Print this page
Remove most KEBE comments and accompanying unused code or variables/fields.
zone.h mismerge
OS-338 Kstat counters to show "slow" VFS operations
OS-5187 improve /proc/diskstat handling
Reviewed by: Patrick Mooney <patrick.mooney@joyent.com>
OS-5179 flatten zvol entries for /dev and /proc/partitions
Reviewed by: Patrick Mooney <patrick.mooney@joyent.com>
OS-4915 want FX high priority zone configuration option
OS-4925 ps pri shows misleading value for zone in RT scheduling class
Reviewed by: Patrick Mooney <patrick.mooney@joyent.com>
OS-4781 would like to be able to add CT_PR_EV_EXIT to fatal event set of current contract
Bad mismerge of zone.h
OS-3820 lxbrand ptrace(2): the next generation
OS-3685 lxbrand PTRACE_O_TRACEFORK race condition
OS-3834 lxbrand 64-bit strace(1) reports 64-bit process as using x32 ABI
OS-3794 lxbrand panic on init signal death
Reviewed by: Robert Mustacchi <rm@joyent.com>
Reviewed by: Jerry Jelinek <jerry.jelinek@joyent.com>
Reviewed by: Bryan Cantrill <bryan@joyent.com>
OS-3050 Would like boot_time and init_pid for zones exposed to GZ kstats
OS-3429 Expose zone's init exit status
OS-803 make phys mem cap a bit harder
OS-1043 minimize vm_getusage impact
OS-11 rcapd behaves poorly when under extreme load
OS-399 zone phys. mem. cap should be a rctl and have associated kstat
| 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 2015 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, 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
|
↓ open down ↓ |
54 lines elided |
↑ open up ↑ |
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 -#define ZONE_ATTR_PHYS_MCAP 12
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
107 +#define ZONE_ATTR_DID 18
108 +#define ZONE_ATTR_PMCAP_PAGEOUT 19
106 109 #define ZONE_ATTR_INITNORESTART 20
110 +#define ZONE_ATTR_PG_FLT_DELAY 21
111 +#define ZONE_ATTR_RSS 22
112 +#define ZONE_ATTR_APP_SVC_CT 23
113 +#define ZONE_ATTR_SCHED_FIXEDHI 24
107 114
108 115 /* Start of the brand-specific attribute namespace */
109 116 #define ZONE_ATTR_BRAND_ATTRS 32768
110 117
111 118 #define ZONE_FS_ALLOWED_MAX 1024
112 119
113 120 #define ZONE_EVENT_CHANNEL "com.sun:zones:status"
114 121 #define ZONE_EVENT_STATUS_CLASS "status"
115 122 #define ZONE_EVENT_STATUS_SUBCLASS "change"
116 123
117 124 #define ZONE_EVENT_UNINITIALIZED "uninitialized"
118 125 #define ZONE_EVENT_INITIALIZED "initialized"
119 126 #define ZONE_EVENT_READY "ready"
120 127 #define ZONE_EVENT_RUNNING "running"
121 128 #define ZONE_EVENT_SHUTTING_DOWN "shutting_down"
122 129
123 130 #define ZONE_CB_NAME "zonename"
124 131 #define ZONE_CB_NEWSTATE "newstate"
125 132 #define ZONE_CB_OLDSTATE "oldstate"
126 133 #define ZONE_CB_TIMESTAMP "when"
127 134 #define ZONE_CB_ZONEID "zoneid"
128 135
129 136 /*
130 137 * Exit values that may be returned by scripts or programs invoked by various
131 138 * zone commands.
132 139 *
133 140 * These are defined as:
134 141 *
135 142 * ZONE_SUBPROC_OK
136 143 * ===============
137 144 * The subprocess completed successfully.
138 145 *
139 146 * ZONE_SUBPROC_USAGE
140 147 * ==================
141 148 * The subprocess failed with a usage message, or a usage message should
142 149 * be output in its behalf.
143 150 *
144 151 * ZONE_SUBPROC_NOTCOMPLETE
145 152 * ========================
146 153 * The subprocess did not complete, but the actions performed by the
147 154 * subprocess require no recovery actions by the user.
148 155 *
149 156 * For example, if the subprocess were called by "zoneadm install," the
150 157 * installation of the zone did not succeed but the user need not perform
151 158 * a "zoneadm uninstall" before attempting another install.
152 159 *
153 160 * ZONE_SUBPROC_FATAL
154 161 * ==================
155 162 * The subprocess failed in a fatal manner, usually one that will require
156 163 * some type of recovery action by the user.
157 164 *
158 165 * For example, if the subprocess were called by "zoneadm install," the
159 166 * installation of the zone did not succeed and the user will need to
160 167 * perform a "zoneadm uninstall" before another install attempt is
161 168 * possible.
162 169 *
163 170 * The non-success exit values are large to avoid accidental collision
164 171 * with values used internally by some commands (e.g. "Z_ERR" and
165 172 * "Z_USAGE" as used by zoneadm.)
166 173 */
167 174 #define ZONE_SUBPROC_OK 0
168 175 #define ZONE_SUBPROC_USAGE 253
169 176 #define ZONE_SUBPROC_NOTCOMPLETE 254
170 177 #define ZONE_SUBPROC_FATAL 255
171 178
172 179 #ifdef _SYSCALL32
173 180 typedef struct {
174 181 caddr32_t zone_name;
175 182 caddr32_t zone_root;
176 183 caddr32_t zone_privs;
177 184 size32_t zone_privssz;
178 185 caddr32_t rctlbuf;
179 186 size32_t rctlbufsz;
180 187 caddr32_t extended_error;
181 188 caddr32_t zfsbuf;
182 189 size32_t zfsbufsz;
183 190 int match; /* match level */
184 191 uint32_t doi; /* DOI for label */
185 192 caddr32_t label; /* label associated with zone */
186 193 int flags;
187 194 } zone_def32;
188 195 #endif
189 196 typedef struct {
190 197 const char *zone_name;
191 198 const char *zone_root;
192 199 const struct priv_set *zone_privs;
193 200 size_t zone_privssz;
194 201 const char *rctlbuf;
195 202 size_t rctlbufsz;
196 203 int *extended_error;
197 204 const char *zfsbuf;
198 205 size_t zfsbufsz;
199 206 int match; /* match level */
200 207 uint32_t doi; /* DOI for label */
201 208 const bslabel_t *label; /* label associated with zone */
202 209 int flags;
203 210 } zone_def;
204 211
205 212 /* extended error information */
206 213 #define ZE_UNKNOWN 0 /* No extended error info */
207 214 #define ZE_CHROOTED 1 /* tried to zone_create from chroot */
208 215 #define ZE_AREMOUNTS 2 /* there are mounts within the zone */
209 216 #define ZE_LABELINUSE 3 /* label is already in use by some other zone */
210 217
211 218 /*
212 219 * zone_status values
213 220 *
214 221 * You must modify zone_status_names in mdb(1M)'s genunix module
215 222 * (genunix/zone.c) when you modify this enum.
216 223 */
217 224 typedef enum {
218 225 ZONE_IS_UNINITIALIZED = 0,
219 226 ZONE_IS_INITIALIZED,
220 227 ZONE_IS_READY,
221 228 ZONE_IS_BOOTING,
222 229 ZONE_IS_RUNNING,
223 230 ZONE_IS_SHUTTING_DOWN,
224 231 ZONE_IS_EMPTY,
225 232 ZONE_IS_DOWN,
226 233 ZONE_IS_DYING,
227 234 ZONE_IS_DEAD
228 235 } zone_status_t;
229 236 #define ZONE_MIN_STATE ZONE_IS_UNINITIALIZED
230 237 #define ZONE_MAX_STATE ZONE_IS_DEAD
231 238
232 239 /*
233 240 * Valid commands which may be issued by zoneadm to zoneadmd. The kernel also
234 241 * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
235 242 */
236 243 typedef enum zone_cmd {
|
↓ open down ↓ |
120 lines elided |
↑ open up ↑ |
237 244 Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
238 245 Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
239 246 } zone_cmd_t;
240 247
241 248 /*
242 249 * The structure of a request to zoneadmd.
243 250 */
244 251 typedef struct zone_cmd_arg {
245 252 uint64_t uniqid; /* unique "generation number" */
246 253 zone_cmd_t cmd; /* requested action */
247 - uint32_t _pad; /* need consistent 32/64 bit alignmt */
254 + int status; /* init status on shutdown */
255 + uint32_t debug; /* enable brand hook debug */
248 256 char locale[MAXPATHLEN]; /* locale in which to render messages */
249 257 char bootbuf[BOOTARGS_MAX]; /* arguments passed to zone_boot() */
258 + /* Needed for 32/64 zoneadm -> zoneadmd door arg size check. */
259 + int pad;
250 260 } zone_cmd_arg_t;
251 261
252 262 /*
253 263 * Structure of zoneadmd's response to a request. A NULL return value means
254 264 * the caller should attempt to restart zoneadmd and retry.
255 265 */
256 266 typedef struct zone_cmd_rval {
257 267 int rval; /* return value of request */
258 268 char errbuf[1]; /* variable-sized buffer containing error messages */
259 269 } zone_cmd_rval_t;
260 270
261 271 /*
262 272 * The zone support infrastructure uses the zone name as a component
263 273 * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
264 274 * in length, so ZONENAME_MAX is limited by that.
265 275 */
266 276 #define ZONENAME_MAX 64
267 277
268 278 #define GLOBAL_ZONENAME "global"
269 279
270 280 /*
271 281 * Extended Regular expression (see regex(5)) which matches all valid zone
272 282 * names.
273 283 */
274 284 #define ZONENAME_REGEXP "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
275 285
276 286 /*
277 287 * Where the zones support infrastructure places temporary files.
278 288 */
279 289 #define ZONES_TMPDIR "/var/run/zones"
280 290
281 291 /*
282 292 * The path to the door used by clients to communicate with zoneadmd.
283 293 */
284 294 #define ZONE_DOOR_PATH ZONES_TMPDIR "/%s.zoneadmd_door"
285 295
286 296
287 297 /* zone_flags */
288 298 /*
289 299 * Threads that read or write the following flag must hold zone_lock.
290 300 */
291 301 #define ZF_REFCOUNTS_LOGGED 0x1 /* a thread logged the zone's refs */
292 302
293 303 /*
294 304 * The following threads are set when the zone is created and never changed.
295 305 * Threads that test for these flags don't have to hold zone_lock.
296 306 */
297 307 #define ZF_HASHED_LABEL 0x2 /* zone has a unique label */
298 308 #define ZF_IS_SCRATCH 0x4 /* scratch zone */
299 309 #define ZF_NET_EXCL 0x8 /* Zone has an exclusive IP stack */
300 310
301 311
302 312 /* zone_create flags */
303 313 #define ZCF_NET_EXCL 0x1 /* Create a zone with exclusive IP */
304 314
305 315 /* zone network properties */
306 316 #define ZONE_NETWORK_ADDRESS 1
307 317 #define ZONE_NETWORK_DEFROUTER 2
308 318
309 319 #define ZONE_NET_ADDRNAME "address"
310 320 #define ZONE_NET_RTRNAME "route"
311 321
312 322 typedef struct zone_net_data {
313 323 int zn_type;
314 324 int zn_len;
315 325 datalink_id_t zn_linkid;
316 326 uint8_t zn_val[1];
317 327 } zone_net_data_t;
318 328
319 329
320 330 #ifdef _KERNEL
321 331
322 332 /*
323 333 * We need to protect the definition of 'list_t' from userland applications and
324 334 * libraries which may be defining ther own versions.
325 335 */
326 336 #include <sys/list.h>
327 337 #include <sys/loadavg.h>
328 338
329 339 #define GLOBAL_ZONEUNIQID 0 /* uniqid of the global zone */
330 340
331 341 struct pool;
332 342 struct brand;
333 343
334 344 /*
335 345 * Each of these constants identifies a kernel subsystem that acquires and
336 346 * releases zone references. Each subsystem that invokes
337 347 * zone_hold_ref() and zone_rele_ref() should specify the
338 348 * zone_ref_subsys_t constant associated with the subsystem. Tracked holds
339 349 * help users and developers quickly identify subsystems that stall zone
340 350 * shutdowns indefinitely.
341 351 *
342 352 * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
343 353 * when you modify this enumeration.
344 354 */
345 355 typedef enum zone_ref_subsys {
346 356 ZONE_REF_NFS, /* NFS */
347 357 ZONE_REF_NFSV4, /* NFSv4 */
348 358 ZONE_REF_SMBFS, /* SMBFS */
349 359 ZONE_REF_MNTFS, /* MNTFS */
350 360 ZONE_REF_LOFI, /* LOFI devices */
351 361 ZONE_REF_VFS, /* VFS infrastructure */
352 362 ZONE_REF_IPC, /* IPC infrastructure */
353 363 ZONE_REF_NUM_SUBSYS /* This must be the last entry. */
354 364 } zone_ref_subsys_t;
355 365
356 366 /*
357 367 * zone_ref represents a general-purpose references to a zone. Each zone's
358 368 * references are linked into the zone's zone_t::zone_ref_list. This allows
359 369 * debuggers to walk zones' references.
360 370 */
361 371 typedef struct zone_ref {
362 372 struct zone *zref_zone; /* the zone to which the reference refers */
363 373 list_node_t zref_linkage; /* linkage for zone_t::zone_ref_list */
364 374 } zone_ref_t;
365 375
366 376 /*
367 377 * Structure to record list of ZFS datasets exported to a zone.
368 378 */
369 379 typedef struct zone_dataset {
370 380 char *zd_dataset;
371 381 list_node_t zd_linkage;
372 382 } zone_dataset_t;
373 383
374 384 /*
375 385 * structure for zone kstats
376 386 */
377 387 typedef struct zone_kstat {
378 388 kstat_named_t zk_zonename;
379 389 kstat_named_t zk_usage;
380 390 kstat_named_t zk_value;
381 391 } zone_kstat_t;
|
↓ open down ↓ |
122 lines elided |
↑ open up ↑ |
382 392
383 393 struct cpucap;
384 394
385 395 typedef struct {
386 396 hrtime_t cycle_start;
387 397 uint_t cycle_cnt;
388 398 hrtime_t zone_avg_cnt;
389 399 } sys_zio_cntr_t;
390 400
391 401 typedef struct {
402 + kstat_named_t zv_zonename;
403 + kstat_named_t zv_nread;
404 + kstat_named_t zv_reads;
405 + kstat_named_t zv_rtime;
406 + kstat_named_t zv_rlentime;
407 + kstat_named_t zv_rcnt;
408 + kstat_named_t zv_nwritten;
409 + kstat_named_t zv_writes;
410 + kstat_named_t zv_wtime;
411 + kstat_named_t zv_wlentime;
412 + kstat_named_t zv_wcnt;
413 + kstat_named_t zv_10ms_ops;
414 + kstat_named_t zv_100ms_ops;
415 + kstat_named_t zv_1s_ops;
416 + kstat_named_t zv_10s_ops;
417 + kstat_named_t zv_delay_cnt;
418 + kstat_named_t zv_delay_time;
419 +} zone_vfs_kstat_t;
420 +
421 +typedef struct {
422 + kstat_named_t zz_zonename;
423 + kstat_named_t zz_nread;
424 + kstat_named_t zz_reads;
425 + kstat_named_t zz_rtime;
426 + kstat_named_t zz_rlentime;
427 + kstat_named_t zz_nwritten;
428 + kstat_named_t zz_writes;
429 + kstat_named_t zz_waittime;
430 +} zone_zfs_kstat_t;
431 +
432 +typedef struct {
392 433 kstat_named_t zm_zonename;
434 + kstat_named_t zm_rss;
435 + kstat_named_t zm_phys_cap;
436 + kstat_named_t zm_swap;
437 + kstat_named_t zm_swap_cap;
438 + kstat_named_t zm_nover;
439 + kstat_named_t zm_pagedout;
393 440 kstat_named_t zm_pgpgin;
394 441 kstat_named_t zm_anonpgin;
395 442 kstat_named_t zm_execpgin;
396 443 kstat_named_t zm_fspgin;
397 444 kstat_named_t zm_anon_alloc_fail;
445 + kstat_named_t zm_pf_throttle;
446 + kstat_named_t zm_pf_throttle_usec;
398 447 } zone_mcap_kstat_t;
399 448
400 449 typedef struct {
401 450 kstat_named_t zm_zonename; /* full name, kstat truncates name */
402 451 kstat_named_t zm_utime;
403 452 kstat_named_t zm_stime;
404 453 kstat_named_t zm_wtime;
405 454 kstat_named_t zm_avenrun1;
406 455 kstat_named_t zm_avenrun5;
407 456 kstat_named_t zm_avenrun15;
408 457 kstat_named_t zm_ffcap;
409 458 kstat_named_t zm_ffnoproc;
410 459 kstat_named_t zm_ffnomem;
411 460 kstat_named_t zm_ffmisc;
412 461 kstat_named_t zm_nested_intp;
413 462 kstat_named_t zm_init_pid;
414 463 kstat_named_t zm_boot_time;
415 464 } zone_misc_kstat_t;
416 465
417 466 typedef struct zone {
418 467 /*
419 468 * zone_name is never modified once set.
420 469 */
421 470 char *zone_name; /* zone's configuration name */
422 471 /*
423 472 * zone_nodename and zone_domain are never freed once allocated.
424 473 */
425 474 char *zone_nodename; /* utsname.nodename equivalent */
426 475 char *zone_domain; /* srpc_domain equivalent */
427 476 /*
428 477 * zone_hostid is used for per-zone hostid emulation.
429 478 * Currently it isn't modified after it's set (so no locks protect
430 479 * accesses), but that might have to change when we allow
431 480 * administrators to change running zones' properties.
432 481 *
433 482 * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
434 483 * that zone_get_hostid() will function correctly.
435 484 */
436 485 uint32_t zone_hostid; /* zone's hostid, HW_INVALID_HOSTID */
437 486 /* if not emulated */
438 487 /*
439 488 * zone_lock protects the following fields of a zone_t:
440 489 * zone_ref
441 490 * zone_cred_ref
442 491 * zone_subsys_ref
443 492 * zone_ref_list
444 493 * zone_ntasks
445 494 * zone_flags
446 495 * zone_zsd
447 496 * zone_pfexecd
448 497 */
449 498 kmutex_t zone_lock;
450 499 /*
451 500 * zone_linkage is the zone's linkage into the active or
452 501 * death-row list. The field is protected by zonehash_lock.
453 502 */
454 503 list_node_t zone_linkage;
455 504 zoneid_t zone_id; /* ID of zone */
456 505 uint_t zone_ref; /* count of zone_hold()s on zone */
457 506 uint_t zone_cred_ref; /* count of zone_hold_cred()s on zone */
458 507 /*
459 508 * Fixed-sized array of subsystem-specific reference counts
460 509 * The sum of all of the counts must be less than or equal to zone_ref.
461 510 * The array is indexed by the counts' subsystems' zone_ref_subsys_t
462 511 * constants.
463 512 */
464 513 uint_t zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
465 514 list_t zone_ref_list; /* list of zone_ref_t structs */
466 515 /*
467 516 * zone_rootvp and zone_rootpath can never be modified once set.
468 517 */
469 518 struct vnode *zone_rootvp; /* zone's root vnode */
470 519 char *zone_rootpath; /* Path to zone's root + '/' */
471 520 ushort_t zone_flags; /* misc flags */
472 521 zone_status_t zone_status; /* protected by zone_status_lock */
473 522 uint_t zone_ntasks; /* number of tasks executing in zone */
474 523 kmutex_t zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
475 524 /* counters in projects and tasks */
476 525 /* that are within the zone */
477 526 rctl_qty_t zone_nlwps; /* number of lwps in zone */
478 527 rctl_qty_t zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
479 528 rctl_qty_t zone_shmmax; /* System V shared memory usage */
480 529 ipc_rqty_t zone_ipc; /* System V IPC id resource usage */
481 530
482 531 uint_t zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
483 532 uint32_t zone_shares; /* FSS shares allocated to zone */
484 533 rctl_set_t *zone_rctls; /* zone-wide (zone.*) rctls */
485 534 kmutex_t zone_mem_lock; /* protects zone_locked_mem and */
486 535 /* kpd_locked_mem for all */
487 536 /* projects in zone. */
488 537 /* Also protects zone_max_swap */
489 538 /* grab after p_lock, before rcs_lock */
490 539 rctl_qty_t zone_locked_mem; /* bytes of locked memory in */
491 540 /* zone */
492 541 rctl_qty_t zone_locked_mem_ctl; /* Current locked memory */
493 542 /* limit. Protected by */
494 543 /* zone_rctls->rcs_lock */
495 544 rctl_qty_t zone_max_swap; /* bytes of swap reserved by zone */
496 545 rctl_qty_t zone_max_swap_ctl; /* current swap limit. */
497 546 /* Protected by */
|
↓ open down ↓ |
90 lines elided |
↑ open up ↑ |
498 547 /* zone_rctls->rcs_lock */
499 548 kmutex_t zone_rctl_lock; /* protects zone_max_lofi */
500 549 rctl_qty_t zone_max_lofi; /* lofi devs for zone */
501 550 rctl_qty_t zone_max_lofi_ctl; /* current lofi limit. */
502 551 /* Protected by */
503 552 /* zone_rctls->rcs_lock */
504 553 list_t zone_zsd; /* list of Zone-Specific Data values */
505 554 kcondvar_t zone_cv; /* used to signal state changes */
506 555 struct proc *zone_zsched; /* Dummy kernel "zsched" process */
507 556 pid_t zone_proc_initpid; /* pid of "init" for this zone */
508 - char *zone_initname; /* fs path to 'init' */
557 + char *zone_initname; /* fs path to 'init' */
558 + int zone_init_status; /* init's exit status */
509 559 int zone_boot_err; /* for zone_boot() if boot fails */
510 560 char *zone_bootargs; /* arguments passed via zone_boot() */
511 - uint64_t zone_phys_mcap; /* physical memory cap */
561 + rctl_qty_t zone_phys_mem_ctl; /* current phys. memory limit */
512 562 /*
513 563 * zone_kthreads is protected by zone_status_lock.
514 564 */
515 565 kthread_t *zone_kthreads; /* kernel threads in zone */
516 566 struct priv_set *zone_privset; /* limit set for zone */
517 567 /*
518 568 * zone_vfslist is protected by vfs_list_lock().
519 569 */
520 570 struct vfs *zone_vfslist; /* list of FS's mounted in zone */
521 571 uint64_t zone_uniqid; /* unique zone generation number */
522 572 struct cred *zone_kcred; /* kcred-like, zone-limited cred */
523 573 /*
524 574 * zone_pool is protected by pool_lock().
525 575 */
526 576 struct pool *zone_pool; /* pool the zone is bound to */
527 577 hrtime_t zone_pool_mod; /* last pool bind modification time */
528 578 /* zone_psetid is protected by cpu_lock */
529 579 psetid_t zone_psetid; /* pset the zone is bound to */
530 580
531 581 time_t zone_boot_time; /* Similar to boot_time */
532 582
533 583 /*
534 584 * The following two can be read without holding any locks. They are
535 585 * updated under cpu_lock.
536 586 */
537 587 int zone_ncpus; /* zone's idea of ncpus */
538 588 int zone_ncpus_online; /* zone's idea of ncpus_online */
|
↓ open down ↓ |
17 lines elided |
↑ open up ↑ |
539 589 /*
540 590 * List of ZFS datasets exported to this zone.
541 591 */
542 592 list_t zone_datasets; /* list of datasets */
543 593
544 594 ts_label_t *zone_slabel; /* zone sensitivity label */
545 595 int zone_match; /* require label match for packets */
546 596 tsol_mlp_list_t zone_mlps; /* MLPs on zone-private addresses */
547 597
548 598 boolean_t zone_restart_init; /* Restart init if it dies? */
599 + boolean_t zone_reboot_on_init_exit; /* Reboot if init dies? */
600 + boolean_t zone_setup_app_contract; /* setup contract? */
549 601 struct brand *zone_brand; /* zone's brand */
550 602 void *zone_brand_data; /* store brand specific data */
551 603 id_t zone_defaultcid; /* dflt scheduling class id */
604 + boolean_t zone_fixed_hipri; /* fixed sched. hi prio */
552 605 kstat_t *zone_swapresv_kstat;
553 606 kstat_t *zone_lockedmem_kstat;
554 607 /*
555 608 * zone_dl_list is protected by zone_lock
556 609 */
557 610 list_t zone_dl_list;
558 611 netstack_t *zone_netstack;
559 612 struct cpucap *zone_cpucap; /* CPU caps data */
560 613
561 614 /*
562 615 * Data and counters used for ZFS fair-share disk IO.
563 616 */
564 617 rctl_qty_t zone_zfs_io_pri; /* ZFS IO priority */
565 618 uint_t zone_zfs_queued[2]; /* sync I/O enqueued count */
|
↓ open down ↓ |
4 lines elided |
↑ open up ↑ |
566 619 uint64_t zone_zfs_weight; /* used to prevent starvation */
567 620 uint64_t zone_io_util; /* IO utilization metric */
568 621 boolean_t zone_io_util_above_avg; /* IO util percent > avg. */
569 622 uint16_t zone_io_delay; /* IO delay on logical r/w */
570 623 kmutex_t zone_stg_io_lock; /* protects IO window data */
571 624 sys_zio_cntr_t zone_rd_ops; /* Counters for ZFS reads, */
572 625 sys_zio_cntr_t zone_wr_ops; /* writes and */
573 626 sys_zio_cntr_t zone_lwr_ops; /* logical writes. */
574 627
575 628 /*
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 + /*
636 + * kstats and counters for VFS ops and bytes.
637 + */
638 + kmutex_t zone_vfs_lock; /* protects VFS statistics */
639 + kstat_t *zone_vfs_ksp;
640 + kstat_io_t zone_vfs_rwstats;
641 + zone_vfs_kstat_t *zone_vfs_stats;
642 +
643 + /*
644 + * kstats for ZFS I/O ops and bytes.
645 + */
646 + kmutex_t zone_zfs_lock; /* protects ZFS statistics */
647 + kstat_io_t zone_zfs_rwstats;
648 + zone_zfs_kstat_t *zone_zfs_stats;
649 +
650 + /*
576 651 * Solaris Auditing per-zone audit context
577 652 */
578 653 struct au_kcontext *zone_audit_kctxt;
579 654 /*
580 655 * For private use by mntfs.
581 656 */
582 657 struct mntelem *zone_mntfs_db;
583 658 krwlock_t zone_mntfs_db_lock;
584 659
585 660 struct klpd_reg *zone_pfexecd;
586 661
587 662 char *zone_fs_allowed;
588 663 rctl_qty_t zone_nprocs; /* number of processes in the zone */
589 664 rctl_qty_t zone_nprocs_ctl; /* current limit protected by */
590 665 /* zone_rctls->rcs_lock */
591 666 kstat_t *zone_nprocs_kstat;
592 667
668 + /*
669 + * kstats and counters for physical memory capping.
670 + */
671 + rctl_qty_t zone_phys_mem; /* current bytes of phys. mem. (RSS) */
672 + kstat_t *zone_physmem_kstat;
673 + uint64_t zone_mcap_nover; /* # of times over phys. cap */
674 + uint64_t zone_mcap_pagedout; /* bytes of mem. paged out */
593 675 kmutex_t zone_mcap_lock; /* protects mcap statistics */
594 676 kstat_t *zone_mcap_ksp;
595 677 zone_mcap_kstat_t *zone_mcap_stats;
596 678 uint64_t zone_pgpgin; /* pages paged in */
597 679 uint64_t zone_anonpgin; /* anon pages paged in */
598 680 uint64_t zone_execpgin; /* exec pages paged in */
599 681 uint64_t zone_fspgin; /* fs pages paged in */
600 682 uint64_t zone_anon_alloc_fail; /* cnt of anon alloc fails */
683 + uint64_t zone_pf_throttle; /* cnt of page flt throttles */
684 + uint64_t zone_pf_throttle_usec; /* time of page flt throttles */
601 685
686 + /* Num usecs to throttle page fault when zone is over phys. mem cap */
687 + uint32_t zone_pg_flt_delay;
688 +
602 689 /*
603 690 * Misc. kstats and counters for zone cpu-usage aggregation.
604 691 * The zone_Xtime values are the sum of the micro-state accounting
605 692 * values for all threads that are running or have run in the zone.
606 693 * This is tracked in msacct.c as threads change state.
607 694 * The zone_stime is the sum of the LMS_SYSTEM times.
608 695 * The zone_utime is the sum of the LMS_USER times.
609 696 * The zone_wtime is the sum of the LMS_WAIT_CPU times.
610 697 * As with per-thread micro-state accounting values, these values are
611 698 * not scaled to nanosecs. The scaling is done by the
612 699 * zone_misc_kstat_update function when kstats are requested.
613 700 */
614 701 kmutex_t zone_misc_lock; /* protects misc statistics */
615 702 kstat_t *zone_misc_ksp;
616 703 zone_misc_kstat_t *zone_misc_stats;
617 704 uint64_t zone_stime; /* total system time */
618 705 uint64_t zone_utime; /* total user time */
619 706 uint64_t zone_wtime; /* total time waiting in runq */
620 707 /* fork-fail kstat tracking */
621 708 uint32_t zone_ffcap; /* hit an rctl cap */
622 709 uint32_t zone_ffnoproc; /* get proc/lwp error */
623 710 uint32_t zone_ffnomem; /* as_dup/memory error */
624 711 uint32_t zone_ffmisc; /* misc. other error */
625 712
626 713 uint32_t zone_nested_intp; /* nested interp. kstat */
627 714
628 715 struct loadavg_s zone_loadavg; /* loadavg for this zone */
629 716 uint64_t zone_hp_avenrun[3]; /* high-precision avenrun */
630 717 int zone_avenrun[3]; /* FSCALED avg. run queue len */
631 718
632 719 /*
633 720 * FSS stats updated once per second by fss_decay_usage.
634 721 */
635 722 uint32_t zone_fss_gen; /* FSS generation cntr */
636 723 uint64_t zone_run_ticks; /* tot # of ticks running */
637 724
638 725 /*
639 726 * DTrace-private per-zone state
640 727 */
641 728 int zone_dtrace_getf; /* # of unprivileged getf()s */
642 729
643 730 /*
644 731 * Synchronization primitives used to synchronize between mounts and
645 732 * zone creation/destruction.
646 733 */
647 734 int zone_mounts_in_progress;
648 735 kcondvar_t zone_mount_cv;
649 736 kmutex_t zone_mount_lock;
650 737 } zone_t;
651 738
652 739 /*
653 740 * Special value of zone_psetid to indicate that pools are disabled.
654 741 */
655 742 #define ZONE_PS_INVAL PS_MYID
656 743
657 744
658 745 extern zone_t zone0;
659 746 extern zone_t *global_zone;
660 747 extern uint_t maxzones;
661 748 extern rctl_hndl_t rc_zone_nlwps;
662 749 extern rctl_hndl_t rc_zone_nprocs;
663 750
664 751 extern long zone(int, void *, void *, void *, void *);
665 752 extern void zone_zsd_init(void);
666 753 extern void zone_init(void);
667 754 extern void zone_hold(zone_t *);
668 755 extern void zone_rele(zone_t *);
669 756 extern void zone_init_ref(zone_ref_t *);
670 757 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
671 758 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
672 759 extern void zone_cred_hold(zone_t *);
673 760 extern void zone_cred_rele(zone_t *);
674 761 extern void zone_task_hold(zone_t *);
675 762 extern void zone_task_rele(zone_t *);
676 763 extern zone_t *zone_find_by_id(zoneid_t);
677 764 extern zone_t *zone_find_by_label(const ts_label_t *);
678 765 extern zone_t *zone_find_by_name(char *);
679 766 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
680 767 extern zone_t *zone_find_by_path(const char *);
681 768 extern zoneid_t getzoneid(void);
682 769 extern zone_t *zone_find_by_id_nolock(zoneid_t);
683 770 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
684 771 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
685 772 extern void zone_loadavg_update();
686 773
687 774 /*
688 775 * Zone-specific data (ZSD) APIs
689 776 */
690 777 /*
691 778 * The following is what code should be initializing its zone_key_t to if it
692 779 * calls zone_getspecific() without necessarily knowing that zone_key_create()
693 780 * has been called on the key.
694 781 */
695 782 #define ZONE_KEY_UNINITIALIZED 0
696 783
697 784 typedef uint_t zone_key_t;
698 785
699 786 extern void zone_key_create(zone_key_t *, void *(*)(zoneid_t),
700 787 void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
701 788 extern int zone_key_delete(zone_key_t);
702 789 extern void *zone_getspecific(zone_key_t, zone_t *);
703 790 extern int zone_setspecific(zone_key_t, zone_t *, const void *);
704 791
705 792 /*
706 793 * The definition of a zsd_entry is truly private to zone.c and is only
707 794 * placed here so it can be shared with mdb.
708 795 *
709 796 * State maintained for each zone times each registered key, which tracks
710 797 * the state of the create, shutdown and destroy callbacks.
711 798 *
712 799 * zsd_flags is used to keep track of pending actions to avoid holding locks
713 800 * when calling the create/shutdown/destroy callbacks, since doing so
714 801 * could lead to deadlocks.
715 802 */
716 803 struct zsd_entry {
717 804 zone_key_t zsd_key; /* Key used to lookup value */
718 805 void *zsd_data; /* Caller-managed value */
719 806 /*
720 807 * Callbacks to be executed when a zone is created, shutdown, and
721 808 * destroyed, respectively.
722 809 */
723 810 void *(*zsd_create)(zoneid_t);
724 811 void (*zsd_shutdown)(zoneid_t, void *);
725 812 void (*zsd_destroy)(zoneid_t, void *);
726 813 list_node_t zsd_linkage;
727 814 uint16_t zsd_flags; /* See below */
728 815 kcondvar_t zsd_cv;
729 816 };
730 817
731 818 /*
732 819 * zsd_flags
733 820 */
734 821 #define ZSD_CREATE_NEEDED 0x0001
735 822 #define ZSD_CREATE_INPROGRESS 0x0002
736 823 #define ZSD_CREATE_COMPLETED 0x0004
737 824 #define ZSD_SHUTDOWN_NEEDED 0x0010
738 825 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
739 826 #define ZSD_SHUTDOWN_COMPLETED 0x0040
740 827 #define ZSD_DESTROY_NEEDED 0x0100
741 828 #define ZSD_DESTROY_INPROGRESS 0x0200
742 829 #define ZSD_DESTROY_COMPLETED 0x0400
743 830
744 831 #define ZSD_CREATE_ALL \
745 832 (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
746 833 #define ZSD_SHUTDOWN_ALL \
747 834 (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
748 835 #define ZSD_DESTROY_ALL \
749 836 (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
750 837
751 838 #define ZSD_ALL_INPROGRESS \
752 839 (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
753 840
754 841 /*
755 842 * Macros to help with zone visibility restrictions.
756 843 */
757 844
758 845 /*
759 846 * Is process in the global zone?
760 847 */
761 848 #define INGLOBALZONE(p) \
762 849 ((p)->p_zone == global_zone)
763 850
764 851 /*
765 852 * Can process view objects in given zone?
766 853 */
767 854 #define HASZONEACCESS(p, zoneid) \
768 855 ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
769 856
770 857 /*
771 858 * Convenience macro to see if a resolved path is visible from within a
772 859 * given zone.
773 860 *
774 861 * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
775 862 * two strings must be equal. Since the rootpathlen has a trailing '/',
776 863 * we want to skip everything in the path up to (but not including) the
777 864 * trailing '/'.
778 865 */
779 866 #define ZONE_PATH_VISIBLE(path, zone) \
780 867 (strncmp((path), (zone)->zone_rootpath, \
781 868 (zone)->zone_rootpathlen - 1) == 0)
782 869
783 870 /*
784 871 * Convenience macro to go from the global view of a path to that seen
785 872 * from within said zone. It is the responsibility of the caller to
786 873 * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
787 874 * in fact visible from within the zone.
788 875 */
789 876 #define ZONE_PATH_TRANSLATE(path, zone) \
790 877 (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
791 878 (path) + (zone)->zone_rootpathlen - 2)
792 879
793 880 /*
794 881 * Special processes visible in all zones.
795 882 */
796 883 #define ZONE_SPECIALPID(x) ((x) == 0 || (x) == 1)
797 884
798 885 /*
799 886 * Zone-safe version of thread_create() to be used when the caller wants to
800 887 * create a kernel thread to run within the current zone's context.
801 888 */
802 889 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
803 890 pri_t);
804 891 extern void zthread_exit(void);
805 892
806 893 /*
807 894 * Functions for an external observer to register interest in a zone's status
808 895 * change. Observers will be woken up when the zone status equals the status
809 896 * argument passed in (in the case of zone_status_timedwait, the function may
810 897 * also return because of a timeout; zone_status_wait_sig may return early due
811 898 * to a signal being delivered; zone_status_timedwait_sig may return for any of
812 899 * the above reasons).
813 900 *
814 901 * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
815 902 * cv_wait_sig() respectively.
816 903 */
817 904 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
818 905 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
819 906 extern void zone_status_wait(zone_t *, zone_status_t);
820 907 extern int zone_status_wait_sig(zone_t *, zone_status_t);
821 908
822 909 /*
823 910 * Get the status of the zone (at the time it was called). The state may
824 911 * have progressed by the time it is returned.
825 912 */
826 913 extern zone_status_t zone_status_get(zone_t *);
827 914
828 915 /*
829 916 * Safely get the hostid of the specified zone (defaults to machine's hostid
830 917 * if the specified zone doesn't emulate a hostid). Passing NULL retrieves
831 918 * the global zone's (i.e., physical system's) hostid.
832 919 */
833 920 extern uint32_t zone_get_hostid(zone_t *);
834 921
835 922 /*
836 923 * Get the "kcred" credentials corresponding to the given zone.
837 924 */
838 925 extern struct cred *zone_get_kcred(zoneid_t);
839 926
840 927 /*
841 928 * Get/set the pool the zone is currently bound to.
842 929 */
843 930 extern struct pool *zone_pool_get(zone_t *);
844 931 extern void zone_pool_set(zone_t *, struct pool *);
845 932
846 933 /*
847 934 * Get/set the pset the zone is currently using.
848 935 */
849 936 extern psetid_t zone_pset_get(zone_t *);
850 937 extern void zone_pset_set(zone_t *, psetid_t);
851 938
|
↓ open down ↓ |
240 lines elided |
↑ open up ↑ |
852 939 /*
853 940 * Get the number of cpus/online-cpus visible from the given zone.
854 941 */
855 942 extern int zone_ncpus_get(zone_t *);
856 943 extern int zone_ncpus_online_get(zone_t *);
857 944
858 945 /*
859 946 * Returns true if the named pool/dataset is visible in the current zone.
860 947 */
861 948 extern int zone_dataset_visible(const char *, int *);
949 +extern int zone_dataset_visible_inzone(zone_t *, const char *, int *);
862 950
863 951 /*
864 952 * zone version of kadmin()
865 953 */
866 954 extern int zone_kadmin(int, int, const char *, cred_t *);
867 955 extern void zone_shutdown_global(void);
868 956
869 957 extern void mount_in_progress(zone_t *);
870 958 extern void mount_completed(zone_t *);
871 959
872 960 extern int zone_walk(int (*)(zone_t *, void *), void *);
873 961
874 962 extern rctl_hndl_t rc_zone_locked_mem;
875 963 extern rctl_hndl_t rc_zone_max_swap;
964 +extern rctl_hndl_t rc_zone_phys_mem;
876 965 extern rctl_hndl_t rc_zone_max_lofi;
877 966
878 967 #endif /* _KERNEL */
879 968
880 969 #ifdef __cplusplus
881 970 }
882 971 #endif
883 972
884 973 #endif /* _SYS_ZONE_H */
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX