Covered Code, include this CDDL HEADER in each
14 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 15 * If applicable, add the following below this CDDL HEADER, with the
16 16 * fields enclosed by brackets "[]" replaced with your own identifying
17 17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 18 *
19 19 * CDDL HEADER END
20 20 */
21 21 /*
22 22 * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
23 + * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
23 24 */
24 25
25 26 #ifndef _SYS_ZONE_H
26 27 #define _SYS_ZONE_H
27 28
28 29 #include <sys/types.h>
29 30 #include <sys/mutex.h>
30 31 #include <sys/param.h>
31 32 #include <sys/rctl.h>
32 33 #include <sys/ipc_rctl.h>
33 34 #include <sys/pset.h>
34 35 #include <sys/tsol/label.h>
35 36 #include <sys/cred.h>
36 37 #include <sys/netstack.h>
37 38 #include <sys/uadmin.h>
38 39 #include <sys/ksynch.h>
39 40 #include <sys/socket_impl.h>
40 41 #include <netinet/in.h>
41 42
42 43 #ifdef __cplusplus
43 44 extern "C" {
44 45 #endif
45 46
46 47 /*
47 48 * NOTE
48 49 *
49 50 * The contents of this file are private to the implementation of
50 51 * Solaris and are subject to change at any time without notice.
51 52 * Applications and drivers using these interfaces may fail to
52 53 * run on future releases.
53 54 */
54 55
55 56 /* Available both in kernel and for user space */
56 57
57 58 /* zone id restrictions and special ids */
58 59 #define MAX_ZONEID 9999
59 60 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */
60 61 #define MIN_ZONEID 0 /* minimum zone ID on system */
61 62 #define GLOBAL_ZONEID 0
62 63 #define ZONEID_WIDTH 4 /* for printf */
63 64
64 65 /*
65 66 * Special zoneid_t token to refer to all zones.
66 67 */
67 68 #define ALL_ZONES (-1)
68 69
69 70 /* system call subcodes */
70 71 #define ZONE_CREATE 0
71 72 #define ZONE_DESTROY 1
72 73 #define ZONE_GETATTR 2
73 74 #define ZONE_ENTER 3
74 75 #define ZONE_LIST 4
75 76 #define ZONE_SHUTDOWN 5
76 77 #define ZONE_LOOKUP 6
77 78 #define ZONE_BOOT 7
78 79 #define ZONE_VERSION 8
79 80 #define ZONE_SETATTR 9
80 81 #define ZONE_ADD_DATALINK 10
81 82 #define ZONE_DEL_DATALINK 11
82 83 #define ZONE_CHECK_DATALINK 12
83 84 #define ZONE_LIST_DATALINK 13
84 85
85 86 /* zone attributes */
86 87 #define ZONE_ATTR_ROOT 1
87 88 #define ZONE_ATTR_NAME 2
88 89 #define ZONE_ATTR_STATUS 3
89 90 #define ZONE_ATTR_PRIVSET 4
90 91 #define ZONE_ATTR_UNIQID 5
91 92 #define ZONE_ATTR_POOLID 6
92 93 #define ZONE_ATTR_INITPID 7
93 94 #define ZONE_ATTR_SLBL 8
94 95 #define ZONE_ATTR_INITNAME 9
95 96 #define ZONE_ATTR_BOOTARGS 10
96 97 #define ZONE_ATTR_BRAND 11
97 98 #define ZONE_ATTR_PHYS_MCAP 12
98 99 #define ZONE_ATTR_SCHED_CLASS 13
99 100 #define ZONE_ATTR_FLAGS 14
100 101 #define ZONE_ATTR_HOSTID 15
101 102 #define ZONE_ATTR_FS_ALLOWED 16
102 103 #define ZONE_ATTR_NETWORK 17
103 104
104 105 /* Start of the brand-specific attribute namespace */
105 106 #define ZONE_ATTR_BRAND_ATTRS 32768
106 107
107 108 #define ZONE_FS_ALLOWED_MAX 1024
108 109
109 110 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
110 111 #define ZONE_EVENT_STATUS_CLASS "status"
111 112 #define ZONE_EVENT_STATUS_SUBCLASS "change"
112 113
113 114 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
114 115 #define ZONE_EVENT_INITIALIZED "initialized"
115 116 #define ZONE_EVENT_READY "ready"
116 117 #define ZONE_EVENT_RUNNING "running"
117 118 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
118 119
119 120 #define ZONE_CB_NAME "zonename"
120 121 #define ZONE_CB_NEWSTATE "newstate"
121 122 #define ZONE_CB_OLDSTATE "oldstate"
122 123 #define ZONE_CB_TIMESTAMP "when"
123 124 #define ZONE_CB_ZONEID "zoneid"
124 125
125 126 /*
126 127 * Exit values that may be returned by scripts or programs invoked by various
127 128 * zone commands.
128 129 *
129 130 * These are defined as:
130 131 *
131 132 * ZONE_SUBPROC_OK
132 133 * ===============
133 134 * The subprocess completed successfully.
134 135 *
135 136 * ZONE_SUBPROC_USAGE
136 137 * ==================
137 138 * The subprocess failed with a usage message, or a usage message should
138 139 * be output in its behalf.
139 140 *
140 141 * ZONE_SUBPROC_NOTCOMPLETE
141 142 * ========================
142 143 * The subprocess did not complete, but the actions performed by the
143 144 * subprocess require no recovery actions by the user.
144 145 *
145 146 * For example, if the subprocess were called by "zoneadm install," the
146 147 * installation of the zone did not succeed but the user need not perform
147 148 * a "zoneadm uninstall" before attempting another install.
148 149 *
149 150 * ZONE_SUBPROC_FATAL
150 151 * ==================
151 152 * The subprocess failed in a fatal manner, usually one that will require
152 153 * some type of recovery action by the user.
153 154 *
154 155 * For example, if the subprocess were called by "zoneadm install," the
155 156 * installation of the zone did not succeed and the user will need to
156 157 * perform a "zoneadm uninstall" before another install attempt is
157 158 * possible.
158 159 *
159 160 * The non-success exit values are large to avoid accidental collision
160 161 * with values used internally by some commands (e.g. "Z_ERR" and
161 162 * "Z_USAGE" as used by zoneadm.)
162 163 */
163 164 #define ZONE_SUBPROC_OK 0
164 165 #define ZONE_SUBPROC_USAGE 253
165 166 #define ZONE_SUBPROC_NOTCOMPLETE 254
166 167 #define ZONE_SUBPROC_FATAL 255
167 168
168 169 #ifdef _SYSCALL32
169 170 typedef struct {
170 171 caddr32_t zone_name;
171 172 caddr32_t zone_root;
172 173 caddr32_t zone_privs;
173 174 size32_t zone_privssz;
174 175 caddr32_t rctlbuf;
175 176 size32_t rctlbufsz;
176 177 caddr32_t extended_error;
177 178 caddr32_t zfsbuf;
178 179 size32_t zfsbufsz;
179 180 int match; /* match level */
180 181 uint32_t doi; /* DOI for label */
181 182 caddr32_t label; /* label associated with zone */
182 183 int flags;
183 184 } zone_def32;
184 185 #endif
185 186 typedef struct {
186 187 const char *zone_name;
187 188 const char *zone_root;
188 189 const struct priv_set *zone_privs;
189 190 size_t zone_privssz;
190 191 const char *rctlbuf;
191 192 size_t rctlbufsz;
192 193 int *extended_error;
193 194 const char *zfsbuf;
194 195 size_t zfsbufsz;
195 196 int match; /* match level */
196 197 uint32_t doi; /* DOI for label */
197 198 const bslabel_t *label; /* label associated with zone */
198 199 int flags;
199 200 } zone_def;
200 201
201 202 /* extended error information */
202 203 #define ZE_UNKNOWN 0 /* No extended error info */
203 204 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
204 205 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
205 206 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
206 207
207 208 /*
208 209 * zone_status values
209 210 *
210 211 * You must modify zone_status_names in mdb(1M)'s genunix module
211 212 * (genunix/zone.c) when you modify this enum.
212 213 */
213 214 typedef enum {
214 215 ZONE_IS_UNINITIALIZED = 0,
215 216 ZONE_IS_INITIALIZED,
216 217 ZONE_IS_READY,
217 218 ZONE_IS_BOOTING,
218 219 ZONE_IS_RUNNING,
219 220 ZONE_IS_SHUTTING_DOWN,
220 221 ZONE_IS_EMPTY,
221 222 ZONE_IS_DOWN,
222 223 ZONE_IS_DYING,
223 224 ZONE_IS_DEAD
|
↓ open down ↓ |
191 lines elided |
↑ open up ↑ |
224 225 } zone_status_t;
225 226 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
226 227 #define ZONE_MAX_STATE ZONE_IS_DEAD
227 228
228 229 /*
229 230 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
230 231 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
231 232 */
232 233 typedef enum zone_cmd {
233 234 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
234 - Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT
235 + Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
235 236 } zone_cmd_t;
236 237
237 238 /*
238 239 * The structure of a request to zoneadmd.
239 240 */
240 241 typedef struct zone_cmd_arg {
241 242 uint64_t uniqid; /* unique "generation number" */
242 243 zone_cmd_t cmd; /* requested action */
243 244 uint32_t _pad; /* need consistent 32/64 bit alignmt */
244 245 char locale[MAXPATHLEN]; /* locale in which to render messages */
245 246 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
246 247 } zone_cmd_arg_t;
247 248
248 249 /*
249 250 * Structure of zoneadmd's response to a request. A NULL return value means
250 251 * the caller should attempt to restart zoneadmd and retry.
251 252 */
252 253 typedef struct zone_cmd_rval {
253 254 int rval; /* return value of request */
254 255 char errbuf[1]; /* variable-sized buffer containing error messages */
255 256 } zone_cmd_rval_t;
256 257
257 258 /*
258 259 * The zone support infrastructure uses the zone name as a component
259 260 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
260 261 * in length, so ZONENAME_MAX is limited by that.
261 262 */
262 263 #define ZONENAME_MAX 64
263 264
264 265 #define GLOBAL_ZONENAME "global"
265 266
266 267 /*
267 268 * Extended Regular expression (see regex(5)) which matches all valid zone
268 269 * names.
269 270 */
270 271 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
271 272
272 273 /*
273 274 * Where the zones support infrastructure places temporary files.
274 275 */
275 276 #define ZONES_TMPDIR "/var/run/zones"
276 277
277 278 /*
278 279 * The path to the door used by clients to communicate with zoneadmd.
279 280 */
280 281 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
281 282
282 283
283 284 /* zone_flags */
284 285 /*
285 286 * Threads that read or write the following flag must hold zone_lock.
286 287 */
287 288 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
288 289
289 290 /*
290 291 * The following threads are set when the zone is created and never changed.
291 292 * Threads that test for these flags don't have to hold zone_lock.
292 293 */
293 294 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
294 295 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
295 296 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
296 297
297 298
298 299 /* zone_create flags */
299 300 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
300 301
301 302 /* zone network properties */
302 303 #define ZONE_NETWORK_ADDRESS 1
303 304 #define ZONE_NETWORK_DEFROUTER 2
304 305
305 306 #define ZONE_NET_ADDRNAME "address"
306 307 #define ZONE_NET_RTRNAME "route"
307 308
308 309 typedef struct zone_net_data {
309 310 int zn_type;
310 311 int zn_len;
311 312 datalink_id_t zn_linkid;
312 313 uint8_t zn_val[1];
313 314 } zone_net_data_t;
314 315
315 316
316 317 #ifdef _KERNEL
317 318
318 319 /*
319 320 * We need to protect the definition of 'list_t' from userland applications and
320 321 * libraries which may be defining ther own versions.
321 322 */
322 323 #include <sys/list.h>
323 324
324 325 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
325 326
326 327 struct pool;
327 328 struct brand;
328 329
329 330 /*
330 331 * Each of these constants identifies a kernel subsystem that acquires and
331 332 * releases zone references. Each subsystem that invokes
332 333 * zone_hold_ref() and zone_rele_ref() should specify the
333 334 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
334 335 * help users and developers quickly identify subsystems that stall zone
335 336 * shutdowns indefinitely.
336 337 *
337 338 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
338 339 * when you modify this enumeration.
339 340 */
340 341 typedef enum zone_ref_subsys {
341 342 ZONE_REF_NFS, /* NFS */
342 343 ZONE_REF_NFSV4, /* NFSv4 */
343 344 ZONE_REF_SMBFS, /* SMBFS */
344 345 ZONE_REF_MNTFS, /* MNTFS */
345 346 ZONE_REF_LOFI, /* LOFI devices */
346 347 ZONE_REF_VFS, /* VFS infrastructure */
347 348 ZONE_REF_IPC, /* IPC infrastructure */
348 349 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
349 350 } zone_ref_subsys_t;
350 351
351 352 /*
352 353 * zone_ref represents a general-purpose references to a zone. Each zone's
353 354 * references are linked into the zone's zone_t::zone_ref_list. This allows
354 355 * debuggers to walk zones' references.
355 356 */
356 357 typedef struct zone_ref {
357 358 struct zone *zref_zone; /* the zone to which the reference refers */
358 359 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
359 360 } zone_ref_t;
360 361
361 362 /*
362 363 * Structure to record list of ZFS datasets exported to a zone.
363 364 */
364 365 typedef struct zone_dataset {
365 366 char *zd_dataset;
366 367 list_node_t zd_linkage;
367 368 } zone_dataset_t;
368 369
369 370 /*
370 371 * structure for zone kstats
371 372 */
372 373 typedef struct zone_kstat {
373 374 kstat_named_t zk_zonename;
374 375 kstat_named_t zk_usage;
375 376 kstat_named_t zk_value;
376 377 } zone_kstat_t;
377 378
378 379 struct cpucap;
379 380
380 381 typedef struct zone {
381 382 /*
382 383 * zone_name is never modified once set.
383 384 */
384 385 char *zone_name; /* zone's configuration name */
385 386 /*
386 387 * zone_nodename and zone_domain are never freed once allocated.
387 388 */
388 389 char *zone_nodename; /* utsname.nodename equivalent */
389 390 char *zone_domain; /* srpc_domain equivalent */
390 391 /*
391 392 * zone_hostid is used for per-zone hostid emulation.
392 393 * Currently it isn't modified after it's set (so no locks protect
393 394 * accesses), but that might have to change when we allow
394 395 * administrators to change running zones' properties.
395 396 *
396 397 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
397 398 * that zone_get_hostid() will function correctly.
398 399 */
399 400 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
400 401 /* if not emulated */
401 402 /*
402 403 * zone_lock protects the following fields of a zone_t:
403 404 * zone_ref
404 405 * zone_cred_ref
405 406 * zone_subsys_ref
406 407 * zone_ref_list
407 408 * zone_ntasks
408 409 * zone_flags
409 410 * zone_zsd
410 411 * zone_pfexecd
411 412 */
412 413 kmutex_t zone_lock;
413 414 /*
414 415 * zone_linkage is the zone's linkage into the active or
415 416 * death-row list. The field is protected by zonehash_lock.
416 417 */
417 418 list_node_t zone_linkage;
418 419 zoneid_t zone_id; /* ID of zone */
419 420 uint_t zone_ref; /* count of zone_hold()s on zone */
420 421 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
421 422 /*
422 423 * Fixed-sized array of subsystem-specific reference counts
423 424 * The sum of all of the counts must be less than or equal to zone_ref.
424 425 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
425 426 * constants.
426 427 */
427 428 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
428 429 list_t zone_ref_list; /* list of zone_ref_t structs */
429 430 /*
430 431 * zone_rootvp and zone_rootpath can never be modified once set.
431 432 */
432 433 struct vnode *zone_rootvp; /* zone's root vnode */
433 434 char *zone_rootpath; /* Path to zone's root + '/' */
434 435 ushort_t zone_flags; /* misc flags */
435 436 zone_status_t zone_status; /* protected by zone_status_lock */
436 437 uint_t zone_ntasks; /* number of tasks executing in zone */
437 438 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
438 439 /* counters in projects and tasks */
439 440 /* that are within the zone */
440 441 rctl_qty_t zone_nlwps; /* number of lwps in zone */
441 442 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
442 443 rctl_qty_t zone_shmmax; /* System V shared memory usage */
443 444 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
444 445
445 446 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
446 447 uint32_t zone_shares; /* FSS shares allocated to zone */
447 448 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
448 449 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
449 450 /* kpd_locked_mem for all */
450 451 /* projects in zone. */
451 452 /* Also protects zone_max_swap */
452 453 /* grab after p_lock, before rcs_lock */
453 454 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
454 455 /* zone */
455 456 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
456 457 /* limit. Protected by */
457 458 /* zone_rctls->rcs_lock */
458 459 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
459 460 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
460 461 /* Protected by */
461 462 /* zone_rctls->rcs_lock */
462 463 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
463 464 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
464 465 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
465 466 /* Protected by */
466 467 /* zone_rctls->rcs_lock */
467 468 list_t zone_zsd; /* list of Zone-Specific Data values */
468 469 kcondvar_t zone_cv; /* used to signal state changes */
469 470 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
470 471 pid_t zone_proc_initpid; /* pid of "init" for this zone */
471 472 char *zone_initname; /* fs path to 'init' */
472 473 int zone_boot_err; /* for zone_boot() if boot fails */
473 474 char *zone_bootargs; /* arguments passed via zone_boot() */
474 475 uint64_t zone_phys_mcap; /* physical memory cap */
475 476 /*
476 477 * zone_kthreads is protected by zone_status_lock.
477 478 */
478 479 kthread_t *zone_kthreads; /* kernel threads in zone */
479 480 struct priv_set *zone_privset; /* limit set for zone */
480 481 /*
481 482 * zone_vfslist is protected by vfs_list_lock().
482 483 */
483 484 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
484 485 uint64_t zone_uniqid; /* unique zone generation number */
485 486 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
486 487 /*
487 488 * zone_pool is protected by pool_lock().
488 489 */
489 490 struct pool *zone_pool; /* pool the zone is bound to */
490 491 hrtime_t zone_pool_mod; /* last pool bind modification time */
491 492 /* zone_psetid is protected by cpu_lock */
492 493 psetid_t zone_psetid; /* pset the zone is bound to */
493 494 /*
494 495 * The following two can be read without holding any locks. They are
495 496 * updated under cpu_lock.
496 497 */
497 498 int zone_ncpus; /* zone's idea of ncpus */
498 499 int zone_ncpus_online; /* zone's idea of ncpus_online */
499 500 /*
500 501 * List of ZFS datasets exported to this zone.
501 502 */
502 503 list_t zone_datasets; /* list of datasets */
503 504
504 505 ts_label_t *zone_slabel; /* zone sensitivity label */
505 506 int zone_match; /* require label match for packets */
506 507 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
507 508
508 509 boolean_t zone_restart_init; /* Restart init if it dies? */
509 510 struct brand *zone_brand; /* zone's brand */
510 511 void *zone_brand_data; /* store brand specific data */
511 512 id_t zone_defaultcid; /* dflt scheduling class id */
512 513 kstat_t *zone_swapresv_kstat;
513 514 kstat_t *zone_lockedmem_kstat;
514 515 /*
515 516 * zone_dl_list is protected by zone_lock
516 517 */
517 518 list_t zone_dl_list;
518 519 netstack_t *zone_netstack;
519 520 struct cpucap *zone_cpucap; /* CPU caps data */
520 521 /*
521 522 * Solaris Auditing per-zone audit context
522 523 */
523 524 struct au_kcontext *zone_audit_kctxt;
524 525 /*
525 526 * For private use by mntfs.
526 527 */
527 528 struct mntelem *zone_mntfs_db;
528 529 krwlock_t zone_mntfs_db_lock;
529 530
530 531 struct klpd_reg *zone_pfexecd;
531 532
532 533 char *zone_fs_allowed;
533 534 rctl_qty_t zone_nprocs; /* number of processes in the zone */
534 535 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
535 536 /* zone_rctls->rcs_lock */
536 537 kstat_t *zone_nprocs_kstat;
537 538 } zone_t;
538 539
539 540 /*
540 541 * Special value of zone_psetid to indicate that pools are disabled.
541 542 */
542 543 #define ZONE_PS_INVAL PS_MYID
543 544
544 545
545 546 extern zone_t zone0;
546 547 extern zone_t *global_zone;
547 548 extern uint_t maxzones;
548 549 extern rctl_hndl_t rc_zone_nlwps;
549 550 extern rctl_hndl_t rc_zone_nprocs;
550 551
551 552 extern long zone(int, void *, void *, void *, void *);
552 553 extern void zone_zsd_init(void);
553 554 extern void zone_init(void);
554 555 extern void zone_hold(zone_t *);
555 556 extern void zone_rele(zone_t *);
556 557 extern void zone_init_ref(zone_ref_t *);
557 558 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
558 559 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
559 560 extern void zone_cred_hold(zone_t *);
560 561 extern void zone_cred_rele(zone_t *);
561 562 extern void zone_task_hold(zone_t *);
562 563 extern void zone_task_rele(zone_t *);
563 564 extern zone_t *zone_find_by_id(zoneid_t);
564 565 extern zone_t *zone_find_by_label(const ts_label_t *);
565 566 extern zone_t *zone_find_by_name(char *);
566 567 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
567 568 extern zone_t *zone_find_by_path(const char *);
568 569 extern zoneid_t getzoneid(void);
569 570 extern zone_t *zone_find_by_id_nolock(zoneid_t);
570 571 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
571 572 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
572 573
573 574 /*
574 575 * Zone-specific data (ZSD) APIs
575 576 */
576 577 /*
577 578 * The following is what code should be initializing its zone_key_t to if it
578 579 * calls zone_getspecific() without necessarily knowing that zone_key_create()
579 580 * has been called on the key.
580 581 */
581 582 #define ZONE_KEY_UNINITIALIZED 0
582 583
583 584 typedef uint_t zone_key_t;
584 585
585 586 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
586 587 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
587 588 extern int zone_key_delete(zone_key_t);
588 589 extern void *zone_getspecific(zone_key_t, zone_t *);
589 590 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
590 591
591 592 /*
592 593 * The definition of a zsd_entry is truly private to zone.c and is only
593 594 * placed here so it can be shared with mdb.
594 595 *
595 596 * State maintained for each zone times each registered key, which tracks
596 597 * the state of the create, shutdown and destroy callbacks.
597 598 *
598 599 * zsd_flags is used to keep track of pending actions to avoid holding locks
599 600 * when calling the create/shutdown/destroy callbacks, since doing so
600 601 * could lead to deadlocks.
601 602 */
602 603 struct zsd_entry {
603 604 zone_key_t zsd_key; /* Key used to lookup value */
604 605 void *zsd_data; /* Caller-managed value */
605 606 /*
606 607 * Callbacks to be executed when a zone is created, shutdown, and
607 608 * destroyed, respectively.
608 609 */
609 610 void *(*zsd_create)(zoneid_t);
610 611 void (*zsd_shutdown)(zoneid_t, void *);
611 612 void (*zsd_destroy)(zoneid_t, void *);
612 613 list_node_t zsd_linkage;
613 614 uint16_t zsd_flags; /* See below */
614 615 kcondvar_t zsd_cv;
615 616 };
616 617
617 618 /*
618 619 * zsd_flags
619 620 */
620 621 #define ZSD_CREATE_NEEDED 0x0001
621 622 #define ZSD_CREATE_INPROGRESS 0x0002
622 623 #define ZSD_CREATE_COMPLETED 0x0004
623 624 #define ZSD_SHUTDOWN_NEEDED 0x0010
624 625 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
625 626 #define ZSD_SHUTDOWN_COMPLETED 0x0040
626 627 #define ZSD_DESTROY_NEEDED 0x0100
627 628 #define ZSD_DESTROY_INPROGRESS 0x0200
628 629 #define ZSD_DESTROY_COMPLETED 0x0400
629 630
630 631 #define ZSD_CREATE_ALL \
631 632 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
632 633 #define ZSD_SHUTDOWN_ALL \
633 634 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
634 635 #define ZSD_DESTROY_ALL \
635 636 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
636 637
637 638 #define ZSD_ALL_INPROGRESS \
638 639 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
639 640
640 641 /*
641 642 * Macros to help with zone visibility restrictions.
642 643 */
643 644
644 645 /*
645 646 * Is process in the global zone?
646 647 */
647 648 #define INGLOBALZONE(p) \
648 649 ((p)->p_zone == global_zone)
649 650
650 651 /*
651 652 * Can process view objects in given zone?
652 653 */
653 654 #define HASZONEACCESS(p, zoneid) \
654 655 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
655 656
656 657 /*
657 658 * Convenience macro to see if a resolved path is visible from within a
658 659 * given zone.
659 660 *
660 661 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
661 662 * two strings must be equal. Since the rootpathlen has a trailing '/',
662 663 * we want to skip everything in the path up to (but not including) the
663 664 * trailing '/'.
664 665 */
665 666 #define ZONE_PATH_VISIBLE(path, zone) \
666 667 (strncmp((path), (zone)->zone_rootpath, \
667 668 (zone)->zone_rootpathlen - 1) == 0)
668 669
669 670 /*
670 671 * Convenience macro to go from the global view of a path to that seen
671 672 * from within said zone. It is the responsibility of the caller to
672 673 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
673 674 * in fact visible from within the zone.
674 675 */
675 676 #define ZONE_PATH_TRANSLATE(path, zone) \
676 677 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
677 678 (path) + (zone)->zone_rootpathlen - 2)
678 679
679 680 /*
680 681 * Special processes visible in all zones.
681 682 */
682 683 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
683 684
684 685 /*
685 686 * Zone-safe version of thread_create() to be used when the caller wants to
686 687 * create a kernel thread to run within the current zone's context.
687 688 */
688 689 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
689 690 pri_t);
690 691 extern void zthread_exit(void);
691 692
692 693 /*
693 694 * Functions for an external observer to register interest in a zone's status
694 695 * change. Observers will be woken up when the zone status equals the status
695 696 * argument passed in (in the case of zone_status_timedwait, the function may
696 697 * also return because of a timeout; zone_status_wait_sig may return early due
697 698 * to a signal being delivered; zone_status_timedwait_sig may return for any of
698 699 * the above reasons).
699 700 *
700 701 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
701 702 * cv_wait_sig() respectively.
702 703 */
703 704 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
704 705 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
705 706 extern void zone_status_wait(zone_t *, zone_status_t);
706 707 extern int zone_status_wait_sig(zone_t *, zone_status_t);
707 708
708 709 /*
709 710 * Get the status of the zone (at the time it was called). The state may
710 711 * have progressed by the time it is returned.
711 712 */
712 713 extern zone_status_t zone_status_get(zone_t *);
713 714
714 715 /*
715 716 * Safely get the hostid of the specified zone (defaults to machine's hostid
716 717 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
717 718 * the global zone's (i.e., physical system's) hostid.
718 719 */
719 720 extern uint32_t zone_get_hostid(zone_t *);
720 721
721 722 /*
722 723 * Get the "kcred" credentials corresponding to the given zone.
723 724 */
724 725 extern struct cred *zone_get_kcred(zoneid_t);
725 726
726 727 /*
727 728 * Get/set the pool the zone is currently bound to.
728 729 */
729 730 extern struct pool *zone_pool_get(zone_t *);
730 731 extern void zone_pool_set(zone_t *, struct pool *);
731 732
732 733 /*
733 734 * Get/set the pset the zone is currently using.
734 735 */
735 736 extern psetid_t zone_pset_get(zone_t *);
736 737 extern void zone_pset_set(zone_t *, psetid_t);
737 738
738 739 /*
739 740 * Get the number of cpus/online-cpus visible from the given zone.
740 741 */
741 742 extern int zone_ncpus_get(zone_t *);
742 743 extern int zone_ncpus_online_get(zone_t *);
743 744
744 745 /*
745 746 * Returns true if the named pool/dataset is visible in the current zone.
746 747 */
747 748 extern int zone_dataset_visible(const char *, int *);
748 749
749 750 /*
750 751 * zone version of kadmin()
751 752 */
752 753 extern int zone_kadmin(int, int, const char *, cred_t *);
753 754 extern void zone_shutdown_global(void);
754 755
755 756 extern void mount_in_progress(void);
756 757 extern void mount_completed(void);
757 758
758 759 extern int zone_walk(int (*)(zone_t *, void *), void *);
759 760
760 761 extern rctl_hndl_t rc_zone_locked_mem;
761 762 extern rctl_hndl_t rc_zone_max_swap;
762 763 extern rctl_hndl_t rc_zone_max_lofi;
763 764
764 765 #endif /* _KERNEL */
765 766
766 767 #ifdef __cplusplus
767 768 }
768 769 #endif
769 770
770 771 #endif /* _SYS_ZONE_H */
|
↓ open down ↓ |
526 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX