Print this page
| 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 2015 Joyent, Inc. All rights reserved.
23 24 * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
24 25 * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
25 26 * Copyright 2016, Joyent, Inc.
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_PMCAP_NOVER 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_DID 18
107 108 #define ZONE_ATTR_PMCAP_PAGEOUT 19
108 109 #define ZONE_ATTR_INITNORESTART 20
109 110 #define ZONE_ATTR_PG_FLT_DELAY 21
110 111 #define ZONE_ATTR_RSS 22
111 112 #define ZONE_ATTR_APP_SVC_CT 23
112 113 #define ZONE_ATTR_SCHED_FIXEDHI 24
113 114
114 115 /* Start of the brand-specific attribute namespace */
115 116 #define ZONE_ATTR_BRAND_ATTRS 32768
116 117
117 118 #define ZONE_FS_ALLOWED_MAX 1024
118 119
119 120 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
120 121 #define ZONE_EVENT_STATUS_CLASS "status"
121 122 #define ZONE_EVENT_STATUS_SUBCLASS "change"
122 123
123 124 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
124 125 #define ZONE_EVENT_INITIALIZED "initialized"
125 126 #define ZONE_EVENT_READY "ready"
126 127 #define ZONE_EVENT_RUNNING "running"
127 128 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
128 129
129 130 #define ZONE_CB_NAME "zonename"
130 131 #define ZONE_CB_NEWSTATE "newstate"
131 132 #define ZONE_CB_OLDSTATE "oldstate"
132 133 #define ZONE_CB_TIMESTAMP "when"
133 134 #define ZONE_CB_ZONEID "zoneid"
134 135
135 136 /*
136 137 * Exit values that may be returned by scripts or programs invoked by various
137 138 * zone commands.
138 139 *
139 140 * These are defined as:
140 141 *
141 142 * ZONE_SUBPROC_OK
142 143 * ===============
143 144 * The subprocess completed successfully.
144 145 *
145 146 * ZONE_SUBPROC_USAGE
146 147 * ==================
147 148 * The subprocess failed with a usage message, or a usage message should
148 149 * be output in its behalf.
149 150 *
150 151 * ZONE_SUBPROC_NOTCOMPLETE
151 152 * ========================
152 153 * The subprocess did not complete, but the actions performed by the
153 154 * subprocess require no recovery actions by the user.
154 155 *
155 156 * For example, if the subprocess were called by "zoneadm install," the
156 157 * installation of the zone did not succeed but the user need not perform
157 158 * a "zoneadm uninstall" before attempting another install.
158 159 *
159 160 * ZONE_SUBPROC_FATAL
160 161 * ==================
161 162 * The subprocess failed in a fatal manner, usually one that will require
162 163 * some type of recovery action by the user.
163 164 *
164 165 * For example, if the subprocess were called by "zoneadm install," the
165 166 * installation of the zone did not succeed and the user will need to
166 167 * perform a "zoneadm uninstall" before another install attempt is
167 168 * possible.
168 169 *
169 170 * The non-success exit values are large to avoid accidental collision
170 171 * with values used internally by some commands (e.g. "Z_ERR" and
171 172 * "Z_USAGE" as used by zoneadm.)
172 173 */
173 174 #define ZONE_SUBPROC_OK 0
174 175 #define ZONE_SUBPROC_USAGE 253
175 176 #define ZONE_SUBPROC_NOTCOMPLETE 254
176 177 #define ZONE_SUBPROC_FATAL 255
177 178
178 179 #ifdef _SYSCALL32
179 180 typedef struct {
180 181 caddr32_t zone_name;
181 182 caddr32_t zone_root;
182 183 caddr32_t zone_privs;
|
↓ open down ↓ |
150 lines elided |
↑ open up ↑ |
183 184 size32_t zone_privssz;
184 185 caddr32_t rctlbuf;
185 186 size32_t rctlbufsz;
186 187 caddr32_t extended_error;
187 188 caddr32_t zfsbuf;
188 189 size32_t zfsbufsz;
189 190 int match; /* match level */
190 191 uint32_t doi; /* DOI for label */
191 192 caddr32_t label; /* label associated with zone */
192 193 int flags;
193 - zoneid_t zoneid; /* requested zoneid */
194 194 } zone_def32;
195 195 #endif
196 196 typedef struct {
197 197 const char *zone_name;
198 198 const char *zone_root;
199 199 const struct priv_set *zone_privs;
200 200 size_t zone_privssz;
201 201 const char *rctlbuf;
202 202 size_t rctlbufsz;
203 203 int *extended_error;
204 204 const char *zfsbuf;
205 205 size_t zfsbufsz;
206 206 int match; /* match level */
207 207 uint32_t doi; /* DOI for label */
208 208 const bslabel_t *label; /* label associated with zone */
209 209 int flags;
210 - zoneid_t zoneid; /* requested zoneid */
211 210 } zone_def;
212 211
213 212 /* extended error information */
214 213 #define ZE_UNKNOWN 0 /* No extended error info */
215 214 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
216 215 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
217 216 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
218 217
219 218 /*
220 219 * zone_status values
221 220 *
222 221 * You must modify zone_status_names in mdb(1M)'s genunix module
223 222 * (genunix/zone.c) when you modify this enum.
224 223 */
225 224 typedef enum {
226 225 ZONE_IS_UNINITIALIZED = 0,
227 226 ZONE_IS_INITIALIZED,
228 227 ZONE_IS_READY,
229 228 ZONE_IS_BOOTING,
230 229 ZONE_IS_RUNNING,
231 230 ZONE_IS_SHUTTING_DOWN,
232 231 ZONE_IS_EMPTY,
233 232 ZONE_IS_DOWN,
234 233 ZONE_IS_DYING,
235 234 ZONE_IS_DEAD
236 235 } zone_status_t;
237 236 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
238 237 #define ZONE_MAX_STATE ZONE_IS_DEAD
239 238
240 239 /*
241 240 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
242 241 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
243 242 */
244 243 typedef enum zone_cmd {
245 244 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
246 245 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
247 246 } zone_cmd_t;
248 247
249 248 /*
250 249 * The structure of a request to zoneadmd.
251 250 */
252 251 typedef struct zone_cmd_arg {
253 252 uint64_t uniqid; /* unique "generation number" */
254 253 zone_cmd_t cmd; /* requested action */
255 254 int status; /* init status on shutdown */
256 255 uint32_t debug; /* enable brand hook debug */
257 256 char locale[MAXPATHLEN]; /* locale in which to render messages */
258 257 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
259 258 /* Needed for 32/64 zoneadm -> zoneadmd door arg size check. */
260 259 int pad;
261 260 } zone_cmd_arg_t;
262 261
263 262 /*
264 263 * Structure of zoneadmd's response to a request. A NULL return value means
265 264 * the caller should attempt to restart zoneadmd and retry.
266 265 */
267 266 typedef struct zone_cmd_rval {
268 267 int rval; /* return value of request */
269 268 char errbuf[1]; /* variable-sized buffer containing error messages */
270 269 } zone_cmd_rval_t;
271 270
272 271 /*
273 272 * The zone support infrastructure uses the zone name as a component
274 273 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
275 274 * in length, so ZONENAME_MAX is limited by that.
276 275 */
277 276 #define ZONENAME_MAX 64
278 277
279 278 #define GLOBAL_ZONENAME "global"
280 279
281 280 /*
282 281 * Extended Regular expression (see regex(5)) which matches all valid zone
283 282 * names.
284 283 */
285 284 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
286 285
287 286 /*
288 287 * Where the zones support infrastructure places temporary files.
289 288 */
290 289 #define ZONES_TMPDIR "/var/run/zones"
291 290
292 291 /*
293 292 * The path to the door used by clients to communicate with zoneadmd.
294 293 */
295 294 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
296 295
297 296
298 297 /* zone_flags */
299 298 /*
300 299 * Threads that read or write the following flag must hold zone_lock.
301 300 */
302 301 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
303 302
304 303 /*
305 304 * The following threads are set when the zone is created and never changed.
306 305 * Threads that test for these flags don't have to hold zone_lock.
307 306 */
308 307 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
309 308 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
310 309 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
311 310
312 311
313 312 /* zone_create flags */
314 313 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
315 314
316 315 /* zone network properties */
317 316 #define ZONE_NETWORK_ADDRESS 1
318 317 #define ZONE_NETWORK_DEFROUTER 2
319 318
320 319 #define ZONE_NET_ADDRNAME "address"
321 320 #define ZONE_NET_RTRNAME "route"
322 321
323 322 typedef struct zone_net_data {
324 323 int zn_type;
325 324 int zn_len;
326 325 datalink_id_t zn_linkid;
327 326 uint8_t zn_val[1];
328 327 } zone_net_data_t;
329 328
330 329
331 330 #ifdef _KERNEL
332 331
333 332 /*
334 333 * We need to protect the definition of 'list_t' from userland applications and
335 334 * libraries which may be defining ther own versions.
336 335 */
337 336 #include <sys/list.h>
338 337 #include <sys/loadavg.h>
339 338
340 339 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
341 340
342 341 struct pool;
343 342 struct brand;
344 343
345 344 /*
346 345 * Each of these constants identifies a kernel subsystem that acquires and
347 346 * releases zone references. Each subsystem that invokes
348 347 * zone_hold_ref() and zone_rele_ref() should specify the
349 348 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
350 349 * help users and developers quickly identify subsystems that stall zone
351 350 * shutdowns indefinitely.
352 351 *
353 352 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
354 353 * when you modify this enumeration.
355 354 */
356 355 typedef enum zone_ref_subsys {
357 356 ZONE_REF_NFS, /* NFS */
358 357 ZONE_REF_NFSV4, /* NFSv4 */
359 358 ZONE_REF_SMBFS, /* SMBFS */
360 359 ZONE_REF_MNTFS, /* MNTFS */
361 360 ZONE_REF_LOFI, /* LOFI devices */
362 361 ZONE_REF_VFS, /* VFS infrastructure */
363 362 ZONE_REF_IPC, /* IPC infrastructure */
364 363 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
365 364 } zone_ref_subsys_t;
366 365
367 366 /*
368 367 * zone_ref represents a general-purpose references to a zone. Each zone's
369 368 * references are linked into the zone's zone_t::zone_ref_list. This allows
370 369 * debuggers to walk zones' references.
371 370 */
372 371 typedef struct zone_ref {
373 372 struct zone *zref_zone; /* the zone to which the reference refers */
374 373 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
375 374 } zone_ref_t;
|
↓ open down ↓ |
155 lines elided |
↑ open up ↑ |
376 375
377 376 /*
378 377 * Structure to record list of ZFS datasets exported to a zone.
379 378 */
380 379 typedef struct zone_dataset {
381 380 char *zd_dataset;
382 381 list_node_t zd_linkage;
383 382 } zone_dataset_t;
384 383
385 384 /*
386 - * structure for rctl zone kstats
385 + * structure for zone kstats
387 386 */
388 387 typedef struct zone_kstat {
389 388 kstat_named_t zk_zonename;
390 389 kstat_named_t zk_usage;
391 390 kstat_named_t zk_value;
392 391 } zone_kstat_t;
393 392
394 393 struct cpucap;
395 394
396 395 typedef struct {
397 396 hrtime_t cycle_start;
398 397 uint_t cycle_cnt;
399 398 hrtime_t zone_avg_cnt;
400 399 } sys_zio_cntr_t;
401 400
402 401 typedef struct {
403 402 kstat_named_t zv_zonename;
404 403 kstat_named_t zv_nread;
405 404 kstat_named_t zv_reads;
406 405 kstat_named_t zv_rtime;
407 406 kstat_named_t zv_rlentime;
408 407 kstat_named_t zv_rcnt;
409 408 kstat_named_t zv_nwritten;
410 409 kstat_named_t zv_writes;
411 410 kstat_named_t zv_wtime;
412 411 kstat_named_t zv_wlentime;
413 412 kstat_named_t zv_wcnt;
414 413 kstat_named_t zv_10ms_ops;
415 414 kstat_named_t zv_100ms_ops;
416 415 kstat_named_t zv_1s_ops;
417 416 kstat_named_t zv_10s_ops;
418 417 kstat_named_t zv_delay_cnt;
419 418 kstat_named_t zv_delay_time;
420 419 } zone_vfs_kstat_t;
421 420
422 421 typedef struct {
423 422 kstat_named_t zz_zonename;
424 423 kstat_named_t zz_nread;
425 424 kstat_named_t zz_reads;
426 425 kstat_named_t zz_rtime;
427 426 kstat_named_t zz_rlentime;
428 427 kstat_named_t zz_nwritten;
429 428 kstat_named_t zz_writes;
430 429 kstat_named_t zz_waittime;
431 430 } zone_zfs_kstat_t;
432 431
433 432 typedef struct {
434 433 kstat_named_t zm_zonename;
435 434 kstat_named_t zm_rss;
436 435 kstat_named_t zm_phys_cap;
437 436 kstat_named_t zm_swap;
438 437 kstat_named_t zm_swap_cap;
439 438 kstat_named_t zm_nover;
440 439 kstat_named_t zm_pagedout;
441 440 kstat_named_t zm_pgpgin;
442 441 kstat_named_t zm_anonpgin;
443 442 kstat_named_t zm_execpgin;
444 443 kstat_named_t zm_fspgin;
445 444 kstat_named_t zm_anon_alloc_fail;
446 445 kstat_named_t zm_pf_throttle;
447 446 kstat_named_t zm_pf_throttle_usec;
448 447 } zone_mcap_kstat_t;
449 448
450 449 typedef struct {
451 450 kstat_named_t zm_zonename; /* full name, kstat truncates name */
|
↓ open down ↓ |
55 lines elided |
↑ open up ↑ |
452 451 kstat_named_t zm_utime;
453 452 kstat_named_t zm_stime;
454 453 kstat_named_t zm_wtime;
455 454 kstat_named_t zm_avenrun1;
456 455 kstat_named_t zm_avenrun5;
457 456 kstat_named_t zm_avenrun15;
458 457 kstat_named_t zm_ffcap;
459 458 kstat_named_t zm_ffnoproc;
460 459 kstat_named_t zm_ffnomem;
461 460 kstat_named_t zm_ffmisc;
462 - kstat_named_t zm_mfseglim;
463 461 kstat_named_t zm_nested_intp;
464 462 kstat_named_t zm_init_pid;
465 463 kstat_named_t zm_boot_time;
466 464 } zone_misc_kstat_t;
467 465
468 466 typedef struct zone {
469 467 /*
470 468 * zone_name is never modified once set.
471 469 */
472 470 char *zone_name; /* zone's configuration name */
473 471 /*
474 472 * zone_nodename and zone_domain are never freed once allocated.
475 473 */
476 474 char *zone_nodename; /* utsname.nodename equivalent */
477 475 char *zone_domain; /* srpc_domain equivalent */
478 476 /*
479 477 * zone_hostid is used for per-zone hostid emulation.
480 478 * Currently it isn't modified after it's set (so no locks protect
481 479 * accesses), but that might have to change when we allow
482 480 * administrators to change running zones' properties.
483 481 *
484 482 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
485 483 * that zone_get_hostid() will function correctly.
486 484 */
487 485 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
488 486 /* if not emulated */
489 487 /*
490 488 * zone_lock protects the following fields of a zone_t:
491 489 * zone_ref
492 490 * zone_cred_ref
493 491 * zone_subsys_ref
494 492 * zone_ref_list
495 493 * zone_ntasks
496 494 * zone_flags
|
↓ open down ↓ |
24 lines elided |
↑ open up ↑ |
497 495 * zone_zsd
498 496 * zone_pfexecd
499 497 */
500 498 kmutex_t zone_lock;
501 499 /*
502 500 * zone_linkage is the zone's linkage into the active or
503 501 * death-row list. The field is protected by zonehash_lock.
504 502 */
505 503 list_node_t zone_linkage;
506 504 zoneid_t zone_id; /* ID of zone */
507 - zoneid_t zone_did; /* persistent debug ID of zone */
508 505 uint_t zone_ref; /* count of zone_hold()s on zone */
509 506 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
510 507 /*
511 508 * Fixed-sized array of subsystem-specific reference counts
512 509 * The sum of all of the counts must be less than or equal to zone_ref.
513 510 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
514 511 * constants.
515 512 */
516 513 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
517 514 list_t zone_ref_list; /* list of zone_ref_t structs */
518 515 /*
519 516 * zone_rootvp and zone_rootpath can never be modified once set.
520 517 */
521 518 struct vnode *zone_rootvp; /* zone's root vnode */
522 519 char *zone_rootpath; /* Path to zone's root + '/' */
523 520 ushort_t zone_flags; /* misc flags */
524 521 zone_status_t zone_status; /* protected by zone_status_lock */
525 522 uint_t zone_ntasks; /* number of tasks executing in zone */
526 523 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
527 524 /* counters in projects and tasks */
528 525 /* that are within the zone */
529 526 rctl_qty_t zone_nlwps; /* number of lwps in zone */
530 527 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
531 528 rctl_qty_t zone_shmmax; /* System V shared memory usage */
532 529 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
533 530
534 531 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
535 532 uint32_t zone_shares; /* FSS shares allocated to zone */
536 533 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
537 534 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
538 535 /* kpd_locked_mem for all */
539 536 /* projects in zone. */
540 537 /* Also protects zone_max_swap */
541 538 /* grab after p_lock, before rcs_lock */
542 539 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
543 540 /* zone */
544 541 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
545 542 /* limit. Protected by */
546 543 /* zone_rctls->rcs_lock */
547 544 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
548 545 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
549 546 /* Protected by */
550 547 /* zone_rctls->rcs_lock */
551 548 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
552 549 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
553 550 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
554 551 /* Protected by */
555 552 /* zone_rctls->rcs_lock */
556 553 list_t zone_zsd; /* list of Zone-Specific Data values */
557 554 kcondvar_t zone_cv; /* used to signal state changes */
558 555 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
559 556 pid_t zone_proc_initpid; /* pid of "init" for this zone */
560 557 char *zone_initname; /* fs path to 'init' */
561 558 int zone_init_status; /* init's exit status */
562 559 int zone_boot_err; /* for zone_boot() if boot fails */
563 560 char *zone_bootargs; /* arguments passed via zone_boot() */
564 561 rctl_qty_t zone_phys_mem_ctl; /* current phys. memory limit */
565 562 /*
566 563 * zone_kthreads is protected by zone_status_lock.
567 564 */
568 565 kthread_t *zone_kthreads; /* kernel threads in zone */
569 566 struct priv_set *zone_privset; /* limit set for zone */
570 567 /*
571 568 * zone_vfslist is protected by vfs_list_lock().
572 569 */
573 570 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
574 571 uint64_t zone_uniqid; /* unique zone generation number */
575 572 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
576 573 /*
577 574 * zone_pool is protected by pool_lock().
578 575 */
579 576 struct pool *zone_pool; /* pool the zone is bound to */
580 577 hrtime_t zone_pool_mod; /* last pool bind modification time */
581 578 /* zone_psetid is protected by cpu_lock */
582 579 psetid_t zone_psetid; /* pset the zone is bound to */
583 580
584 581 time_t zone_boot_time; /* Similar to boot_time */
585 582
586 583 /*
587 584 * The following two can be read without holding any locks. They are
588 585 * updated under cpu_lock.
589 586 */
590 587 int zone_ncpus; /* zone's idea of ncpus */
591 588 int zone_ncpus_online; /* zone's idea of ncpus_online */
592 589 /*
593 590 * List of ZFS datasets exported to this zone.
594 591 */
595 592 list_t zone_datasets; /* list of datasets */
596 593
597 594 ts_label_t *zone_slabel; /* zone sensitivity label */
598 595 int zone_match; /* require label match for packets */
599 596 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
600 597
601 598 boolean_t zone_restart_init; /* Restart init if it dies? */
602 599 boolean_t zone_reboot_on_init_exit; /* Reboot if init dies? */
603 600 boolean_t zone_setup_app_contract; /* setup contract? */
604 601 struct brand *zone_brand; /* zone's brand */
605 602 void *zone_brand_data; /* store brand specific data */
606 603 id_t zone_defaultcid; /* dflt scheduling class id */
607 604 boolean_t zone_fixed_hipri; /* fixed sched. hi prio */
608 605 kstat_t *zone_swapresv_kstat;
609 606 kstat_t *zone_lockedmem_kstat;
610 607 /*
611 608 * zone_dl_list is protected by zone_lock
612 609 */
613 610 list_t zone_dl_list;
614 611 netstack_t *zone_netstack;
615 612 struct cpucap *zone_cpucap; /* CPU caps data */
616 613
617 614 /*
618 615 * Data and counters used for ZFS fair-share disk IO.
619 616 */
620 617 rctl_qty_t zone_zfs_io_pri; /* ZFS IO priority */
621 618 uint_t zone_zfs_queued[2]; /* sync I/O enqueued count */
|
↓ open down ↓ |
104 lines elided |
↑ open up ↑ |
622 619 uint64_t zone_zfs_weight; /* used to prevent starvation */
623 620 uint64_t zone_io_util; /* IO utilization metric */
624 621 boolean_t zone_io_util_above_avg; /* IO util percent > avg. */
625 622 uint16_t zone_io_delay; /* IO delay on logical r/w */
626 623 kmutex_t zone_stg_io_lock; /* protects IO window data */
627 624 sys_zio_cntr_t zone_rd_ops; /* Counters for ZFS reads, */
628 625 sys_zio_cntr_t zone_wr_ops; /* writes and */
629 626 sys_zio_cntr_t zone_lwr_ops; /* logical writes. */
630 627
631 628 /*
629 + * kstats and counters for I/O ops and bytes.
630 + */
631 + kmutex_t zone_io_lock; /* protects I/O statistics */
632 + kstat_t *zone_io_ksp;
633 + kstat_io_t *zone_io_kiop;
634 +
635 + /*
632 636 * kstats and counters for VFS ops and bytes.
633 637 */
634 638 kmutex_t zone_vfs_lock; /* protects VFS statistics */
635 639 kstat_t *zone_vfs_ksp;
636 640 kstat_io_t zone_vfs_rwstats;
637 641 zone_vfs_kstat_t *zone_vfs_stats;
638 642
639 643 /*
640 644 * kstats for ZFS I/O ops and bytes.
641 645 */
642 646 kmutex_t zone_zfs_lock; /* protects ZFS statistics */
643 - kstat_t *zone_zfs_ksp;
644 647 kstat_io_t zone_zfs_rwstats;
645 648 zone_zfs_kstat_t *zone_zfs_stats;
646 649
647 650 /*
648 651 * Solaris Auditing per-zone audit context
649 652 */
650 653 struct au_kcontext *zone_audit_kctxt;
651 654 /*
652 655 * For private use by mntfs.
653 656 */
654 657 struct mntelem *zone_mntfs_db;
655 658 krwlock_t zone_mntfs_db_lock;
656 659
657 660 struct klpd_reg *zone_pfexecd;
658 661
659 662 char *zone_fs_allowed;
660 663 rctl_qty_t zone_nprocs; /* number of processes in the zone */
661 664 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
662 665 /* zone_rctls->rcs_lock */
663 666 kstat_t *zone_nprocs_kstat;
664 667
665 668 /*
666 669 * kstats and counters for physical memory capping.
667 670 */
668 671 rctl_qty_t zone_phys_mem; /* current bytes of phys. mem. (RSS) */
669 672 kstat_t *zone_physmem_kstat;
670 673 uint64_t zone_mcap_nover; /* # of times over phys. cap */
671 674 uint64_t zone_mcap_pagedout; /* bytes of mem. paged out */
672 675 kmutex_t zone_mcap_lock; /* protects mcap statistics */
673 676 kstat_t *zone_mcap_ksp;
674 677 zone_mcap_kstat_t *zone_mcap_stats;
675 678 uint64_t zone_pgpgin; /* pages paged in */
676 679 uint64_t zone_anonpgin; /* anon pages paged in */
677 680 uint64_t zone_execpgin; /* exec pages paged in */
678 681 uint64_t zone_fspgin; /* fs pages paged in */
679 682 uint64_t zone_anon_alloc_fail; /* cnt of anon alloc fails */
680 683 uint64_t zone_pf_throttle; /* cnt of page flt throttles */
681 684 uint64_t zone_pf_throttle_usec; /* time of page flt throttles */
682 685
683 686 /* Num usecs to throttle page fault when zone is over phys. mem cap */
684 687 uint32_t zone_pg_flt_delay;
685 688
686 689 /*
687 690 * Misc. kstats and counters for zone cpu-usage aggregation.
688 691 * The zone_Xtime values are the sum of the micro-state accounting
689 692 * values for all threads that are running or have run in the zone.
690 693 * This is tracked in msacct.c as threads change state.
691 694 * The zone_stime is the sum of the LMS_SYSTEM times.
692 695 * The zone_utime is the sum of the LMS_USER times.
693 696 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
694 697 * As with per-thread micro-state accounting values, these values are
695 698 * not scaled to nanosecs. The scaling is done by the
696 699 * zone_misc_kstat_update function when kstats are requested.
697 700 */
698 701 kmutex_t zone_misc_lock; /* protects misc statistics */
699 702 kstat_t *zone_misc_ksp;
|
↓ open down ↓ |
46 lines elided |
↑ open up ↑ |
700 703 zone_misc_kstat_t *zone_misc_stats;
701 704 uint64_t zone_stime; /* total system time */
702 705 uint64_t zone_utime; /* total user time */
703 706 uint64_t zone_wtime; /* total time waiting in runq */
704 707 /* fork-fail kstat tracking */
705 708 uint32_t zone_ffcap; /* hit an rctl cap */
706 709 uint32_t zone_ffnoproc; /* get proc/lwp error */
707 710 uint32_t zone_ffnomem; /* as_dup/memory error */
708 711 uint32_t zone_ffmisc; /* misc. other error */
709 712
710 - uint32_t zone_mfseglim; /* map failure (# segs limit) */
711 -
712 713 uint32_t zone_nested_intp; /* nested interp. kstat */
713 714
714 715 struct loadavg_s zone_loadavg; /* loadavg for this zone */
715 716 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
716 717 int zone_avenrun[3]; /* FSCALED avg. run queue len */
717 718
718 719 /*
719 720 * FSS stats updated once per second by fss_decay_usage.
720 721 */
721 722 uint32_t zone_fss_gen; /* FSS generation cntr */
722 723 uint64_t zone_run_ticks; /* tot # of ticks running */
723 724
724 725 /*
725 726 * DTrace-private per-zone state
726 727 */
727 728 int zone_dtrace_getf; /* # of unprivileged getf()s */
728 729
729 730 /*
730 731 * Synchronization primitives used to synchronize between mounts and
731 732 * zone creation/destruction.
732 733 */
733 734 int zone_mounts_in_progress;
734 735 kcondvar_t zone_mount_cv;
735 736 kmutex_t zone_mount_lock;
736 737 } zone_t;
737 738
738 739 /*
739 740 * Special value of zone_psetid to indicate that pools are disabled.
740 741 */
741 742 #define ZONE_PS_INVAL PS_MYID
742 743
743 744
744 745 extern zone_t zone0;
745 746 extern zone_t *global_zone;
746 747 extern uint_t maxzones;
747 748 extern rctl_hndl_t rc_zone_nlwps;
748 749 extern rctl_hndl_t rc_zone_nprocs;
749 750
750 751 extern long zone(int, void *, void *, void *, void *);
751 752 extern void zone_zsd_init(void);
752 753 extern void zone_init(void);
753 754 extern void zone_hold(zone_t *);
754 755 extern void zone_rele(zone_t *);
755 756 extern void zone_init_ref(zone_ref_t *);
756 757 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
757 758 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
|
↓ open down ↓ |
36 lines elided |
↑ open up ↑ |
758 759 extern void zone_cred_hold(zone_t *);
759 760 extern void zone_cred_rele(zone_t *);
760 761 extern void zone_task_hold(zone_t *);
761 762 extern void zone_task_rele(zone_t *);
762 763 extern zone_t *zone_find_by_id(zoneid_t);
763 764 extern zone_t *zone_find_by_label(const ts_label_t *);
764 765 extern zone_t *zone_find_by_name(char *);
765 766 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
766 767 extern zone_t *zone_find_by_path(const char *);
767 768 extern zoneid_t getzoneid(void);
768 -extern zoneid_t getzonedid(void);
769 769 extern zone_t *zone_find_by_id_nolock(zoneid_t);
770 770 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
771 771 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
772 772 extern void zone_loadavg_update();
773 773
774 774 /*
775 775 * Zone-specific data (ZSD) APIs
776 776 */
777 777 /*
778 778 * The following is what code should be initializing its zone_key_t to if it
779 779 * calls zone_getspecific() without necessarily knowing that zone_key_create()
780 780 * has been called on the key.
781 781 */
782 782 #define ZONE_KEY_UNINITIALIZED 0
783 783
784 784 typedef uint_t zone_key_t;
785 785
786 786 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
787 787 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
788 788 extern int zone_key_delete(zone_key_t);
789 789 extern void *zone_getspecific(zone_key_t, zone_t *);
790 790 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
791 791
792 792 /*
793 793 * The definition of a zsd_entry is truly private to zone.c and is only
794 794 * placed here so it can be shared with mdb.
795 795 *
796 796 * State maintained for each zone times each registered key, which tracks
797 797 * the state of the create, shutdown and destroy callbacks.
798 798 *
799 799 * zsd_flags is used to keep track of pending actions to avoid holding locks
800 800 * when calling the create/shutdown/destroy callbacks, since doing so
801 801 * could lead to deadlocks.
802 802 */
803 803 struct zsd_entry {
804 804 zone_key_t zsd_key; /* Key used to lookup value */
805 805 void *zsd_data; /* Caller-managed value */
806 806 /*
807 807 * Callbacks to be executed when a zone is created, shutdown, and
808 808 * destroyed, respectively.
809 809 */
810 810 void *(*zsd_create)(zoneid_t);
811 811 void (*zsd_shutdown)(zoneid_t, void *);
812 812 void (*zsd_destroy)(zoneid_t, void *);
813 813 list_node_t zsd_linkage;
814 814 uint16_t zsd_flags; /* See below */
815 815 kcondvar_t zsd_cv;
816 816 };
817 817
818 818 /*
819 819 * zsd_flags
820 820 */
821 821 #define ZSD_CREATE_NEEDED 0x0001
822 822 #define ZSD_CREATE_INPROGRESS 0x0002
823 823 #define ZSD_CREATE_COMPLETED 0x0004
824 824 #define ZSD_SHUTDOWN_NEEDED 0x0010
825 825 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
826 826 #define ZSD_SHUTDOWN_COMPLETED 0x0040
827 827 #define ZSD_DESTROY_NEEDED 0x0100
828 828 #define ZSD_DESTROY_INPROGRESS 0x0200
829 829 #define ZSD_DESTROY_COMPLETED 0x0400
830 830
831 831 #define ZSD_CREATE_ALL \
832 832 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
833 833 #define ZSD_SHUTDOWN_ALL \
834 834 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
835 835 #define ZSD_DESTROY_ALL \
836 836 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
837 837
838 838 #define ZSD_ALL_INPROGRESS \
839 839 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
840 840
841 841 /*
842 842 * Macros to help with zone visibility restrictions.
843 843 */
844 844
845 845 /*
846 846 * Is process in the global zone?
847 847 */
848 848 #define INGLOBALZONE(p) \
849 849 ((p)->p_zone == global_zone)
850 850
851 851 /*
852 852 * Can process view objects in given zone?
853 853 */
854 854 #define HASZONEACCESS(p, zoneid) \
855 855 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
856 856
857 857 /*
858 858 * Convenience macro to see if a resolved path is visible from within a
859 859 * given zone.
860 860 *
861 861 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
862 862 * two strings must be equal. Since the rootpathlen has a trailing '/',
863 863 * we want to skip everything in the path up to (but not including) the
864 864 * trailing '/'.
865 865 */
866 866 #define ZONE_PATH_VISIBLE(path, zone) \
867 867 (strncmp((path), (zone)->zone_rootpath, \
868 868 (zone)->zone_rootpathlen - 1) == 0)
869 869
870 870 /*
871 871 * Convenience macro to go from the global view of a path to that seen
872 872 * from within said zone. It is the responsibility of the caller to
873 873 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
874 874 * in fact visible from within the zone.
875 875 */
876 876 #define ZONE_PATH_TRANSLATE(path, zone) \
877 877 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
878 878 (path) + (zone)->zone_rootpathlen - 2)
879 879
880 880 /*
881 881 * Special processes visible in all zones.
882 882 */
883 883 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
884 884
885 885 /*
886 886 * Zone-safe version of thread_create() to be used when the caller wants to
887 887 * create a kernel thread to run within the current zone's context.
888 888 */
889 889 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
890 890 pri_t);
891 891 extern void zthread_exit(void);
892 892
893 893 /*
894 894 * Functions for an external observer to register interest in a zone's status
895 895 * change. Observers will be woken up when the zone status equals the status
896 896 * argument passed in (in the case of zone_status_timedwait, the function may
897 897 * also return because of a timeout; zone_status_wait_sig may return early due
898 898 * to a signal being delivered; zone_status_timedwait_sig may return for any of
899 899 * the above reasons).
900 900 *
901 901 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
902 902 * cv_wait_sig() respectively.
903 903 */
904 904 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
905 905 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
906 906 extern void zone_status_wait(zone_t *, zone_status_t);
907 907 extern int zone_status_wait_sig(zone_t *, zone_status_t);
908 908
909 909 /*
910 910 * Get the status of the zone (at the time it was called). The state may
911 911 * have progressed by the time it is returned.
912 912 */
913 913 extern zone_status_t zone_status_get(zone_t *);
914 914
915 915 /*
916 916 * Safely get the hostid of the specified zone (defaults to machine's hostid
917 917 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
918 918 * the global zone's (i.e., physical system's) hostid.
919 919 */
920 920 extern uint32_t zone_get_hostid(zone_t *);
921 921
922 922 /*
923 923 * Get the "kcred" credentials corresponding to the given zone.
924 924 */
925 925 extern struct cred *zone_get_kcred(zoneid_t);
926 926
927 927 /*
928 928 * Get/set the pool the zone is currently bound to.
929 929 */
930 930 extern struct pool *zone_pool_get(zone_t *);
931 931 extern void zone_pool_set(zone_t *, struct pool *);
932 932
933 933 /*
934 934 * Get/set the pset the zone is currently using.
935 935 */
936 936 extern psetid_t zone_pset_get(zone_t *);
937 937 extern void zone_pset_set(zone_t *, psetid_t);
938 938
939 939 /*
940 940 * Get the number of cpus/online-cpus visible from the given zone.
941 941 */
942 942 extern int zone_ncpus_get(zone_t *);
943 943 extern int zone_ncpus_online_get(zone_t *);
944 944
945 945 /*
946 946 * Returns true if the named pool/dataset is visible in the current zone.
947 947 */
948 948 extern int zone_dataset_visible(const char *, int *);
949 949 extern int zone_dataset_visible_inzone(zone_t *, const char *, int *);
950 950
951 951 /*
952 952 * zone version of kadmin()
953 953 */
954 954 extern int zone_kadmin(int, int, const char *, cred_t *);
955 955 extern void zone_shutdown_global(void);
956 956
957 957 extern void mount_in_progress(zone_t *);
958 958 extern void mount_completed(zone_t *);
959 959
960 960 extern int zone_walk(int (*)(zone_t *, void *), void *);
961 961
962 962 extern rctl_hndl_t rc_zone_locked_mem;
963 963 extern rctl_hndl_t rc_zone_max_swap;
964 964 extern rctl_hndl_t rc_zone_phys_mem;
965 965 extern rctl_hndl_t rc_zone_max_lofi;
966 966
967 967 #endif /* _KERNEL */
968 968
969 969 #ifdef __cplusplus
970 970 }
971 971 #endif
972 972
973 973 #endif /* _SYS_ZONE_H */
|
↓ open down ↓ |
195 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX