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