1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
  23  * Copyright 2015 Joyent, Inc. All rights reserved.
  24  * Copyright 2014 Nexenta Systems, Inc. All rights reserved.
  25  * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
  26  */
  27 
  28 #ifndef _SYS_ZONE_H
  29 #define _SYS_ZONE_H
  30 
  31 #include <sys/types.h>
  32 #include <sys/mutex.h>
  33 #include <sys/param.h>
  34 #include <sys/rctl.h>
  35 #include <sys/ipc_rctl.h>
  36 #include <sys/pset.h>
  37 #include <sys/tsol/label.h>
  38 #include <sys/cred.h>
  39 #include <sys/netstack.h>
  40 #include <sys/uadmin.h>
  41 #include <sys/ksynch.h>
  42 #include <sys/socket_impl.h>
  43 #include <sys/secflags.h>
  44 #include <netinet/in.h>
  45 
  46 #ifdef  __cplusplus
  47 extern "C" {
  48 #endif
  49 
  50 /*
  51  * NOTE
  52  *
  53  * The contents of this file are private to the implementation of
  54  * Solaris and are subject to change at any time without notice.
  55  * Applications and drivers using these interfaces may fail to
  56  * run on future releases.
  57  */
  58 
  59 /* Available both in kernel and for user space */
  60 
  61 /* zone id restrictions and special ids */
  62 #define MAX_ZONEID      9999
  63 #define MIN_USERZONEID  1       /* lowest user-creatable zone ID */
  64 #define MIN_ZONEID      0       /* minimum zone ID on system */
  65 #define GLOBAL_ZONEID   0
  66 #define ZONEID_WIDTH    4       /* for printf */
  67 
  68 /*
  69  * Special zoneid_t token to refer to all zones.
  70  */
  71 #define ALL_ZONES       (-1)
  72 
  73 /* system call subcodes */
  74 #define ZONE_CREATE             0
  75 #define ZONE_DESTROY            1
  76 #define ZONE_GETATTR            2
  77 #define ZONE_ENTER              3
  78 #define ZONE_LIST               4
  79 #define ZONE_SHUTDOWN           5
  80 #define ZONE_LOOKUP             6
  81 #define ZONE_BOOT               7
  82 #define ZONE_VERSION            8
  83 #define ZONE_SETATTR            9
  84 #define ZONE_ADD_DATALINK       10
  85 #define ZONE_DEL_DATALINK       11
  86 #define ZONE_CHECK_DATALINK     12
  87 #define ZONE_LIST_DATALINK      13
  88 
  89 /* zone attributes */
  90 #define ZONE_ATTR_ROOT          1
  91 #define ZONE_ATTR_NAME          2
  92 #define ZONE_ATTR_STATUS        3
  93 #define ZONE_ATTR_PRIVSET       4
  94 #define ZONE_ATTR_UNIQID        5
  95 #define ZONE_ATTR_POOLID        6
  96 #define ZONE_ATTR_INITPID       7
  97 #define ZONE_ATTR_SLBL          8
  98 #define ZONE_ATTR_INITNAME      9
  99 #define ZONE_ATTR_BOOTARGS      10
 100 #define ZONE_ATTR_BRAND         11
 101 #define ZONE_ATTR_PHYS_MCAP     12
 102 #define ZONE_ATTR_SCHED_CLASS   13
 103 #define ZONE_ATTR_FLAGS         14
 104 #define ZONE_ATTR_HOSTID        15
 105 #define ZONE_ATTR_FS_ALLOWED    16
 106 #define ZONE_ATTR_NETWORK       17
 107 #define ZONE_ATTR_INITNORESTART 20
 108 #define ZONE_ATTR_SECFLAGS      21
 109 
 110 /* Start of the brand-specific attribute namespace */
 111 #define ZONE_ATTR_BRAND_ATTRS   32768
 112 
 113 #define ZONE_FS_ALLOWED_MAX     1024
 114 
 115 #define ZONE_EVENT_CHANNEL      "com.sun:zones:status"
 116 #define ZONE_EVENT_STATUS_CLASS "status"
 117 #define ZONE_EVENT_STATUS_SUBCLASS      "change"
 118 
 119 #define ZONE_EVENT_UNINITIALIZED        "uninitialized"
 120 #define ZONE_EVENT_INITIALIZED          "initialized"
 121 #define ZONE_EVENT_READY                "ready"
 122 #define ZONE_EVENT_RUNNING              "running"
 123 #define ZONE_EVENT_SHUTTING_DOWN        "shutting_down"
 124 
 125 #define ZONE_CB_NAME            "zonename"
 126 #define ZONE_CB_NEWSTATE        "newstate"
 127 #define ZONE_CB_OLDSTATE        "oldstate"
 128 #define ZONE_CB_TIMESTAMP       "when"
 129 #define ZONE_CB_ZONEID          "zoneid"
 130 
 131 /*
 132  * Exit values that may be returned by scripts or programs invoked by various
 133  * zone commands.
 134  *
 135  * These are defined as:
 136  *
 137  *      ZONE_SUBPROC_OK
 138  *      ===============
 139  *      The subprocess completed successfully.
 140  *
 141  *      ZONE_SUBPROC_USAGE
 142  *      ==================
 143  *      The subprocess failed with a usage message, or a usage message should
 144  *      be output in its behalf.
 145  *
 146  *      ZONE_SUBPROC_NOTCOMPLETE
 147  *      ========================
 148  *      The subprocess did not complete, but the actions performed by the
 149  *      subprocess require no recovery actions by the user.
 150  *
 151  *      For example, if the subprocess were called by "zoneadm install," the
 152  *      installation of the zone did not succeed but the user need not perform
 153  *      a "zoneadm uninstall" before attempting another install.
 154  *
 155  *      ZONE_SUBPROC_FATAL
 156  *      ==================
 157  *      The subprocess failed in a fatal manner, usually one that will require
 158  *      some type of recovery action by the user.
 159  *
 160  *      For example, if the subprocess were called by "zoneadm install," the
 161  *      installation of the zone did not succeed and the user will need to
 162  *      perform a "zoneadm uninstall" before another install attempt is
 163  *      possible.
 164  *
 165  *      The non-success exit values are large to avoid accidental collision
 166  *      with values used internally by some commands (e.g. "Z_ERR" and
 167  *      "Z_USAGE" as used by zoneadm.)
 168  */
 169 #define ZONE_SUBPROC_OK                 0
 170 #define ZONE_SUBPROC_USAGE              253
 171 #define ZONE_SUBPROC_NOTCOMPLETE        254
 172 #define ZONE_SUBPROC_FATAL              255
 173 
 174 #ifdef _SYSCALL32
 175 typedef struct {
 176         caddr32_t zone_name;
 177         caddr32_t zone_root;
 178         caddr32_t zone_privs;
 179         size32_t zone_privssz;
 180         caddr32_t rctlbuf;
 181         size32_t rctlbufsz;
 182         caddr32_t extended_error;
 183         caddr32_t zfsbuf;
 184         size32_t  zfsbufsz;
 185         int match;                      /* match level */
 186         uint32_t doi;                   /* DOI for label */
 187         caddr32_t label;                /* label associated with zone */
 188         int flags;
 189 } zone_def32;
 190 #endif
 191 typedef struct {
 192         const char *zone_name;
 193         const char *zone_root;
 194         const struct priv_set *zone_privs;
 195         size_t zone_privssz;
 196         const char *rctlbuf;
 197         size_t rctlbufsz;
 198         int *extended_error;
 199         const char *zfsbuf;
 200         size_t zfsbufsz;
 201         int match;                      /* match level */
 202         uint32_t doi;                   /* DOI for label */
 203         const bslabel_t *label;         /* label associated with zone */
 204         int flags;
 205 } zone_def;
 206 
 207 /* extended error information */
 208 #define ZE_UNKNOWN      0       /* No extended error info */
 209 #define ZE_CHROOTED     1       /* tried to zone_create from chroot */
 210 #define ZE_AREMOUNTS    2       /* there are mounts within the zone */
 211 #define ZE_LABELINUSE   3       /* label is already in use by some other zone */
 212 
 213 /*
 214  * zone_status values
 215  *
 216  * You must modify zone_status_names in mdb(1M)'s genunix module
 217  * (genunix/zone.c) when you modify this enum.
 218  */
 219 typedef enum {
 220         ZONE_IS_UNINITIALIZED = 0,
 221         ZONE_IS_INITIALIZED,
 222         ZONE_IS_READY,
 223         ZONE_IS_BOOTING,
 224         ZONE_IS_RUNNING,
 225         ZONE_IS_SHUTTING_DOWN,
 226         ZONE_IS_EMPTY,
 227         ZONE_IS_DOWN,
 228         ZONE_IS_DYING,
 229         ZONE_IS_DEAD
 230 } zone_status_t;
 231 #define ZONE_MIN_STATE          ZONE_IS_UNINITIALIZED
 232 #define ZONE_MAX_STATE          ZONE_IS_DEAD
 233 
 234 /*
 235  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
 236  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
 237  */
 238 typedef enum zone_cmd {
 239         Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
 240         Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
 241 } zone_cmd_t;
 242 
 243 /*
 244  * The structure of a request to zoneadmd.
 245  */
 246 typedef struct zone_cmd_arg {
 247         uint64_t        uniqid;         /* unique "generation number" */
 248         zone_cmd_t      cmd;            /* requested action */
 249         uint32_t        _pad;           /* need consistent 32/64 bit alignmt */
 250         char locale[MAXPATHLEN];        /* locale in which to render messages */
 251         char bootbuf[BOOTARGS_MAX];     /* arguments passed to zone_boot() */
 252 } zone_cmd_arg_t;
 253 
 254 /*
 255  * Structure of zoneadmd's response to a request.  A NULL return value means
 256  * the caller should attempt to restart zoneadmd and retry.
 257  */
 258 typedef struct zone_cmd_rval {
 259         int rval;                       /* return value of request */
 260         char errbuf[1]; /* variable-sized buffer containing error messages */
 261 } zone_cmd_rval_t;
 262 
 263 /*
 264  * The zone support infrastructure uses the zone name as a component
 265  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
 266  * in length, so ZONENAME_MAX is limited by that.
 267  */
 268 #define ZONENAME_MAX            64
 269 
 270 #define GLOBAL_ZONENAME         "global"
 271 
 272 /*
 273  * Extended Regular expression (see regex(5)) which matches all valid zone
 274  * names.
 275  */
 276 #define ZONENAME_REGEXP         "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
 277 
 278 /*
 279  * Where the zones support infrastructure places temporary files.
 280  */
 281 #define ZONES_TMPDIR            "/var/run/zones"
 282 
 283 /*
 284  * The path to the door used by clients to communicate with zoneadmd.
 285  */
 286 #define ZONE_DOOR_PATH          ZONES_TMPDIR "/%s.zoneadmd_door"
 287 
 288 
 289 /* zone_flags */
 290 /*
 291  * Threads that read or write the following flag must hold zone_lock.
 292  */
 293 #define ZF_REFCOUNTS_LOGGED     0x1     /* a thread logged the zone's refs */
 294 
 295 /*
 296  * The following threads are set when the zone is created and never changed.
 297  * Threads that test for these flags don't have to hold zone_lock.
 298  */
 299 #define ZF_HASHED_LABEL         0x2     /* zone has a unique label */
 300 #define ZF_IS_SCRATCH           0x4     /* scratch zone */
 301 #define ZF_NET_EXCL             0x8     /* Zone has an exclusive IP stack */
 302 
 303 
 304 /* zone_create flags */
 305 #define ZCF_NET_EXCL            0x1     /* Create a zone with exclusive IP */
 306 
 307 /* zone network properties */
 308 #define ZONE_NETWORK_ADDRESS    1
 309 #define ZONE_NETWORK_DEFROUTER  2
 310 
 311 #define ZONE_NET_ADDRNAME       "address"
 312 #define ZONE_NET_RTRNAME        "route"
 313 
 314 typedef struct zone_net_data {
 315         int zn_type;
 316         int zn_len;
 317         datalink_id_t zn_linkid;
 318         uint8_t zn_val[1];
 319 } zone_net_data_t;
 320 
 321 
 322 #ifdef _KERNEL
 323 
 324 /*
 325  * We need to protect the definition of 'list_t' from userland applications and
 326  * libraries which may be defining ther own versions.
 327  */
 328 #include <sys/list.h>
 329 #include <sys/loadavg.h>
 330 
 331 #define GLOBAL_ZONEUNIQID       0       /* uniqid of the global zone */
 332 
 333 struct pool;
 334 struct brand;
 335 
 336 /*
 337  * Each of these constants identifies a kernel subsystem that acquires and
 338  * releases zone references.  Each subsystem that invokes
 339  * zone_hold_ref() and zone_rele_ref() should specify the
 340  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
 341  * help users and developers quickly identify subsystems that stall zone
 342  * shutdowns indefinitely.
 343  *
 344  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
 345  * when you modify this enumeration.
 346  */
 347 typedef enum zone_ref_subsys {
 348         ZONE_REF_NFS,                   /* NFS */
 349         ZONE_REF_NFSV4,                 /* NFSv4 */
 350         ZONE_REF_SMBFS,                 /* SMBFS */
 351         ZONE_REF_MNTFS,                 /* MNTFS */
 352         ZONE_REF_LOFI,                  /* LOFI devices */
 353         ZONE_REF_VFS,                   /* VFS infrastructure */
 354         ZONE_REF_IPC,                   /* IPC infrastructure */
 355         ZONE_REF_NUM_SUBSYS             /* This must be the last entry. */
 356 } zone_ref_subsys_t;
 357 
 358 /*
 359  * zone_ref represents a general-purpose references to a zone.  Each zone's
 360  * references are linked into the zone's zone_t::zone_ref_list.  This allows
 361  * debuggers to walk zones' references.
 362  */
 363 typedef struct zone_ref {
 364         struct zone     *zref_zone; /* the zone to which the reference refers */
 365         list_node_t     zref_linkage; /* linkage for zone_t::zone_ref_list */
 366 } zone_ref_t;
 367 
 368 /*
 369  * Structure to record list of ZFS datasets exported to a zone.
 370  */
 371 typedef struct zone_dataset {
 372         char            *zd_dataset;
 373         list_node_t     zd_linkage;
 374 } zone_dataset_t;
 375 
 376 /*
 377  * structure for zone kstats
 378  */
 379 typedef struct zone_kstat {
 380         kstat_named_t zk_zonename;
 381         kstat_named_t zk_usage;
 382         kstat_named_t zk_value;
 383 } zone_kstat_t;
 384 
 385 struct cpucap;
 386 
 387 typedef struct {
 388         kstat_named_t   zm_zonename;
 389         kstat_named_t   zm_pgpgin;
 390         kstat_named_t   zm_anonpgin;
 391         kstat_named_t   zm_execpgin;
 392         kstat_named_t   zm_fspgin;
 393         kstat_named_t   zm_anon_alloc_fail;
 394 } zone_mcap_kstat_t;
 395 
 396 typedef struct {
 397         kstat_named_t   zm_zonename;    /* full name, kstat truncates name */
 398         kstat_named_t   zm_utime;
 399         kstat_named_t   zm_stime;
 400         kstat_named_t   zm_wtime;
 401         kstat_named_t   zm_avenrun1;
 402         kstat_named_t   zm_avenrun5;
 403         kstat_named_t   zm_avenrun15;
 404         kstat_named_t   zm_ffcap;
 405         kstat_named_t   zm_ffnoproc;
 406         kstat_named_t   zm_ffnomem;
 407         kstat_named_t   zm_ffmisc;
 408         kstat_named_t   zm_nested_intp;
 409         kstat_named_t   zm_init_pid;
 410         kstat_named_t   zm_boot_time;
 411 } zone_misc_kstat_t;
 412 
 413 typedef struct zone {
 414         /*
 415          * zone_name is never modified once set.
 416          */
 417         char            *zone_name;     /* zone's configuration name */
 418         /*
 419          * zone_nodename and zone_domain are never freed once allocated.
 420          */
 421         char            *zone_nodename; /* utsname.nodename equivalent */
 422         char            *zone_domain;   /* srpc_domain equivalent */
 423         /*
 424          * zone_hostid is used for per-zone hostid emulation.
 425          * Currently it isn't modified after it's set (so no locks protect
 426          * accesses), but that might have to change when we allow
 427          * administrators to change running zones' properties.
 428          *
 429          * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
 430          * that zone_get_hostid() will function correctly.
 431          */
 432         uint32_t        zone_hostid;    /* zone's hostid, HW_INVALID_HOSTID */
 433                                         /* if not emulated */
 434         /*
 435          * 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
 443          *      zone_pfexecd
 444          */
 445         kmutex_t        zone_lock;
 446         /*
 447          * zone_linkage is the zone's linkage into the active or
 448          * death-row list.  The field is protected by zonehash_lock.
 449          */
 450         list_node_t     zone_linkage;
 451         zoneid_t        zone_id;        /* ID of zone */
 452         uint_t          zone_ref;       /* count of zone_hold()s on zone */
 453         uint_t          zone_cred_ref;  /* count of zone_hold_cred()s on zone */
 454         /*
 455          * Fixed-sized array of subsystem-specific reference counts
 456          * The sum of all of the counts must be less than or equal to zone_ref.
 457          * The array is indexed by the counts' subsystems' zone_ref_subsys_t
 458          * constants.
 459          */
 460         uint_t          zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
 461         list_t          zone_ref_list;  /* list of zone_ref_t structs */
 462         /*
 463          * zone_rootvp and zone_rootpath can never be modified once set.
 464          */
 465         struct vnode    *zone_rootvp;   /* zone's root vnode */
 466         char            *zone_rootpath; /* Path to zone's root + '/' */
 467         ushort_t        zone_flags;     /* misc flags */
 468         zone_status_t   zone_status;    /* protected by zone_status_lock */
 469         uint_t          zone_ntasks;    /* number of tasks executing in zone */
 470         kmutex_t        zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
 471                                         /* counters in projects and tasks */
 472                                         /* that are within the zone */
 473         rctl_qty_t      zone_nlwps;     /* number of lwps in zone */
 474         rctl_qty_t      zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
 475         rctl_qty_t      zone_shmmax;    /* System V shared memory usage */
 476         ipc_rqty_t      zone_ipc;       /* System V IPC id resource usage */
 477 
 478         uint_t          zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
 479         uint32_t        zone_shares;    /* FSS shares allocated to zone */
 480         rctl_set_t      *zone_rctls;    /* zone-wide (zone.*) rctls */
 481         kmutex_t        zone_mem_lock;  /* protects zone_locked_mem and */
 482                                         /* kpd_locked_mem for all */
 483                                         /* projects in zone. */
 484                                         /* Also protects zone_max_swap */
 485                                         /* grab after p_lock, before rcs_lock */
 486         rctl_qty_t      zone_locked_mem;        /* bytes of locked memory in */
 487                                                 /* zone */
 488         rctl_qty_t      zone_locked_mem_ctl;    /* Current locked memory */
 489                                                 /* limit.  Protected by */
 490                                                 /* zone_rctls->rcs_lock */
 491         rctl_qty_t      zone_max_swap; /* bytes of swap reserved by zone */
 492         rctl_qty_t      zone_max_swap_ctl;      /* current swap limit. */
 493                                                 /* Protected by */
 494                                                 /* zone_rctls->rcs_lock */
 495         kmutex_t        zone_rctl_lock; /* protects zone_max_lofi */
 496         rctl_qty_t      zone_max_lofi; /* lofi devs for zone */
 497         rctl_qty_t      zone_max_lofi_ctl;      /* current lofi limit. */
 498                                                 /* Protected by */
 499                                                 /* zone_rctls->rcs_lock */
 500         list_t          zone_zsd;       /* list of Zone-Specific Data values */
 501         kcondvar_t      zone_cv;        /* used to signal state changes */
 502         struct proc     *zone_zsched;   /* Dummy kernel "zsched" process */
 503         pid_t           zone_proc_initpid; /* pid of "init" for this zone */
 504         char            *zone_initname; /* fs path to 'init' */
 505         int             zone_boot_err;  /* for zone_boot() if boot fails */
 506         char            *zone_bootargs; /* arguments passed via zone_boot() */
 507         uint64_t        zone_phys_mcap; /* physical memory cap */
 508         /*
 509          * zone_kthreads is protected by zone_status_lock.
 510          */
 511         kthread_t       *zone_kthreads; /* kernel threads in zone */
 512         struct priv_set *zone_privset;  /* limit set for zone */
 513         /*
 514          * zone_vfslist is protected by vfs_list_lock().
 515          */
 516         struct vfs      *zone_vfslist;  /* list of FS's mounted in zone */
 517         uint64_t        zone_uniqid;    /* unique zone generation number */
 518         struct cred     *zone_kcred;    /* kcred-like, zone-limited cred */
 519         /*
 520          * zone_pool is protected by pool_lock().
 521          */
 522         struct pool     *zone_pool;     /* pool the zone is bound to */
 523         hrtime_t        zone_pool_mod;  /* last pool bind modification time */
 524         /* zone_psetid is protected by cpu_lock */
 525         psetid_t        zone_psetid;    /* pset the zone is bound to */
 526 
 527         time_t          zone_boot_time; /* Similar to boot_time */
 528 
 529         /*
 530          * The following two can be read without holding any locks.  They are
 531          * updated under cpu_lock.
 532          */
 533         int             zone_ncpus;  /* zone's idea of ncpus */
 534         int             zone_ncpus_online; /* zone's idea of ncpus_online */
 535         /*
 536          * List of ZFS datasets exported to this zone.
 537          */
 538         list_t          zone_datasets;  /* list of datasets */
 539 
 540         ts_label_t      *zone_slabel;   /* zone sensitivity label */
 541         int             zone_match;     /* require label match for packets */
 542         tsol_mlp_list_t zone_mlps;      /* MLPs on zone-private addresses */
 543 
 544         boolean_t       zone_restart_init;      /* Restart init if it dies? */
 545         struct brand    *zone_brand;            /* zone's brand */
 546         void            *zone_brand_data;       /* store brand specific data */
 547         id_t            zone_defaultcid;        /* dflt scheduling class id */
 548         kstat_t         *zone_swapresv_kstat;
 549         kstat_t         *zone_lockedmem_kstat;
 550         /*
 551          * zone_dl_list is protected by zone_lock
 552          */
 553         list_t          zone_dl_list;
 554         netstack_t      *zone_netstack;
 555         struct cpucap   *zone_cpucap;   /* CPU caps data */
 556         /*
 557          * Solaris Auditing per-zone audit context
 558          */
 559         struct au_kcontext      *zone_audit_kctxt;
 560         /*
 561          * For private use by mntfs.
 562          */
 563         struct mntelem  *zone_mntfs_db;
 564         krwlock_t       zone_mntfs_db_lock;
 565 
 566         struct klpd_reg         *zone_pfexecd;
 567 
 568         char            *zone_fs_allowed;
 569         rctl_qty_t      zone_nprocs;    /* number of processes in the zone */
 570         rctl_qty_t      zone_nprocs_ctl;        /* current limit protected by */
 571                                                 /* zone_rctls->rcs_lock */
 572         kstat_t         *zone_nprocs_kstat;
 573 
 574         kmutex_t        zone_mcap_lock; /* protects mcap statistics */
 575         kstat_t         *zone_mcap_ksp;
 576         zone_mcap_kstat_t *zone_mcap_stats;
 577         uint64_t        zone_pgpgin;            /* pages paged in */
 578         uint64_t        zone_anonpgin;          /* anon pages paged in */
 579         uint64_t        zone_execpgin;          /* exec pages paged in */
 580         uint64_t        zone_fspgin;            /* fs pages paged in */
 581         uint64_t        zone_anon_alloc_fail;   /* cnt of anon alloc fails */
 582 
 583         psecflags_t     zone_secflags; /* default zone security-flags */
 584 
 585         /*
 586          * 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         kmutex_t        zone_misc_lock;         /* protects misc statistics */
 598         kstat_t         *zone_misc_ksp;
 599         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 */
 603         /* fork-fail kstat tracking */
 604         uint32_t        zone_ffcap;             /* hit an rctl cap */
 605         uint32_t        zone_ffnoproc;          /* get proc/lwp error */
 606         uint32_t        zone_ffnomem;           /* as_dup/memory error */
 607         uint32_t        zone_ffmisc;            /* misc. other error */
 608 
 609         uint32_t        zone_nested_intp;       /* nested interp. kstat */
 610 
 611         struct loadavg_s zone_loadavg;          /* loadavg for this zone */
 612         uint64_t        zone_hp_avenrun[3];     /* high-precision avenrun */
 613         int             zone_avenrun[3];        /* FSCALED avg. run queue len */
 614 
 615         /*
 616          * FSS stats updated once per second by fss_decay_usage.
 617          */
 618         uint32_t        zone_fss_gen;           /* FSS generation cntr */
 619         uint64_t        zone_run_ticks;         /* tot # of ticks running */
 620 
 621         /*
 622          * DTrace-private per-zone state
 623          */
 624         int             zone_dtrace_getf;       /* # of unprivileged getf()s */
 625 
 626         /*
 627          * Synchronization primitives used to synchronize between mounts and
 628          * zone creation/destruction.
 629          */
 630         int             zone_mounts_in_progress;
 631         kcondvar_t      zone_mount_cv;
 632         kmutex_t        zone_mount_lock;
 633 } zone_t;
 634 
 635 /*
 636  * Special value of zone_psetid to indicate that pools are disabled.
 637  */
 638 #define ZONE_PS_INVAL   PS_MYID
 639 
 640 
 641 extern zone_t zone0;
 642 extern zone_t *global_zone;
 643 extern uint_t maxzones;
 644 extern rctl_hndl_t rc_zone_nlwps;
 645 extern rctl_hndl_t rc_zone_nprocs;
 646 
 647 extern long zone(int, void *, void *, void *, void *);
 648 extern void zone_zsd_init(void);
 649 extern void zone_init(void);
 650 extern void zone_hold(zone_t *);
 651 extern void zone_rele(zone_t *);
 652 extern void zone_init_ref(zone_ref_t *);
 653 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
 654 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
 655 extern void zone_cred_hold(zone_t *);
 656 extern void zone_cred_rele(zone_t *);
 657 extern void zone_task_hold(zone_t *);
 658 extern void zone_task_rele(zone_t *);
 659 extern zone_t *zone_find_by_id(zoneid_t);
 660 extern zone_t *zone_find_by_label(const ts_label_t *);
 661 extern zone_t *zone_find_by_name(char *);
 662 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
 663 extern zone_t *zone_find_by_path(const char *);
 664 extern zoneid_t getzoneid(void);
 665 extern zone_t *zone_find_by_id_nolock(zoneid_t);
 666 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
 667 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
 668 extern void zone_loadavg_update();
 669 
 670 /*
 671  * Zone-specific data (ZSD) APIs
 672  */
 673 /*
 674  * The following is what code should be initializing its zone_key_t to if it
 675  * calls zone_getspecific() without necessarily knowing that zone_key_create()
 676  * has been called on the key.
 677  */
 678 #define ZONE_KEY_UNINITIALIZED  0
 679 
 680 typedef uint_t zone_key_t;
 681 
 682 extern void     zone_key_create(zone_key_t *, void *(*)(zoneid_t),
 683     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
 684 extern int      zone_key_delete(zone_key_t);
 685 extern void     *zone_getspecific(zone_key_t, zone_t *);
 686 extern int      zone_setspecific(zone_key_t, zone_t *, const void *);
 687 
 688 /*
 689  * The definition of a zsd_entry is truly private to zone.c and is only
 690  * placed here so it can be shared with mdb.
 691  *
 692  * State maintained for each zone times each registered key, which tracks
 693  * the state of the create, shutdown and destroy callbacks.
 694  *
 695  * zsd_flags is used to keep track of pending actions to avoid holding locks
 696  * when calling the create/shutdown/destroy callbacks, since doing so
 697  * could lead to deadlocks.
 698  */
 699 struct zsd_entry {
 700         zone_key_t              zsd_key;        /* Key used to lookup value */
 701         void                    *zsd_data;      /* Caller-managed value */
 702         /*
 703          * Callbacks to be executed when a zone is created, shutdown, and
 704          * destroyed, respectively.
 705          */
 706         void                    *(*zsd_create)(zoneid_t);
 707         void                    (*zsd_shutdown)(zoneid_t, void *);
 708         void                    (*zsd_destroy)(zoneid_t, void *);
 709         list_node_t             zsd_linkage;
 710         uint16_t                zsd_flags;      /* See below */
 711         kcondvar_t              zsd_cv;
 712 };
 713 
 714 /*
 715  * zsd_flags
 716  */
 717 #define ZSD_CREATE_NEEDED       0x0001
 718 #define ZSD_CREATE_INPROGRESS   0x0002
 719 #define ZSD_CREATE_COMPLETED    0x0004
 720 #define ZSD_SHUTDOWN_NEEDED     0x0010
 721 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
 722 #define ZSD_SHUTDOWN_COMPLETED  0x0040
 723 #define ZSD_DESTROY_NEEDED      0x0100
 724 #define ZSD_DESTROY_INPROGRESS  0x0200
 725 #define ZSD_DESTROY_COMPLETED   0x0400
 726 
 727 #define ZSD_CREATE_ALL  \
 728         (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
 729 #define ZSD_SHUTDOWN_ALL        \
 730         (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
 731 #define ZSD_DESTROY_ALL \
 732         (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
 733 
 734 #define ZSD_ALL_INPROGRESS \
 735         (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
 736 
 737 /*
 738  * Macros to help with zone visibility restrictions.
 739  */
 740 
 741 /*
 742  * Is process in the global zone?
 743  */
 744 #define INGLOBALZONE(p) \
 745         ((p)->p_zone == global_zone)
 746 
 747 /*
 748  * Can process view objects in given zone?
 749  */
 750 #define HASZONEACCESS(p, zoneid) \
 751         ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
 752 
 753 /*
 754  * Convenience macro to see if a resolved path is visible from within a
 755  * given zone.
 756  *
 757  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
 758  * two strings must be equal.  Since the rootpathlen has a trailing '/',
 759  * we want to skip everything in the path up to (but not including) the
 760  * trailing '/'.
 761  */
 762 #define ZONE_PATH_VISIBLE(path, zone) \
 763         (strncmp((path), (zone)->zone_rootpath,              \
 764             (zone)->zone_rootpathlen - 1) == 0)
 765 
 766 /*
 767  * Convenience macro to go from the global view of a path to that seen
 768  * from within said zone.  It is the responsibility of the caller to
 769  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
 770  * in fact visible from within the zone.
 771  */
 772 #define ZONE_PATH_TRANSLATE(path, zone) \
 773         (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
 774         (path) + (zone)->zone_rootpathlen - 2)
 775 
 776 /*
 777  * Special processes visible in all zones.
 778  */
 779 #define ZONE_SPECIALPID(x)       ((x) == 0 || (x) == 1)
 780 
 781 /*
 782  * Zone-safe version of thread_create() to be used when the caller wants to
 783  * create a kernel thread to run within the current zone's context.
 784  */
 785 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
 786     pri_t);
 787 extern void zthread_exit(void);
 788 
 789 /*
 790  * Functions for an external observer to register interest in a zone's status
 791  * change.  Observers will be woken up when the zone status equals the status
 792  * argument passed in (in the case of zone_status_timedwait, the function may
 793  * also return because of a timeout; zone_status_wait_sig may return early due
 794  * to a signal being delivered; zone_status_timedwait_sig may return for any of
 795  * the above reasons).
 796  *
 797  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
 798  * cv_wait_sig() respectively.
 799  */
 800 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
 801 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
 802 extern void zone_status_wait(zone_t *, zone_status_t);
 803 extern int zone_status_wait_sig(zone_t *, zone_status_t);
 804 
 805 /*
 806  * Get the status  of the zone (at the time it was called).  The state may
 807  * have progressed by the time it is returned.
 808  */
 809 extern zone_status_t zone_status_get(zone_t *);
 810 
 811 /*
 812  * Safely get the hostid of the specified zone (defaults to machine's hostid
 813  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
 814  * the global zone's (i.e., physical system's) hostid.
 815  */
 816 extern uint32_t zone_get_hostid(zone_t *);
 817 
 818 /*
 819  * Get the "kcred" credentials corresponding to the given zone.
 820  */
 821 extern struct cred *zone_get_kcred(zoneid_t);
 822 
 823 /*
 824  * Get/set the pool the zone is currently bound to.
 825  */
 826 extern struct pool *zone_pool_get(zone_t *);
 827 extern void zone_pool_set(zone_t *, struct pool *);
 828 
 829 /*
 830  * Get/set the pset the zone is currently using.
 831  */
 832 extern psetid_t zone_pset_get(zone_t *);
 833 extern void zone_pset_set(zone_t *, psetid_t);
 834 
 835 /*
 836  * Get the number of cpus/online-cpus visible from the given zone.
 837  */
 838 extern int zone_ncpus_get(zone_t *);
 839 extern int zone_ncpus_online_get(zone_t *);
 840 
 841 /*
 842  * Returns true if the named pool/dataset is visible in the current zone.
 843  */
 844 extern int zone_dataset_visible(const char *, int *);
 845 
 846 /*
 847  * zone version of kadmin()
 848  */
 849 extern int zone_kadmin(int, int, const char *, cred_t *);
 850 extern void zone_shutdown_global(void);
 851 
 852 extern void mount_in_progress(zone_t *);
 853 extern void mount_completed(zone_t *);
 854 
 855 extern int zone_walk(int (*)(zone_t *, void *), void *);
 856 
 857 extern rctl_hndl_t rc_zone_locked_mem;
 858 extern rctl_hndl_t rc_zone_max_swap;
 859 extern rctl_hndl_t rc_zone_max_lofi;
 860 
 861 #endif  /* _KERNEL */
 862 
 863 #ifdef  __cplusplus
 864 }
 865 #endif
 866 
 867 #endif  /* _SYS_ZONE_H */