Print this page
9936 atomic ops in syscall_mstate() induce significant overhead
9942 zone secflags are not initialized correctly
| 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 + * Copyright 2018 Joyent, Inc.
24 24 * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
25 25 * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
26 26 */
27 27
28 28 #ifndef _SYS_ZONE_H
29 29 #define _SYS_ZONE_H
30 30
31 31 #include <sys/types.h>
32 32 #include <sys/mutex.h>
33 33 #include <sys/param.h>
34 34 #include <sys/rctl.h>
|
↓ open down ↓ |
1 lines elided |
↑ open up ↑ |
35 35 #include <sys/ipc_rctl.h>
36 36 #include <sys/pset.h>
37 37 #include <sys/tsol/label.h>
38 38 #include <sys/cred.h>
39 39 #include <sys/netstack.h>
40 40 #include <sys/uadmin.h>
41 41 #include <sys/ksynch.h>
42 42 #include <sys/socket_impl.h>
43 43 #include <sys/secflags.h>
44 44 #include <netinet/in.h>
45 +#include <sys/cpu_uarray.h>
45 46
46 47 #ifdef __cplusplus
47 48 extern "C" {
48 49 #endif
49 50
50 51 /*
51 52 * NOTE
52 53 *
53 54 * The contents of this file are private to the implementation of
54 55 * Solaris and are subject to change at any time without notice.
55 56 * Applications and drivers using these interfaces may fail to
56 57 * run on future releases.
57 58 */
58 59
59 60 /* Available both in kernel and for user space */
60 61
61 62 /* zone id restrictions and special ids */
62 63 #define MAX_ZONEID 9999
63 64 #define MIN_USERZONEID 1 /* lowest user-creatable zone ID */
64 65 #define MIN_ZONEID 0 /* minimum zone ID on system */
65 66 #define GLOBAL_ZONEID 0
66 67 #define ZONEID_WIDTH 4 /* for printf */
67 68
68 69 /*
69 70 * Special zoneid_t token to refer to all zones.
70 71 */
71 72 #define ALL_ZONES (-1)
72 73
73 74 /* system call subcodes */
74 75 #define ZONE_CREATE 0
75 76 #define ZONE_DESTROY 1
76 77 #define ZONE_GETATTR 2
77 78 #define ZONE_ENTER 3
78 79 #define ZONE_LIST 4
79 80 #define ZONE_SHUTDOWN 5
80 81 #define ZONE_LOOKUP 6
81 82 #define ZONE_BOOT 7
82 83 #define ZONE_VERSION 8
83 84 #define ZONE_SETATTR 9
84 85 #define ZONE_ADD_DATALINK 10
85 86 #define ZONE_DEL_DATALINK 11
86 87 #define ZONE_CHECK_DATALINK 12
87 88 #define ZONE_LIST_DATALINK 13
88 89
89 90 /* zone attributes */
90 91 #define ZONE_ATTR_ROOT 1
91 92 #define ZONE_ATTR_NAME 2
92 93 #define ZONE_ATTR_STATUS 3
93 94 #define ZONE_ATTR_PRIVSET 4
94 95 #define ZONE_ATTR_UNIQID 5
95 96 #define ZONE_ATTR_POOLID 6
96 97 #define ZONE_ATTR_INITPID 7
97 98 #define ZONE_ATTR_SLBL 8
98 99 #define ZONE_ATTR_INITNAME 9
99 100 #define ZONE_ATTR_BOOTARGS 10
100 101 #define ZONE_ATTR_BRAND 11
101 102 #define ZONE_ATTR_PHYS_MCAP 12
102 103 #define ZONE_ATTR_SCHED_CLASS 13
103 104 #define ZONE_ATTR_FLAGS 14
104 105 #define ZONE_ATTR_HOSTID 15
105 106 #define ZONE_ATTR_FS_ALLOWED 16
106 107 #define ZONE_ATTR_NETWORK 17
107 108 #define ZONE_ATTR_INITNORESTART 20
108 109 #define ZONE_ATTR_SECFLAGS 21
109 110
110 111 /* Start of the brand-specific attribute namespace */
111 112 #define ZONE_ATTR_BRAND_ATTRS 32768
112 113
113 114 #define ZONE_FS_ALLOWED_MAX 1024
114 115
115 116 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
116 117 #define ZONE_EVENT_STATUS_CLASS "status"
117 118 #define ZONE_EVENT_STATUS_SUBCLASS "change"
118 119
119 120 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
120 121 #define ZONE_EVENT_INITIALIZED "initialized"
121 122 #define ZONE_EVENT_READY "ready"
122 123 #define ZONE_EVENT_RUNNING "running"
123 124 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
124 125
125 126 #define ZONE_CB_NAME "zonename"
126 127 #define ZONE_CB_NEWSTATE "newstate"
127 128 #define ZONE_CB_OLDSTATE "oldstate"
128 129 #define ZONE_CB_TIMESTAMP "when"
129 130 #define ZONE_CB_ZONEID "zoneid"
130 131
131 132 /*
132 133 * Exit values that may be returned by scripts or programs invoked by various
133 134 * zone commands.
134 135 *
135 136 * These are defined as:
136 137 *
137 138 * ZONE_SUBPROC_OK
138 139 * ===============
139 140 * The subprocess completed successfully.
140 141 *
141 142 * ZONE_SUBPROC_USAGE
142 143 * ==================
143 144 * The subprocess failed with a usage message, or a usage message should
144 145 * be output in its behalf.
145 146 *
146 147 * ZONE_SUBPROC_NOTCOMPLETE
147 148 * ========================
148 149 * The subprocess did not complete, but the actions performed by the
149 150 * subprocess require no recovery actions by the user.
150 151 *
151 152 * For example, if the subprocess were called by "zoneadm install," the
152 153 * installation of the zone did not succeed but the user need not perform
153 154 * a "zoneadm uninstall" before attempting another install.
154 155 *
155 156 * ZONE_SUBPROC_FATAL
156 157 * ==================
157 158 * The subprocess failed in a fatal manner, usually one that will require
158 159 * some type of recovery action by the user.
159 160 *
160 161 * For example, if the subprocess were called by "zoneadm install," the
161 162 * installation of the zone did not succeed and the user will need to
162 163 * perform a "zoneadm uninstall" before another install attempt is
163 164 * possible.
164 165 *
165 166 * The non-success exit values are large to avoid accidental collision
166 167 * with values used internally by some commands (e.g. "Z_ERR" and
167 168 * "Z_USAGE" as used by zoneadm.)
168 169 */
169 170 #define ZONE_SUBPROC_OK 0
170 171 #define ZONE_SUBPROC_USAGE 253
171 172 #define ZONE_SUBPROC_NOTCOMPLETE 254
172 173 #define ZONE_SUBPROC_FATAL 255
173 174
174 175 #ifdef _SYSCALL32
175 176 typedef struct {
176 177 caddr32_t zone_name;
177 178 caddr32_t zone_root;
178 179 caddr32_t zone_privs;
179 180 size32_t zone_privssz;
180 181 caddr32_t rctlbuf;
181 182 size32_t rctlbufsz;
182 183 caddr32_t extended_error;
183 184 caddr32_t zfsbuf;
184 185 size32_t zfsbufsz;
185 186 int match; /* match level */
186 187 uint32_t doi; /* DOI for label */
187 188 caddr32_t label; /* label associated with zone */
188 189 int flags;
189 190 } zone_def32;
190 191 #endif
191 192 typedef struct {
192 193 const char *zone_name;
193 194 const char *zone_root;
194 195 const struct priv_set *zone_privs;
195 196 size_t zone_privssz;
196 197 const char *rctlbuf;
197 198 size_t rctlbufsz;
198 199 int *extended_error;
199 200 const char *zfsbuf;
200 201 size_t zfsbufsz;
201 202 int match; /* match level */
202 203 uint32_t doi; /* DOI for label */
203 204 const bslabel_t *label; /* label associated with zone */
204 205 int flags;
205 206 } zone_def;
206 207
207 208 /* extended error information */
208 209 #define ZE_UNKNOWN 0 /* No extended error info */
209 210 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
210 211 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
211 212 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
212 213
213 214 /*
214 215 * zone_status values
215 216 *
216 217 * You must modify zone_status_names in mdb(1M)'s genunix module
217 218 * (genunix/zone.c) when you modify this enum.
218 219 */
219 220 typedef enum {
220 221 ZONE_IS_UNINITIALIZED = 0,
221 222 ZONE_IS_INITIALIZED,
222 223 ZONE_IS_READY,
223 224 ZONE_IS_BOOTING,
224 225 ZONE_IS_RUNNING,
225 226 ZONE_IS_SHUTTING_DOWN,
226 227 ZONE_IS_EMPTY,
227 228 ZONE_IS_DOWN,
228 229 ZONE_IS_DYING,
229 230 ZONE_IS_DEAD
230 231 } zone_status_t;
231 232 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
232 233 #define ZONE_MAX_STATE ZONE_IS_DEAD
233 234
234 235 /*
235 236 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
236 237 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
237 238 */
238 239 typedef enum zone_cmd {
239 240 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
240 241 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
241 242 } zone_cmd_t;
242 243
243 244 /*
244 245 * The structure of a request to zoneadmd.
245 246 */
246 247 typedef struct zone_cmd_arg {
247 248 uint64_t uniqid; /* unique "generation number" */
248 249 zone_cmd_t cmd; /* requested action */
249 250 uint32_t _pad; /* need consistent 32/64 bit alignmt */
250 251 char locale[MAXPATHLEN]; /* locale in which to render messages */
251 252 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
252 253 } zone_cmd_arg_t;
253 254
254 255 /*
255 256 * Structure of zoneadmd's response to a request. A NULL return value means
256 257 * the caller should attempt to restart zoneadmd and retry.
257 258 */
258 259 typedef struct zone_cmd_rval {
259 260 int rval; /* return value of request */
260 261 char errbuf[1]; /* variable-sized buffer containing error messages */
261 262 } zone_cmd_rval_t;
262 263
263 264 /*
264 265 * The zone support infrastructure uses the zone name as a component
265 266 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
266 267 * in length, so ZONENAME_MAX is limited by that.
267 268 */
268 269 #define ZONENAME_MAX 64
269 270
270 271 #define GLOBAL_ZONENAME "global"
271 272
272 273 /*
273 274 * Extended Regular expression (see regex(5)) which matches all valid zone
274 275 * names.
275 276 */
276 277 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
277 278
278 279 /*
279 280 * Where the zones support infrastructure places temporary files.
280 281 */
281 282 #define ZONES_TMPDIR "/var/run/zones"
282 283
283 284 /*
284 285 * The path to the door used by clients to communicate with zoneadmd.
285 286 */
286 287 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
287 288
288 289
289 290 /* zone_flags */
290 291 /*
291 292 * Threads that read or write the following flag must hold zone_lock.
292 293 */
293 294 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
294 295
295 296 /*
296 297 * The following threads are set when the zone is created and never changed.
297 298 * Threads that test for these flags don't have to hold zone_lock.
298 299 */
299 300 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
300 301 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
301 302 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
302 303
303 304
304 305 /* zone_create flags */
305 306 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
306 307
307 308 /* zone network properties */
308 309 #define ZONE_NETWORK_ADDRESS 1
309 310 #define ZONE_NETWORK_DEFROUTER 2
310 311
311 312 #define ZONE_NET_ADDRNAME "address"
312 313 #define ZONE_NET_RTRNAME "route"
313 314
314 315 typedef struct zone_net_data {
315 316 int zn_type;
316 317 int zn_len;
317 318 datalink_id_t zn_linkid;
318 319 uint8_t zn_val[1];
319 320 } zone_net_data_t;
320 321
321 322
322 323 #ifdef _KERNEL
|
↓ open down ↓ |
268 lines elided |
↑ open up ↑ |
323 324
324 325 /*
325 326 * We need to protect the definition of 'list_t' from userland applications and
326 327 * libraries which may be defining ther own versions.
327 328 */
328 329 #include <sys/list.h>
329 330 #include <sys/loadavg.h>
330 331
331 332 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
332 333
334 +/*
335 + * Indexes into ->zone_ustate array, summing the micro state of all threads in a
336 + * particular zone.
337 + */
338 +#define ZONE_USTATE_STIME (0)
339 +#define ZONE_USTATE_UTIME (1)
340 +#define ZONE_USTATE_WTIME (2)
341 +#define ZONE_USTATE_MAX (3)
342 +
333 343 struct pool;
334 344 struct brand;
335 345
336 346 /*
337 347 * Each of these constants identifies a kernel subsystem that acquires and
338 348 * releases zone references. Each subsystem that invokes
339 349 * zone_hold_ref() and zone_rele_ref() should specify the
340 350 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
341 351 * help users and developers quickly identify subsystems that stall zone
342 352 * shutdowns indefinitely.
343 353 *
344 354 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
345 355 * when you modify this enumeration.
346 356 */
347 357 typedef enum zone_ref_subsys {
348 358 ZONE_REF_NFS, /* NFS */
349 359 ZONE_REF_NFSV4, /* NFSv4 */
350 360 ZONE_REF_SMBFS, /* SMBFS */
351 361 ZONE_REF_MNTFS, /* MNTFS */
352 362 ZONE_REF_LOFI, /* LOFI devices */
353 363 ZONE_REF_VFS, /* VFS infrastructure */
354 364 ZONE_REF_IPC, /* IPC infrastructure */
355 365 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
356 366 } zone_ref_subsys_t;
357 367
358 368 /*
359 369 * zone_ref represents a general-purpose references to a zone. Each zone's
360 370 * references are linked into the zone's zone_t::zone_ref_list. This allows
361 371 * debuggers to walk zones' references.
362 372 */
363 373 typedef struct zone_ref {
364 374 struct zone *zref_zone; /* the zone to which the reference refers */
365 375 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
366 376 } zone_ref_t;
367 377
368 378 /*
369 379 * Structure to record list of ZFS datasets exported to a zone.
370 380 */
371 381 typedef struct zone_dataset {
372 382 char *zd_dataset;
373 383 list_node_t zd_linkage;
374 384 } zone_dataset_t;
375 385
376 386 /*
377 387 * structure for zone kstats
378 388 */
379 389 typedef struct zone_kstat {
380 390 kstat_named_t zk_zonename;
381 391 kstat_named_t zk_usage;
382 392 kstat_named_t zk_value;
383 393 } zone_kstat_t;
384 394
385 395 struct cpucap;
386 396
387 397 typedef struct {
388 398 kstat_named_t zm_zonename;
389 399 kstat_named_t zm_pgpgin;
390 400 kstat_named_t zm_anonpgin;
391 401 kstat_named_t zm_execpgin;
392 402 kstat_named_t zm_fspgin;
393 403 kstat_named_t zm_anon_alloc_fail;
394 404 } zone_mcap_kstat_t;
395 405
396 406 typedef struct {
397 407 kstat_named_t zm_zonename; /* full name, kstat truncates name */
398 408 kstat_named_t zm_utime;
399 409 kstat_named_t zm_stime;
400 410 kstat_named_t zm_wtime;
401 411 kstat_named_t zm_avenrun1;
402 412 kstat_named_t zm_avenrun5;
403 413 kstat_named_t zm_avenrun15;
404 414 kstat_named_t zm_ffcap;
405 415 kstat_named_t zm_ffnoproc;
406 416 kstat_named_t zm_ffnomem;
407 417 kstat_named_t zm_ffmisc;
408 418 kstat_named_t zm_nested_intp;
409 419 kstat_named_t zm_init_pid;
410 420 kstat_named_t zm_boot_time;
411 421 } zone_misc_kstat_t;
412 422
413 423 typedef struct zone {
414 424 /*
415 425 * zone_name is never modified once set.
416 426 */
417 427 char *zone_name; /* zone's configuration name */
418 428 /*
419 429 * zone_nodename and zone_domain are never freed once allocated.
420 430 */
421 431 char *zone_nodename; /* utsname.nodename equivalent */
422 432 char *zone_domain; /* srpc_domain equivalent */
423 433 /*
424 434 * zone_hostid is used for per-zone hostid emulation.
425 435 * Currently it isn't modified after it's set (so no locks protect
|
↓ open down ↓ |
83 lines elided |
↑ open up ↑ |
426 436 * accesses), but that might have to change when we allow
427 437 * administrators to change running zones' properties.
428 438 *
429 439 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
430 440 * that zone_get_hostid() will function correctly.
431 441 */
432 442 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
433 443 /* if not emulated */
434 444 /*
435 445 * zone_lock protects the following fields of a zone_t:
436 - * zone_ref
437 - * zone_cred_ref
438 - * zone_subsys_ref
439 - * zone_ref_list
440 - * zone_ntasks
441 - * zone_flags
442 - * zone_zsd
446 + * zone_ref
447 + * zone_cred_ref
448 + * zone_subsys_ref
449 + * zone_ref_list
450 + * zone_ntasks
451 + * zone_flags
452 + * zone_zsd
443 453 * zone_pfexecd
444 454 */
445 455 kmutex_t zone_lock;
446 456 /*
447 457 * zone_linkage is the zone's linkage into the active or
448 458 * death-row list. The field is protected by zonehash_lock.
449 459 */
450 460 list_node_t zone_linkage;
451 461 zoneid_t zone_id; /* ID of zone */
452 462 uint_t zone_ref; /* count of zone_hold()s on zone */
453 463 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
454 464 /*
455 465 * Fixed-sized array of subsystem-specific reference counts
456 466 * The sum of all of the counts must be less than or equal to zone_ref.
457 467 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
458 468 * constants.
459 469 */
460 470 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
461 471 list_t zone_ref_list; /* list of zone_ref_t structs */
462 472 /*
463 473 * zone_rootvp and zone_rootpath can never be modified once set.
464 474 */
465 475 struct vnode *zone_rootvp; /* zone's root vnode */
466 476 char *zone_rootpath; /* Path to zone's root + '/' */
467 477 ushort_t zone_flags; /* misc flags */
468 478 zone_status_t zone_status; /* protected by zone_status_lock */
469 479 uint_t zone_ntasks; /* number of tasks executing in zone */
470 480 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
471 481 /* counters in projects and tasks */
472 482 /* that are within the zone */
473 483 rctl_qty_t zone_nlwps; /* number of lwps in zone */
474 484 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
475 485 rctl_qty_t zone_shmmax; /* System V shared memory usage */
476 486 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
477 487
478 488 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
479 489 uint32_t zone_shares; /* FSS shares allocated to zone */
480 490 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
481 491 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
482 492 /* kpd_locked_mem for all */
483 493 /* projects in zone. */
484 494 /* Also protects zone_max_swap */
485 495 /* grab after p_lock, before rcs_lock */
486 496 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
487 497 /* zone */
488 498 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
489 499 /* limit. Protected by */
490 500 /* zone_rctls->rcs_lock */
491 501 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
492 502 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
493 503 /* Protected by */
494 504 /* zone_rctls->rcs_lock */
495 505 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
496 506 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
497 507 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
498 508 /* Protected by */
499 509 /* zone_rctls->rcs_lock */
500 510 list_t zone_zsd; /* list of Zone-Specific Data values */
501 511 kcondvar_t zone_cv; /* used to signal state changes */
502 512 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
503 513 pid_t zone_proc_initpid; /* pid of "init" for this zone */
504 514 char *zone_initname; /* fs path to 'init' */
505 515 int zone_boot_err; /* for zone_boot() if boot fails */
506 516 char *zone_bootargs; /* arguments passed via zone_boot() */
507 517 uint64_t zone_phys_mcap; /* physical memory cap */
508 518 /*
509 519 * zone_kthreads is protected by zone_status_lock.
510 520 */
511 521 kthread_t *zone_kthreads; /* kernel threads in zone */
512 522 struct priv_set *zone_privset; /* limit set for zone */
513 523 /*
514 524 * zone_vfslist is protected by vfs_list_lock().
515 525 */
516 526 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
517 527 uint64_t zone_uniqid; /* unique zone generation number */
518 528 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
519 529 /*
520 530 * zone_pool is protected by pool_lock().
521 531 */
522 532 struct pool *zone_pool; /* pool the zone is bound to */
523 533 hrtime_t zone_pool_mod; /* last pool bind modification time */
524 534 /* zone_psetid is protected by cpu_lock */
525 535 psetid_t zone_psetid; /* pset the zone is bound to */
526 536
527 537 time_t zone_boot_time; /* Similar to boot_time */
528 538
529 539 /*
530 540 * The following two can be read without holding any locks. They are
531 541 * updated under cpu_lock.
532 542 */
533 543 int zone_ncpus; /* zone's idea of ncpus */
534 544 int zone_ncpus_online; /* zone's idea of ncpus_online */
535 545 /*
|
↓ open down ↓ |
83 lines elided |
↑ open up ↑ |
536 546 * List of ZFS datasets exported to this zone.
537 547 */
538 548 list_t zone_datasets; /* list of datasets */
539 549
540 550 ts_label_t *zone_slabel; /* zone sensitivity label */
541 551 int zone_match; /* require label match for packets */
542 552 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
543 553
544 554 boolean_t zone_restart_init; /* Restart init if it dies? */
545 555 struct brand *zone_brand; /* zone's brand */
546 - void *zone_brand_data; /* store brand specific data */
556 + void *zone_brand_data; /* store brand specific data */
547 557 id_t zone_defaultcid; /* dflt scheduling class id */
548 558 kstat_t *zone_swapresv_kstat;
549 559 kstat_t *zone_lockedmem_kstat;
550 560 /*
551 561 * zone_dl_list is protected by zone_lock
552 562 */
553 563 list_t zone_dl_list;
554 564 netstack_t *zone_netstack;
555 565 struct cpucap *zone_cpucap; /* CPU caps data */
556 566 /*
557 567 * Solaris Auditing per-zone audit context
558 568 */
559 569 struct au_kcontext *zone_audit_kctxt;
560 570 /*
561 571 * For private use by mntfs.
562 572 */
563 573 struct mntelem *zone_mntfs_db;
564 574 krwlock_t zone_mntfs_db_lock;
565 575
566 576 struct klpd_reg *zone_pfexecd;
567 577
568 578 char *zone_fs_allowed;
569 579 rctl_qty_t zone_nprocs; /* number of processes in the zone */
570 580 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
571 581 /* zone_rctls->rcs_lock */
572 582 kstat_t *zone_nprocs_kstat;
573 583
574 584 kmutex_t zone_mcap_lock; /* protects mcap statistics */
575 585 kstat_t *zone_mcap_ksp;
576 586 zone_mcap_kstat_t *zone_mcap_stats;
|
↓ open down ↓ |
20 lines elided |
↑ open up ↑ |
577 587 uint64_t zone_pgpgin; /* pages paged in */
578 588 uint64_t zone_anonpgin; /* anon pages paged in */
579 589 uint64_t zone_execpgin; /* exec pages paged in */
580 590 uint64_t zone_fspgin; /* fs pages paged in */
581 591 uint64_t zone_anon_alloc_fail; /* cnt of anon alloc fails */
582 592
583 593 psecflags_t zone_secflags; /* default zone security-flags */
584 594
585 595 /*
586 596 * Misc. kstats and counters for zone cpu-usage aggregation.
587 - * The zone_Xtime values are the sum of the micro-state accounting
588 - * values for all threads that are running or have run in the zone.
589 - * This is tracked in msacct.c as threads change state.
590 - * The zone_stime is the sum of the LMS_SYSTEM times.
591 - * The zone_utime is the sum of the LMS_USER times.
592 - * The zone_wtime is the sum of the LMS_WAIT_CPU times.
593 - * As with per-thread micro-state accounting values, these values are
594 - * not scaled to nanosecs. The scaling is done by the
595 - * zone_misc_kstat_update function when kstats are requested.
596 597 */
597 598 kmutex_t zone_misc_lock; /* protects misc statistics */
598 599 kstat_t *zone_misc_ksp;
599 600 zone_misc_kstat_t *zone_misc_stats;
600 - uint64_t zone_stime; /* total system time */
601 - uint64_t zone_utime; /* total user time */
602 - uint64_t zone_wtime; /* total time waiting in runq */
601 + /* Accumulated microstate for all threads in this zone. */
602 + cpu_uarray_t *zone_ustate;
603 603 /* fork-fail kstat tracking */
604 604 uint32_t zone_ffcap; /* hit an rctl cap */
605 605 uint32_t zone_ffnoproc; /* get proc/lwp error */
606 606 uint32_t zone_ffnomem; /* as_dup/memory error */
607 607 uint32_t zone_ffmisc; /* misc. other error */
608 608
609 609 uint32_t zone_nested_intp; /* nested interp. kstat */
610 610
611 611 struct loadavg_s zone_loadavg; /* loadavg for this zone */
612 612 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
613 613 int zone_avenrun[3]; /* FSCALED avg. run queue len */
614 614
615 615 /*
616 616 * FSS stats updated once per second by fss_decay_usage.
617 617 */
618 618 uint32_t zone_fss_gen; /* FSS generation cntr */
619 619 uint64_t zone_run_ticks; /* tot # of ticks running */
620 620
621 621 /*
622 622 * DTrace-private per-zone state
623 623 */
624 624 int zone_dtrace_getf; /* # of unprivileged getf()s */
625 625
626 626 /*
627 627 * Synchronization primitives used to synchronize between mounts and
628 628 * zone creation/destruction.
629 629 */
630 630 int zone_mounts_in_progress;
631 631 kcondvar_t zone_mount_cv;
632 632 kmutex_t zone_mount_lock;
633 633 } zone_t;
634 634
635 635 /*
636 636 * Special value of zone_psetid to indicate that pools are disabled.
637 637 */
638 638 #define ZONE_PS_INVAL PS_MYID
639 639
640 640
641 641 extern zone_t zone0;
642 642 extern zone_t *global_zone;
643 643 extern uint_t maxzones;
644 644 extern rctl_hndl_t rc_zone_nlwps;
645 645 extern rctl_hndl_t rc_zone_nprocs;
646 646
647 647 extern long zone(int, void *, void *, void *, void *);
648 648 extern void zone_zsd_init(void);
649 649 extern void zone_init(void);
650 650 extern void zone_hold(zone_t *);
651 651 extern void zone_rele(zone_t *);
652 652 extern void zone_init_ref(zone_ref_t *);
653 653 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
654 654 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
655 655 extern void zone_cred_hold(zone_t *);
656 656 extern void zone_cred_rele(zone_t *);
657 657 extern void zone_task_hold(zone_t *);
658 658 extern void zone_task_rele(zone_t *);
659 659 extern zone_t *zone_find_by_id(zoneid_t);
660 660 extern zone_t *zone_find_by_label(const ts_label_t *);
661 661 extern zone_t *zone_find_by_name(char *);
662 662 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
663 663 extern zone_t *zone_find_by_path(const char *);
664 664 extern zoneid_t getzoneid(void);
665 665 extern zone_t *zone_find_by_id_nolock(zoneid_t);
666 666 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
667 667 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
668 668 extern void zone_loadavg_update();
669 669
670 670 /*
671 671 * Zone-specific data (ZSD) APIs
672 672 */
673 673 /*
|
↓ open down ↓ |
61 lines elided |
↑ open up ↑ |
674 674 * The following is what code should be initializing its zone_key_t to if it
675 675 * calls zone_getspecific() without necessarily knowing that zone_key_create()
676 676 * has been called on the key.
677 677 */
678 678 #define ZONE_KEY_UNINITIALIZED 0
679 679
680 680 typedef uint_t zone_key_t;
681 681
682 682 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
683 683 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
684 -extern int zone_key_delete(zone_key_t);
684 +extern int zone_key_delete(zone_key_t);
685 685 extern void *zone_getspecific(zone_key_t, zone_t *);
686 686 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
687 687
688 688 /*
689 689 * The definition of a zsd_entry is truly private to zone.c and is only
690 690 * placed here so it can be shared with mdb.
691 691 *
692 692 * State maintained for each zone times each registered key, which tracks
693 693 * the state of the create, shutdown and destroy callbacks.
694 694 *
695 695 * zsd_flags is used to keep track of pending actions to avoid holding locks
696 696 * when calling the create/shutdown/destroy callbacks, since doing so
697 697 * could lead to deadlocks.
698 698 */
699 699 struct zsd_entry {
|
↓ open down ↓ |
5 lines elided |
↑ open up ↑ |
700 700 zone_key_t zsd_key; /* Key used to lookup value */
701 701 void *zsd_data; /* Caller-managed value */
702 702 /*
703 703 * Callbacks to be executed when a zone is created, shutdown, and
704 704 * destroyed, respectively.
705 705 */
706 706 void *(*zsd_create)(zoneid_t);
707 707 void (*zsd_shutdown)(zoneid_t, void *);
708 708 void (*zsd_destroy)(zoneid_t, void *);
709 709 list_node_t zsd_linkage;
710 - uint16_t zsd_flags; /* See below */
710 + uint16_t zsd_flags; /* See below */
711 711 kcondvar_t zsd_cv;
712 712 };
713 713
714 714 /*
715 715 * zsd_flags
716 716 */
717 717 #define ZSD_CREATE_NEEDED 0x0001
718 718 #define ZSD_CREATE_INPROGRESS 0x0002
719 719 #define ZSD_CREATE_COMPLETED 0x0004
720 720 #define ZSD_SHUTDOWN_NEEDED 0x0010
721 721 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
722 722 #define ZSD_SHUTDOWN_COMPLETED 0x0040
723 723 #define ZSD_DESTROY_NEEDED 0x0100
724 724 #define ZSD_DESTROY_INPROGRESS 0x0200
725 725 #define ZSD_DESTROY_COMPLETED 0x0400
726 726
727 727 #define ZSD_CREATE_ALL \
728 728 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
729 729 #define ZSD_SHUTDOWN_ALL \
730 730 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
731 731 #define ZSD_DESTROY_ALL \
732 732 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
733 733
734 734 #define ZSD_ALL_INPROGRESS \
735 735 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
736 736
737 737 /*
738 738 * Macros to help with zone visibility restrictions.
739 739 */
740 740
741 741 /*
742 742 * Is process in the global zone?
743 743 */
744 744 #define INGLOBALZONE(p) \
745 745 ((p)->p_zone == global_zone)
746 746
747 747 /*
748 748 * Can process view objects in given zone?
749 749 */
750 750 #define HASZONEACCESS(p, zoneid) \
751 751 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
752 752
753 753 /*
754 754 * Convenience macro to see if a resolved path is visible from within a
755 755 * given zone.
756 756 *
757 757 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
758 758 * two strings must be equal. Since the rootpathlen has a trailing '/',
759 759 * we want to skip everything in the path up to (but not including) the
760 760 * trailing '/'.
761 761 */
762 762 #define ZONE_PATH_VISIBLE(path, zone) \
763 763 (strncmp((path), (zone)->zone_rootpath, \
764 764 (zone)->zone_rootpathlen - 1) == 0)
765 765
766 766 /*
767 767 * Convenience macro to go from the global view of a path to that seen
768 768 * from within said zone. It is the responsibility of the caller to
769 769 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
770 770 * in fact visible from within the zone.
771 771 */
772 772 #define ZONE_PATH_TRANSLATE(path, zone) \
773 773 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
774 774 (path) + (zone)->zone_rootpathlen - 2)
775 775
776 776 /*
777 777 * Special processes visible in all zones.
778 778 */
779 779 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
780 780
781 781 /*
782 782 * Zone-safe version of thread_create() to be used when the caller wants to
783 783 * create a kernel thread to run within the current zone's context.
784 784 */
785 785 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
786 786 pri_t);
787 787 extern void zthread_exit(void);
788 788
789 789 /*
790 790 * Functions for an external observer to register interest in a zone's status
791 791 * change. Observers will be woken up when the zone status equals the status
792 792 * argument passed in (in the case of zone_status_timedwait, the function may
793 793 * also return because of a timeout; zone_status_wait_sig may return early due
794 794 * to a signal being delivered; zone_status_timedwait_sig may return for any of
795 795 * the above reasons).
796 796 *
797 797 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
798 798 * cv_wait_sig() respectively.
799 799 */
800 800 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
801 801 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
802 802 extern void zone_status_wait(zone_t *, zone_status_t);
803 803 extern int zone_status_wait_sig(zone_t *, zone_status_t);
804 804
805 805 /*
806 806 * Get the status of the zone (at the time it was called). The state may
807 807 * have progressed by the time it is returned.
808 808 */
809 809 extern zone_status_t zone_status_get(zone_t *);
810 810
811 811 /*
812 812 * Safely get the hostid of the specified zone (defaults to machine's hostid
813 813 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
814 814 * the global zone's (i.e., physical system's) hostid.
815 815 */
816 816 extern uint32_t zone_get_hostid(zone_t *);
817 817
818 818 /*
819 819 * Get the "kcred" credentials corresponding to the given zone.
820 820 */
821 821 extern struct cred *zone_get_kcred(zoneid_t);
822 822
823 823 /*
824 824 * Get/set the pool the zone is currently bound to.
825 825 */
826 826 extern struct pool *zone_pool_get(zone_t *);
827 827 extern void zone_pool_set(zone_t *, struct pool *);
828 828
829 829 /*
830 830 * Get/set the pset the zone is currently using.
831 831 */
832 832 extern psetid_t zone_pset_get(zone_t *);
833 833 extern void zone_pset_set(zone_t *, psetid_t);
834 834
835 835 /*
836 836 * Get the number of cpus/online-cpus visible from the given zone.
837 837 */
838 838 extern int zone_ncpus_get(zone_t *);
839 839 extern int zone_ncpus_online_get(zone_t *);
840 840
841 841 /*
842 842 * Returns true if the named pool/dataset is visible in the current zone.
843 843 */
844 844 extern int zone_dataset_visible(const char *, int *);
845 845
846 846 /*
847 847 * zone version of kadmin()
848 848 */
849 849 extern int zone_kadmin(int, int, const char *, cred_t *);
850 850 extern void zone_shutdown_global(void);
851 851
852 852 extern void mount_in_progress(zone_t *);
853 853 extern void mount_completed(zone_t *);
854 854
855 855 extern int zone_walk(int (*)(zone_t *, void *), void *);
856 856
857 857 extern rctl_hndl_t rc_zone_locked_mem;
858 858 extern rctl_hndl_t rc_zone_max_swap;
859 859 extern rctl_hndl_t rc_zone_max_lofi;
860 860
861 861 #endif /* _KERNEL */
862 862
863 863 #ifdef __cplusplus
864 864 }
865 865 #endif
866 866
867 867 #endif /* _SYS_ZONE_H */
|
↓ open down ↓ |
147 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX