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 /*
  23  * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
  24  * Copyright 2014 Igor Kozhukhov <ikozhukhov@gmail.com>.
  25  * Copyright 2019 Nexenta Systems, Inc. All rights reserved.
  26  * Copyright 2020 Joyent, Inc.
  27  */
  28 
  29 #ifndef _SYS_ZONE_H
  30 #define _SYS_ZONE_H
  31 
  32 #include <sys/types.h>
  33 #include <sys/param.h>
  34 #include <sys/tsol/label.h>
  35 #include <sys/uadmin.h>
  36 #include <netinet/in.h>
  37 
  38 #ifdef _KERNEL
  39 /*
  40  * Many includes are kernel-only to reduce namespace pollution of
  41  * userland applications.
  42  */
  43 #include <sys/mutex.h>
  44 #include <sys/rctl.h>
  45 #include <sys/ipc_rctl.h>
  46 #include <sys/pset.h>
  47 #include <sys/cred.h>
  48 #include <sys/netstack.h>
  49 #include <sys/ksynch.h>
  50 #include <sys/socket_impl.h>
  51 #include <sys/secflags.h>
  52 #include <sys/cpu_uarray.h>
  53 #include <sys/nvpair.h>
  54 #include <sys/list.h>
  55 #include <sys/loadavg.h>
  56 #include <sys/vnode.h>
  57 #endif  /* _KERNEL */
  58 
  59 #ifdef  __cplusplus
  60 extern "C" {
  61 #endif
  62 
  63 /*
  64  * NOTE
  65  *
  66  * The contents of this file are private to the implementation of
  67  * illumos and are subject to change at any time without notice.
  68  * Applications and drivers using these interfaces may fail to
  69  * run on future releases.
  70  */
  71 
  72 /* Available both in kernel and for user space */
  73 
  74 /*
  75  * zone id restrictions and special ids.
  76  * See 'maxzones' for run-time zone limit.
  77  *
  78  * The current 8k value for MAX_ZONES was originally derived from the virtual
  79  * interface limit in IP when "shared-stack" was the only supported networking
  80  * for zones. The virtual interface limit is the number of addresses allowed
  81  * on an interface (see MAX_ADDRS_PER_IF). Even with exclusive stacks, an 8k
  82  * zone limit is still a reasonable choice at this time, given other limits
  83  * within the kernel. Since we only support 8192 zones (which includes GZ),
  84  * there is no point in allowing MAX_ZONEID > 8k.
  85  */
  86 #define MAX_ZONES       8192
  87 #define MAX_ZONEID      (MAX_ZONES - 1)
  88 #define MIN_USERZONEID  1       /* lowest user-creatable zone ID */
  89 #define MIN_ZONEID      0       /* minimum zone ID on system */
  90 #define GLOBAL_ZONEID   0
  91 #define ZONEID_WIDTH    4       /* for printf */
  92 
  93 /*
  94  * Special zoneid_t token to refer to all zones.
  95  */
  96 #define ALL_ZONES       (-1)
  97 
  98 /* system call subcodes */
  99 #define ZONE_CREATE             0
 100 #define ZONE_DESTROY            1
 101 #define ZONE_GETATTR            2
 102 #define ZONE_ENTER              3
 103 #define ZONE_LIST               4
 104 #define ZONE_SHUTDOWN           5
 105 #define ZONE_LOOKUP             6
 106 #define ZONE_BOOT               7
 107 #define ZONE_VERSION            8
 108 #define ZONE_SETATTR            9
 109 #define ZONE_ADD_DATALINK       10
 110 #define ZONE_DEL_DATALINK       11
 111 #define ZONE_CHECK_DATALINK     12
 112 #define ZONE_LIST_DATALINK      13
 113 
 114 /* zone attributes */
 115 #define ZONE_ATTR_ROOT          1
 116 #define ZONE_ATTR_NAME          2
 117 #define ZONE_ATTR_STATUS        3
 118 #define ZONE_ATTR_PRIVSET       4
 119 #define ZONE_ATTR_UNIQID        5
 120 #define ZONE_ATTR_POOLID        6
 121 #define ZONE_ATTR_INITPID       7
 122 #define ZONE_ATTR_SLBL          8
 123 #define ZONE_ATTR_INITNAME      9
 124 #define ZONE_ATTR_BOOTARGS      10
 125 #define ZONE_ATTR_BRAND         11
 126 #define ZONE_ATTR_SCHED_CLASS   12
 127 #define ZONE_ATTR_FLAGS         13
 128 #define ZONE_ATTR_HOSTID        14
 129 #define ZONE_ATTR_FS_ALLOWED    15
 130 #define ZONE_ATTR_NETWORK       16
 131 #define ZONE_ATTR_DID           17
 132 #define ZONE_ATTR_INITNORESTART 18
 133 #define ZONE_ATTR_APP_SVC_CT    19
 134 #define ZONE_ATTR_SCHED_FIXEDHI 20
 135 #define ZONE_ATTR_SECFLAGS      21
 136 #define ZONE_ATTR_INITRESTART0  22
 137 #define ZONE_ATTR_INITREBOOT    23
 138 
 139 /* Start of the brand-specific attribute namespace */
 140 #define ZONE_ATTR_BRAND_ATTRS   32768
 141 
 142 #define ZONE_FS_ALLOWED_MAX     1024
 143 
 144 #define ZONE_EVENT_CHANNEL      "com.sun:zones:status"
 145 #define ZONE_EVENT_STATUS_CLASS "status"
 146 #define ZONE_EVENT_STATUS_SUBCLASS      "change"
 147 
 148 #define ZONE_EVENT_UNINITIALIZED        "uninitialized"
 149 #define ZONE_EVENT_INITIALIZED          "initialized"
 150 #define ZONE_EVENT_READY                "ready"
 151 #define ZONE_EVENT_RUNNING              "running"
 152 #define ZONE_EVENT_SHUTTING_DOWN        "shutting_down"
 153 #define ZONE_EVENT_FREE                 "free"
 154 
 155 #define ZONE_CB_NAME            "zonename"
 156 #define ZONE_CB_NEWSTATE        "newstate"
 157 #define ZONE_CB_OLDSTATE        "oldstate"
 158 #define ZONE_CB_RESTARTS        "restarts"
 159 #define ZONE_CB_TIMESTAMP       "when"
 160 #define ZONE_CB_ZONEID          "zoneid"
 161 
 162 #define ZONE_EVENT_INIT_CLASS           "init"
 163 #define ZONE_EVENT_INIT_RESTART_SC      "restart"
 164 
 165 /*
 166  * Exit values that may be returned by scripts or programs invoked by various
 167  * zone commands.
 168  *
 169  * These are defined as:
 170  *
 171  *      ZONE_SUBPROC_OK
 172  *      ===============
 173  *      The subprocess completed successfully.
 174  *
 175  *      ZONE_SUBPROC_USAGE
 176  *      ==================
 177  *      The subprocess failed with a usage message, or a usage message should
 178  *      be output in its behalf.
 179  *
 180  *      ZONE_SUBPROC_NOTCOMPLETE
 181  *      ========================
 182  *      The subprocess did not complete, but the actions performed by the
 183  *      subprocess require no recovery actions by the user.
 184  *
 185  *      For example, if the subprocess were called by "zoneadm install," the
 186  *      installation of the zone did not succeed but the user need not perform
 187  *      a "zoneadm uninstall" before attempting another install.
 188  *
 189  *      ZONE_SUBPROC_FATAL
 190  *      ==================
 191  *      The subprocess failed in a fatal manner, usually one that will require
 192  *      some type of recovery action by the user.
 193  *
 194  *      For example, if the subprocess were called by "zoneadm install," the
 195  *      installation of the zone did not succeed and the user will need to
 196  *      perform a "zoneadm uninstall" before another install attempt is
 197  *      possible.
 198  *
 199  *      The non-success exit values are large to avoid accidental collision
 200  *      with values used internally by some commands (e.g. "Z_ERR" and
 201  *      "Z_USAGE" as used by zoneadm.)
 202  */
 203 #define ZONE_SUBPROC_OK                 0
 204 #define ZONE_SUBPROC_USAGE              253
 205 #define ZONE_SUBPROC_NOTCOMPLETE        254
 206 #define ZONE_SUBPROC_FATAL              255
 207 
 208 #ifdef _SYSCALL32
 209 typedef struct {
 210         caddr32_t zone_name;
 211         caddr32_t zone_root;
 212         caddr32_t zone_privs;
 213         size32_t zone_privssz;
 214         caddr32_t rctlbuf;
 215         size32_t rctlbufsz;
 216         caddr32_t extended_error;
 217         caddr32_t zfsbuf;
 218         size32_t  zfsbufsz;
 219         int match;                      /* match level */
 220         uint32_t doi;                   /* DOI for label */
 221         caddr32_t label;                /* label associated with zone */
 222         int flags;
 223         zoneid_t zoneid;                /* requested zoneid */
 224 } zone_def32;
 225 #endif
 226 typedef struct {
 227         const char *zone_name;
 228         const char *zone_root;
 229         const struct priv_set *zone_privs;
 230         size_t zone_privssz;
 231         const char *rctlbuf;
 232         size_t rctlbufsz;
 233         int *extended_error;
 234         const char *zfsbuf;
 235         size_t zfsbufsz;
 236         int match;                      /* match level */
 237         uint32_t doi;                   /* DOI for label */
 238         const bslabel_t *label;         /* label associated with zone */
 239         int flags;
 240         zoneid_t zoneid;                /* requested zoneid */
 241 } zone_def;
 242 
 243 /* extended error information */
 244 #define ZE_UNKNOWN      0       /* No extended error info */
 245 #define ZE_CHROOTED     1       /* tried to zone_create from chroot */
 246 #define ZE_AREMOUNTS    2       /* there are mounts within the zone */
 247 #define ZE_LABELINUSE   3       /* label is already in use by some other zone */
 248 
 249 /*
 250  * zone_status values
 251  *
 252  * You must modify zone_status_names in mdb(1M)'s genunix module
 253  * (genunix/zone.c) when you modify this enum.
 254  */
 255 typedef enum {
 256         ZONE_IS_UNINITIALIZED = 0,
 257         ZONE_IS_INITIALIZED,
 258         ZONE_IS_READY,
 259         ZONE_IS_BOOTING,
 260         ZONE_IS_RUNNING,
 261         ZONE_IS_SHUTTING_DOWN,
 262         ZONE_IS_EMPTY,
 263         ZONE_IS_DOWN,
 264         ZONE_IS_DYING,
 265         ZONE_IS_DEAD,
 266         ZONE_IS_FREE            /* transient state for zone sysevent */
 267 } zone_status_t;
 268 #define ZONE_MIN_STATE          ZONE_IS_UNINITIALIZED
 269 #define ZONE_MAX_STATE          ZONE_IS_DEAD
 270 
 271 /*
 272  * Valid commands which may be issued by zoneadm to zoneadmd.  The kernel also
 273  * communicates with zoneadmd, but only uses Z_REBOOT and Z_HALT.
 274  */
 275 typedef enum zone_cmd {
 276         Z_READY, Z_BOOT, Z_FORCEBOOT, Z_REBOOT, Z_HALT, Z_NOTE_UNINSTALLING,
 277         Z_MOUNT, Z_FORCEMOUNT, Z_UNMOUNT, Z_SHUTDOWN
 278 } zone_cmd_t;
 279 
 280 /*
 281  * The structure of a request to zoneadmd.
 282  */
 283 typedef struct zone_cmd_arg {
 284         uint64_t        uniqid;         /* unique "generation number" */
 285         zone_cmd_t      cmd;            /* requested action */
 286         int             status;         /* init status on shutdown */
 287         uint32_t        debug;          /* enable brand hook debug */
 288         char locale[MAXPATHLEN];        /* locale in which to render messages */
 289         char bootbuf[BOOTARGS_MAX];     /* arguments passed to zone_boot() */
 290         /* Needed for 32/64 zoneadm -> zoneadmd door arg size check. */
 291         int             pad;
 292 } zone_cmd_arg_t;
 293 
 294 /*
 295  * Structure of zoneadmd's response to a request.  A NULL return value means
 296  * the caller should attempt to restart zoneadmd and retry.
 297  */
 298 typedef struct zone_cmd_rval {
 299         int rval;                       /* return value of request */
 300         char errbuf[1]; /* variable-sized buffer containing error messages */
 301 } zone_cmd_rval_t;
 302 
 303 /*
 304  * The zone support infrastructure uses the zone name as a component
 305  * of unix domain (AF_UNIX) sockets, which are limited to 108 characters
 306  * in length, so ZONENAME_MAX is limited by that.
 307  */
 308 #define ZONENAME_MAX            64
 309 
 310 #define GLOBAL_ZONENAME         "global"
 311 
 312 /*
 313  * Extended Regular expression (see regex(5)) which matches all valid zone
 314  * names.
 315  */
 316 #define ZONENAME_REGEXP         "[a-zA-Z0-9][-_.a-zA-Z0-9]{0,62}"
 317 
 318 /*
 319  * Where the zones support infrastructure places temporary files.
 320  */
 321 #define ZONES_TMPDIR            "/var/run/zones"
 322 
 323 /*
 324  * The path to the door used by clients to communicate with zoneadmd.
 325  */
 326 #define ZONE_DOOR_PATH          ZONES_TMPDIR "/%s.zoneadmd_door"
 327 
 328 
 329 /* zone_flags */
 330 /*
 331  * Threads that read or write the following flag must hold zone_lock.
 332  */
 333 #define ZF_REFCOUNTS_LOGGED     0x1     /* a thread logged the zone's refs */
 334 
 335 /*
 336  * The following threads are set when the zone is created and never changed.
 337  * Threads that test for these flags don't have to hold zone_lock.
 338  */
 339 #define ZF_HASHED_LABEL         0x2     /* zone has a unique label */
 340 #define ZF_IS_SCRATCH           0x4     /* scratch zone */
 341 #define ZF_NET_EXCL             0x8     /* Zone has an exclusive IP stack */
 342 
 343 
 344 /* zone_create flags */
 345 #define ZCF_NET_EXCL            0x1     /* Create a zone with exclusive IP */
 346 
 347 /* zone network properties */
 348 #define ZONE_NETWORK_ADDRESS    1
 349 #define ZONE_NETWORK_DEFROUTER  2
 350 
 351 #define ZONE_NET_ADDRNAME       "address"
 352 #define ZONE_NET_RTRNAME        "route"
 353 
 354 typedef struct zone_net_data {
 355         int zn_type;
 356         int zn_len;
 357         datalink_id_t zn_linkid;
 358         uint8_t zn_val[1];
 359 } zone_net_data_t;
 360 
 361 
 362 #ifdef _KERNEL
 363 
 364 #define GLOBAL_ZONEUNIQID       0       /* uniqid of the global zone */
 365 
 366 /*
 367  * Indexes into ->zone_ustate array, summing the micro state of all threads in a
 368  * particular zone.
 369  */
 370 #define ZONE_USTATE_STIME (0)
 371 #define ZONE_USTATE_UTIME (1)
 372 #define ZONE_USTATE_WTIME (2)
 373 #define ZONE_USTATE_MAX (3)
 374 
 375 struct pool;
 376 struct brand;
 377 
 378 /*
 379  * Each of these constants identifies a kernel subsystem that acquires and
 380  * releases zone references.  Each subsystem that invokes
 381  * zone_hold_ref() and zone_rele_ref() should specify the
 382  * zone_ref_subsys_t constant associated with the subsystem.  Tracked holds
 383  * help users and developers quickly identify subsystems that stall zone
 384  * shutdowns indefinitely.
 385  *
 386  * NOTE: You must modify zone_ref_subsys_names in usr/src/uts/common/os/zone.c
 387  * when you modify this enumeration.
 388  */
 389 typedef enum zone_ref_subsys {
 390         ZONE_REF_NFS,                   /* NFS */
 391         ZONE_REF_NFSV4,                 /* NFSv4 */
 392         ZONE_REF_SMBFS,                 /* SMBFS */
 393         ZONE_REF_MNTFS,                 /* MNTFS */
 394         ZONE_REF_LOFI,                  /* LOFI devices */
 395         ZONE_REF_VFS,                   /* VFS infrastructure */
 396         ZONE_REF_IPC,                   /* IPC infrastructure */
 397         ZONE_REF_NUM_SUBSYS             /* This must be the last entry. */
 398 } zone_ref_subsys_t;
 399 
 400 /*
 401  * zone_ref represents a general-purpose references to a zone.  Each zone's
 402  * references are linked into the zone's zone_t::zone_ref_list.  This allows
 403  * debuggers to walk zones' references.
 404  */
 405 typedef struct zone_ref {
 406         struct zone     *zref_zone; /* the zone to which the reference refers */
 407         list_node_t     zref_linkage; /* linkage for zone_t::zone_ref_list */
 408 } zone_ref_t;
 409 
 410 /*
 411  * Structure to record list of ZFS datasets exported to a zone.
 412  */
 413 typedef struct zone_dataset {
 414         char            *zd_dataset;
 415         list_node_t     zd_linkage;
 416 } zone_dataset_t;
 417 
 418 /*
 419  * structure for rctl zone kstats
 420  */
 421 typedef struct zone_kstat {
 422         kstat_named_t zk_zonename;
 423         kstat_named_t zk_usage;
 424         kstat_named_t zk_value;
 425 } zone_kstat_t;
 426 
 427 struct cpucap;
 428 
 429 typedef struct {
 430         hrtime_t        cycle_start;
 431         uint_t          cycle_cnt;
 432         hrtime_t        zone_avg_cnt;
 433 } sys_zio_cntr_t;
 434 
 435 typedef struct {
 436         kstat_named_t   zv_zonename;
 437         kstat_named_t   zv_nread;
 438         kstat_named_t   zv_reads;
 439         kstat_named_t   zv_rtime;
 440         kstat_named_t   zv_rlentime;
 441         kstat_named_t   zv_rcnt;
 442         kstat_named_t   zv_nwritten;
 443         kstat_named_t   zv_writes;
 444         kstat_named_t   zv_wtime;
 445         kstat_named_t   zv_wlentime;
 446         kstat_named_t   zv_wcnt;
 447         kstat_named_t   zv_10ms_ops;
 448         kstat_named_t   zv_100ms_ops;
 449         kstat_named_t   zv_1s_ops;
 450         kstat_named_t   zv_10s_ops;
 451         kstat_named_t   zv_delay_cnt;
 452         kstat_named_t   zv_delay_time;
 453 } zone_vfs_kstat_t;
 454 
 455 typedef struct {
 456         kstat_named_t   zz_zonename;
 457         kstat_named_t   zz_nread;
 458         kstat_named_t   zz_reads;
 459         kstat_named_t   zz_rtime;
 460         kstat_named_t   zz_rlentime;
 461         kstat_named_t   zz_nwritten;
 462         kstat_named_t   zz_writes;
 463         kstat_named_t   zz_waittime;
 464 } zone_zfs_kstat_t;
 465 
 466 typedef struct {
 467         kstat_named_t   zm_zonename;
 468         kstat_named_t   zm_rss;
 469         kstat_named_t   zm_phys_cap;
 470         kstat_named_t   zm_swap;
 471         kstat_named_t   zm_swap_cap;
 472         kstat_named_t   zm_nover;
 473         kstat_named_t   zm_pagedout;
 474         kstat_named_t   zm_pgpgin;
 475         kstat_named_t   zm_anonpgin;
 476         kstat_named_t   zm_execpgin;
 477         kstat_named_t   zm_fspgin;
 478         kstat_named_t   zm_anon_alloc_fail;
 479         kstat_named_t   zm_pf_throttle;
 480         kstat_named_t   zm_pf_throttle_usec;
 481 } zone_mcap_kstat_t;
 482 
 483 typedef struct {
 484         kstat_named_t   zm_zonename;    /* full name, kstat truncates name */
 485         kstat_named_t   zm_utime;
 486         kstat_named_t   zm_stime;
 487         kstat_named_t   zm_wtime;
 488         kstat_named_t   zm_avenrun1;
 489         kstat_named_t   zm_avenrun5;
 490         kstat_named_t   zm_avenrun15;
 491         kstat_named_t   zm_ffcap;
 492         kstat_named_t   zm_ffnoproc;
 493         kstat_named_t   zm_ffnomem;
 494         kstat_named_t   zm_ffmisc;
 495         kstat_named_t   zm_mfseglim;
 496         kstat_named_t   zm_nested_intp;
 497         kstat_named_t   zm_init_pid;
 498         kstat_named_t   zm_init_restarts;
 499         kstat_named_t   zm_boot_time;
 500 } zone_misc_kstat_t;
 501 
 502 typedef struct zone {
 503         /*
 504          * zone_name is never modified once set.
 505          */
 506         char            *zone_name;     /* zone's configuration name */
 507         /*
 508          * zone_nodename and zone_domain are never freed once allocated.
 509          */
 510         char            *zone_nodename; /* utsname.nodename equivalent */
 511         char            *zone_domain;   /* srpc_domain equivalent */
 512         /*
 513          * zone_hostid is used for per-zone hostid emulation.
 514          * Currently it isn't modified after it's set (so no locks protect
 515          * accesses), but that might have to change when we allow
 516          * administrators to change running zones' properties.
 517          *
 518          * The global zone's zone_hostid must always be HW_INVALID_HOSTID so
 519          * that zone_get_hostid() will function correctly.
 520          */
 521         uint32_t        zone_hostid;    /* zone's hostid, HW_INVALID_HOSTID */
 522                                         /* if not emulated */
 523         /*
 524          * zone_lock protects the following fields of a zone_t:
 525          *      zone_ref
 526          *      zone_cred_ref
 527          *      zone_subsys_ref
 528          *      zone_ref_list
 529          *      zone_ntasks
 530          *      zone_flags
 531          *      zone_zsd
 532          *      zone_pfexecd
 533          */
 534         kmutex_t        zone_lock;
 535         /*
 536          * zone_linkage is the zone's linkage into the active or
 537          * death-row list.  The field is protected by zonehash_lock.
 538          */
 539         list_node_t     zone_linkage;
 540         zoneid_t        zone_id;        /* ID of zone */
 541         zoneid_t        zone_did;       /* persistent debug ID of zone */
 542         uint_t          zone_ref;       /* count of zone_hold()s on zone */
 543         uint_t          zone_cred_ref;  /* count of zone_hold_cred()s on zone */
 544         /*
 545          * Fixed-sized array of subsystem-specific reference counts
 546          * The sum of all of the counts must be less than or equal to zone_ref.
 547          * The array is indexed by the counts' subsystems' zone_ref_subsys_t
 548          * constants.
 549          */
 550         uint_t          zone_subsys_ref[ZONE_REF_NUM_SUBSYS];
 551         list_t          zone_ref_list;  /* list of zone_ref_t structs */
 552         /*
 553          * zone_rootvp and zone_rootpath can never be modified once set.
 554          */
 555         struct vnode    *zone_rootvp;   /* zone's root vnode */
 556         char            *zone_rootpath; /* Path to zone's root + '/' */
 557         ushort_t        zone_flags;     /* misc flags */
 558         zone_status_t   zone_status;    /* protected by zone_status_lock */
 559         uint_t          zone_ntasks;    /* number of tasks executing in zone */
 560         kmutex_t        zone_nlwps_lock; /* protects zone_nlwps, and *_nlwps */
 561                                         /* counters in projects and tasks */
 562                                         /* that are within the zone */
 563         rctl_qty_t      zone_nlwps;     /* number of lwps in zone */
 564         rctl_qty_t      zone_nlwps_ctl; /* protected by zone_rctls->rcs_lock */
 565         rctl_qty_t      zone_shmmax;    /* System V shared memory usage */
 566         ipc_rqty_t      zone_ipc;       /* System V IPC id resource usage */
 567 
 568         uint_t          zone_rootpathlen; /* strlen(zone_rootpath) + 1 */
 569         uint32_t        zone_shares;    /* FSS shares allocated to zone */
 570         rctl_set_t      *zone_rctls;    /* zone-wide (zone.*) rctls */
 571         kmutex_t        zone_mem_lock;  /* protects zone_locked_mem and */
 572                                         /* kpd_locked_mem for all */
 573                                         /* projects in zone. */
 574                                         /* Also protects zone_max_swap */
 575                                         /* grab after p_lock, before rcs_lock */
 576         rctl_qty_t      zone_locked_mem;        /* bytes of locked memory in */
 577                                                 /* zone */
 578         rctl_qty_t      zone_locked_mem_ctl;    /* Current locked memory */
 579                                                 /* limit.  Protected by */
 580                                                 /* zone_rctls->rcs_lock */
 581         rctl_qty_t      zone_max_swap; /* bytes of swap reserved by zone */
 582         rctl_qty_t      zone_max_swap_ctl;      /* current swap limit. */
 583                                                 /* Protected by */
 584                                                 /* zone_rctls->rcs_lock */
 585         kmutex_t        zone_rctl_lock; /* protects zone_max_lofi */
 586         rctl_qty_t      zone_max_lofi; /* lofi devs for zone */
 587         rctl_qty_t      zone_max_lofi_ctl;      /* current lofi limit. */
 588                                                 /* Protected by */
 589                                                 /* zone_rctls->rcs_lock */
 590         list_t          zone_zsd;       /* list of Zone-Specific Data values */
 591         kcondvar_t      zone_cv;        /* used to signal state changes */
 592         struct proc     *zone_zsched;   /* Dummy kernel "zsched" process */
 593         pid_t           zone_proc_initpid; /* pid of "init" for this zone */
 594         uint_t          zone_proc_init_restarts; /* times init restarted */
 595         char            *zone_initname;         /* fs path to 'init' */
 596         int             zone_init_status;       /* init's exit status */
 597         int             zone_boot_err;  /* for zone_boot() if boot fails */
 598         char            *zone_bootargs; /* arguments passed via zone_boot() */
 599         /*
 600          * zone_kthreads is protected by zone_status_lock.
 601          */
 602         kthread_t       *zone_kthreads; /* kernel threads in zone */
 603         struct priv_set *zone_privset;  /* limit set for zone */
 604         /*
 605          * zone_vfslist is protected by vfs_list_lock().
 606          */
 607         struct vfs      *zone_vfslist;  /* list of FS's mounted in zone */
 608         uint64_t        zone_uniqid;    /* unique zone generation number */
 609         struct cred     *zone_kcred;    /* kcred-like, zone-limited cred */
 610         /*
 611          * zone_pool is protected by pool_lock().
 612          */
 613         struct pool     *zone_pool;     /* pool the zone is bound to */
 614         hrtime_t        zone_pool_mod;  /* last pool bind modification time */
 615         /* zone_psetid is protected by cpu_lock */
 616         psetid_t        zone_psetid;    /* pset the zone is bound to */
 617 
 618         time_t          zone_boot_time; /* Similar to boot_time */
 619 
 620         /*
 621          * The following two can be read without holding any locks.  They are
 622          * updated under cpu_lock.
 623          */
 624         int             zone_ncpus;  /* zone's idea of ncpus */
 625         int             zone_ncpus_online; /* zone's idea of ncpus_online */
 626         /*
 627          * List of ZFS datasets exported to this zone.
 628          */
 629         list_t          zone_datasets;  /* list of datasets */
 630 
 631         ts_label_t      *zone_slabel;   /* zone sensitivity label */
 632         int             zone_match;     /* require label match for packets */
 633         tsol_mlp_list_t zone_mlps;      /* MLPs on zone-private addresses */
 634 
 635         boolean_t       zone_restart_init;      /* Restart init if it dies? */
 636         boolean_t       zone_reboot_on_init_exit; /* Reboot if init dies? */
 637         boolean_t       zone_restart_init_0;    /* Restart only if it exits 0 */
 638         boolean_t       zone_setup_app_contract; /* setup contract? */
 639         struct brand    *zone_brand;            /* zone's brand */
 640         void            *zone_brand_data;       /* store brand specific data */
 641         id_t            zone_defaultcid;        /* dflt scheduling class id */
 642         boolean_t       zone_fixed_hipri;       /* fixed sched. hi prio */
 643         kstat_t         *zone_swapresv_kstat;
 644         kstat_t         *zone_lockedmem_kstat;
 645         /*
 646          * zone_dl_list is protected by zone_lock
 647          */
 648         list_t          zone_dl_list;
 649         netstack_t      *zone_netstack;
 650         struct cpucap   *zone_cpucap;   /* CPU caps data */
 651 
 652         /*
 653          * kstats and counters for VFS ops and bytes.
 654          */
 655         kmutex_t        zone_vfs_lock;          /* protects VFS statistics */
 656         kstat_t         *zone_vfs_ksp;
 657         kstat_io_t      zone_vfs_rwstats;
 658         zone_vfs_kstat_t *zone_vfs_stats;
 659 
 660         /*
 661          * kstats for ZFS I/O ops and bytes.
 662          */
 663         kmutex_t        zone_zfs_lock;          /* protects ZFS statistics */
 664         kstat_t         *zone_zfs_ksp;
 665         zone_zfs_kstat_t *zone_zfs_stats;
 666 
 667         /*
 668          * illumos Auditing per-zone audit context
 669          */
 670         struct au_kcontext      *zone_audit_kctxt;
 671         /*
 672          * For private use by mntfs.
 673          */
 674         struct mntelem  *zone_mntfs_db;
 675         krwlock_t       zone_mntfs_db_lock;
 676 
 677         struct klpd_reg         *zone_pfexecd;
 678 
 679         char            *zone_fs_allowed;
 680         rctl_qty_t      zone_nprocs;    /* number of processes in the zone */
 681         rctl_qty_t      zone_nprocs_ctl;        /* current limit protected by */
 682                                                 /* zone_rctls->rcs_lock */
 683         kstat_t         *zone_nprocs_kstat;
 684 
 685         /*
 686          * kstats and counters for physical memory capping.
 687          */
 688         kstat_t         *zone_physmem_kstat;
 689         kmutex_t        zone_mcap_lock;         /* protects mcap statistics */
 690         kstat_t         *zone_mcap_ksp;
 691         zone_mcap_kstat_t *zone_mcap_stats;
 692         uint64_t        zone_pgpgin;            /* pages paged in */
 693         uint64_t        zone_anonpgin;          /* anon pages paged in */
 694         uint64_t        zone_execpgin;          /* exec pages paged in */
 695         uint64_t        zone_fspgin;            /* fs pages paged in */
 696         uint64_t        zone_anon_alloc_fail;   /* cnt of anon alloc fails */
 697 
 698         psecflags_t     zone_secflags; /* default zone security-flags */
 699 
 700         /*
 701          * Misc. kstats and counters for zone cpu-usage aggregation.
 702          */
 703         kmutex_t        zone_misc_lock;         /* protects misc statistics */
 704         kstat_t         *zone_misc_ksp;
 705         zone_misc_kstat_t *zone_misc_stats;
 706         /* Accumulated microstate for all threads in this zone. */
 707         cpu_uarray_t    *zone_ustate;
 708         /* fork-fail kstat tracking */
 709         uint32_t        zone_ffcap;             /* hit an rctl cap */
 710         uint32_t        zone_ffnoproc;          /* get proc/lwp error */
 711         uint32_t        zone_ffnomem;           /* as_dup/memory error */
 712         uint32_t        zone_ffmisc;            /* misc. other error */
 713 
 714         uint32_t        zone_mfseglim;          /* map failure (# segs limit) */
 715 
 716         uint32_t        zone_nested_intp;       /* nested interp. kstat */
 717 
 718         struct loadavg_s zone_loadavg;          /* loadavg for this zone */
 719         uint64_t        zone_hp_avenrun[3];     /* high-precision avenrun */
 720         int             zone_avenrun[3];        /* FSCALED avg. run queue len */
 721 
 722         /*
 723          * FSS stats updated once per second by fss_decay_usage.
 724          */
 725         uint32_t        zone_fss_gen;           /* FSS generation cntr */
 726         uint64_t        zone_run_ticks;         /* tot # of ticks running */
 727 
 728         /*
 729          * DTrace-private per-zone state
 730          */
 731         int             zone_dtrace_getf;       /* # of unprivileged getf()s */
 732 
 733         /*
 734          * Synchronization primitives used to synchronize between mounts and
 735          * zone creation/destruction.
 736          */
 737         int             zone_mounts_in_progress;
 738         kcondvar_t      zone_mount_cv;
 739         kmutex_t        zone_mount_lock;
 740 } zone_t;
 741 
 742 /*
 743  * Data and counters used for ZFS fair-share disk IO.
 744  */
 745 typedef struct zone_zfs_io {
 746         uint16_t        zpers_zfs_io_pri;       /* ZFS IO priority - 16k max */
 747         uint_t          zpers_zfs_queued[2];    /* sync I/O enqueued count */
 748         sys_zio_cntr_t  zpers_rd_ops;           /* Counters for ZFS reads, */
 749         sys_zio_cntr_t  zpers_wr_ops;           /* writes, and */
 750         sys_zio_cntr_t  zpers_lwr_ops;          /* logical writes. */
 751         kstat_io_t      zpers_zfs_rwstats;
 752         uint64_t        zpers_io_util;          /* IO utilization metric */
 753         uint64_t        zpers_zfs_rd_waittime;
 754         uint8_t         zpers_io_delay;         /* IO delay on logical r/w */
 755         uint8_t         zpers_zfs_weight;       /* used to prevent starvation */
 756         uint8_t         zpers_io_util_above_avg; /* IO util percent > avg. */
 757 } zone_zfs_io_t;
 758 
 759 /*
 760  * "Persistent" zone data which can be accessed idependently of the zone_t.
 761  */
 762 typedef struct zone_persist {
 763         kmutex_t        zpers_zfs_lock; /* Protects zpers_zfsp references */
 764         zone_zfs_io_t   *zpers_zfsp;    /* ZFS fair-share IO data */
 765         uint8_t         zpers_over;     /* currently over cap */
 766         uint32_t        zpers_pg_cnt;   /* current RSS in pages */
 767         uint32_t        zpers_pg_limit; /* current RRS limit in pages */
 768         uint32_t        zpers_nover;    /* # of times over phys. cap */
 769 #ifndef DEBUG
 770         uint64_t        zpers_pg_out;   /* # pages flushed */
 771 #else
 772         /*
 773          * To conserve memory, some detailed kstats are only kept for DEBUG
 774          * builds.
 775          */
 776         uint64_t        zpers_zfs_rd_waittime;
 777 
 778         uint64_t        zpers_pg_anon;          /* # clean anon pages flushed */
 779         uint64_t        zpers_pg_anondirty;     /* # dirty anon pages flushed */
 780         uint64_t        zpers_pg_fs;            /* # clean fs pages flushed */
 781         uint64_t        zpers_pg_fsdirty;       /* # dirty fs pages flushed */
 782 #endif
 783 } zone_persist_t;
 784 
 785 typedef enum zone_pageout_op {
 786         ZPO_DIRTY, ZPO_FS, ZPO_ANON, ZPO_ANONDIRTY
 787 } zone_pageout_op_t;
 788 
 789 /*
 790  * Special value of zone_psetid to indicate that pools are disabled.
 791  */
 792 #define ZONE_PS_INVAL   PS_MYID
 793 
 794 extern zone_t zone0;
 795 extern zone_t *global_zone;
 796 extern uint_t maxzones;
 797 extern rctl_hndl_t rc_zone_nlwps;
 798 extern rctl_hndl_t rc_zone_nprocs;
 799 
 800 extern long zone(int, void *, void *, void *, void *);
 801 extern void zone_zsd_init(void);
 802 extern void zone_init(void);
 803 extern void zone_hold(zone_t *);
 804 extern void zone_rele(zone_t *);
 805 extern void zone_init_ref(zone_ref_t *);
 806 extern void zone_hold_ref(zone_t *, zone_ref_t *, zone_ref_subsys_t);
 807 extern void zone_rele_ref(zone_ref_t *, zone_ref_subsys_t);
 808 extern void zone_cred_hold(zone_t *);
 809 extern void zone_cred_rele(zone_t *);
 810 extern void zone_task_hold(zone_t *);
 811 extern void zone_task_rele(zone_t *);
 812 extern zone_t *zone_find_by_id(zoneid_t);
 813 extern zone_t *zone_find_by_label(const ts_label_t *);
 814 extern zone_t *zone_find_by_name(char *);
 815 extern zone_t *zone_find_by_any_path(const char *, boolean_t);
 816 extern zone_t *zone_find_by_path(const char *);
 817 extern zoneid_t getzoneid(void);
 818 extern zoneid_t getzonedid(void);
 819 extern zone_t *zone_find_by_id_nolock(zoneid_t);
 820 extern int zone_datalink_walk(zoneid_t, int (*)(datalink_id_t, void *), void *);
 821 extern int zone_check_datalink(zoneid_t *, datalink_id_t);
 822 extern void zone_loadavg_update();
 823 
 824 /*
 825  * Zone-specific data (ZSD) APIs
 826  */
 827 /*
 828  * The following is what code should be initializing its zone_key_t to if it
 829  * calls zone_getspecific() without necessarily knowing that zone_key_create()
 830  * has been called on the key.
 831  */
 832 #define ZONE_KEY_UNINITIALIZED  0
 833 
 834 typedef uint_t zone_key_t;
 835 
 836 extern void     zone_key_create(zone_key_t *, void *(*)(zoneid_t),
 837     void (*)(zoneid_t, void *), void (*)(zoneid_t, void *));
 838 extern int      zone_key_delete(zone_key_t);
 839 extern void     *zone_getspecific(zone_key_t, zone_t *);
 840 extern int      zone_setspecific(zone_key_t, zone_t *, const void *);
 841 
 842 /*
 843  * The definition of a zsd_entry is truly private to zone.c and is only
 844  * placed here so it can be shared with mdb.
 845  *
 846  * State maintained for each zone times each registered key, which tracks
 847  * the state of the create, shutdown and destroy callbacks.
 848  *
 849  * zsd_flags is used to keep track of pending actions to avoid holding locks
 850  * when calling the create/shutdown/destroy callbacks, since doing so
 851  * could lead to deadlocks.
 852  */
 853 struct zsd_entry {
 854         zone_key_t              zsd_key;        /* Key used to lookup value */
 855         void                    *zsd_data;      /* Caller-managed value */
 856         /*
 857          * Callbacks to be executed when a zone is created, shutdown, and
 858          * destroyed, respectively.
 859          */
 860         void                    *(*zsd_create)(zoneid_t);
 861         void                    (*zsd_shutdown)(zoneid_t, void *);
 862         void                    (*zsd_destroy)(zoneid_t, void *);
 863         list_node_t             zsd_linkage;
 864         uint16_t                zsd_flags;      /* See below */
 865         kcondvar_t              zsd_cv;
 866 };
 867 
 868 /*
 869  * zsd_flags
 870  */
 871 #define ZSD_CREATE_NEEDED       0x0001
 872 #define ZSD_CREATE_INPROGRESS   0x0002
 873 #define ZSD_CREATE_COMPLETED    0x0004
 874 #define ZSD_SHUTDOWN_NEEDED     0x0010
 875 #define ZSD_SHUTDOWN_INPROGRESS 0x0020
 876 #define ZSD_SHUTDOWN_COMPLETED  0x0040
 877 #define ZSD_DESTROY_NEEDED      0x0100
 878 #define ZSD_DESTROY_INPROGRESS  0x0200
 879 #define ZSD_DESTROY_COMPLETED   0x0400
 880 
 881 #define ZSD_CREATE_ALL  \
 882         (ZSD_CREATE_NEEDED|ZSD_CREATE_INPROGRESS|ZSD_CREATE_COMPLETED)
 883 #define ZSD_SHUTDOWN_ALL        \
 884         (ZSD_SHUTDOWN_NEEDED|ZSD_SHUTDOWN_INPROGRESS|ZSD_SHUTDOWN_COMPLETED)
 885 #define ZSD_DESTROY_ALL \
 886         (ZSD_DESTROY_NEEDED|ZSD_DESTROY_INPROGRESS|ZSD_DESTROY_COMPLETED)
 887 
 888 #define ZSD_ALL_INPROGRESS \
 889         (ZSD_CREATE_INPROGRESS|ZSD_SHUTDOWN_INPROGRESS|ZSD_DESTROY_INPROGRESS)
 890 
 891 /*
 892  * Macros to help with zone visibility restrictions.
 893  */
 894 
 895 /*
 896  * Is process in the global zone?
 897  */
 898 #define INGLOBALZONE(p) \
 899         ((p)->p_zone == global_zone)
 900 
 901 /*
 902  * Can process view objects in given zone?
 903  */
 904 #define HASZONEACCESS(p, zoneid) \
 905         ((p)->p_zone->zone_id == (zoneid) || INGLOBALZONE(p))
 906 
 907 /*
 908  * Convenience macro to see if a resolved path is visible from within a
 909  * given zone.
 910  *
 911  * The basic idea is that the first (zone_rootpathlen - 1) bytes of the
 912  * two strings must be equal.  Since the rootpathlen has a trailing '/',
 913  * we want to skip everything in the path up to (but not including) the
 914  * trailing '/'.
 915  */
 916 #define ZONE_PATH_VISIBLE(path, zone) \
 917         (strncmp((path), (zone)->zone_rootpath,              \
 918             (zone)->zone_rootpathlen - 1) == 0)
 919 
 920 /*
 921  * Convenience macro to go from the global view of a path to that seen
 922  * from within said zone.  It is the responsibility of the caller to
 923  * ensure that the path is a resolved one (ie, no '..'s or '.'s), and is
 924  * in fact visible from within the zone.
 925  */
 926 #define ZONE_PATH_TRANSLATE(path, zone) \
 927         (ASSERT(ZONE_PATH_VISIBLE(path, zone)), \
 928         (path) + (zone)->zone_rootpathlen - 2)
 929 
 930 /*
 931  * Special processes visible in all zones.
 932  */
 933 #define ZONE_SPECIALPID(x)       ((x) == 0 || (x) == 1)
 934 
 935 /*
 936  * A root vnode of the current zone.
 937  *
 938  * NOTE: It may be necessary (initialization time for file sharing where an
 939  * NGZ loads a file-sharing kernel module that does zsd initialization) to NOT
 940  * use this macro. One should ASSERT() that curzone == active ZSD (an
 941  * ASSERTion that's not always true at ZSD initialization time) during regular
 942  * use of this macro.
 943  */
 944 #define ZONE_ROOTVP()   (curzone->zone_rootvp)
 945 
 946 /*
 947  * Since a zone's root isn't necessarily an actual filesystem boundary
 948  * (i.e. VROOT may not be set on zone->zone_rootvp) we need to not assume it.
 949  * This macro helps in checking if a vnode is the current zone's rootvp.
 950  * NOTE:  Using the VN_ prefix, even though it's defined here in zone.h.
 951  * NOTE2: See above warning about ZONE_ROOTVP().
 952  */
 953 #define VN_IS_CURZONEROOT(vp)   (VN_CMP(vp, ZONE_ROOTVP()))
 954 
 955 /*
 956  * Zone-safe version of thread_create() to be used when the caller wants to
 957  * create a kernel thread to run within the current zone's context.
 958  */
 959 extern kthread_t *zthread_create(caddr_t, size_t, void (*)(), void *, size_t,
 960     pri_t);
 961 extern void zthread_exit(void);
 962 
 963 /*
 964  * Functions for an external observer to register interest in a zone's status
 965  * change.  Observers will be woken up when the zone status equals the status
 966  * argument passed in (in the case of zone_status_timedwait, the function may
 967  * also return because of a timeout; zone_status_wait_sig may return early due
 968  * to a signal being delivered; zone_status_timedwait_sig may return for any of
 969  * the above reasons).
 970  *
 971  * Otherwise these behave identically to cv_timedwait(), cv_wait(), and
 972  * cv_wait_sig() respectively.
 973  */
 974 extern clock_t zone_status_timedwait(zone_t *, clock_t, zone_status_t);
 975 extern clock_t zone_status_timedwait_sig(zone_t *, clock_t, zone_status_t);
 976 extern void zone_status_wait(zone_t *, zone_status_t);
 977 extern int zone_status_wait_sig(zone_t *, zone_status_t);
 978 
 979 /*
 980  * Get the status  of the zone (at the time it was called).  The state may
 981  * have progressed by the time it is returned.
 982  */
 983 extern zone_status_t zone_status_get(zone_t *);
 984 
 985 /*
 986  * Safely get the hostid of the specified zone (defaults to machine's hostid
 987  * if the specified zone doesn't emulate a hostid).  Passing NULL retrieves
 988  * the global zone's (i.e., physical system's) hostid.
 989  */
 990 extern uint32_t zone_get_hostid(zone_t *);
 991 
 992 /*
 993  * Get the "kcred" credentials corresponding to the given zone.
 994  */
 995 extern struct cred *zone_get_kcred(zoneid_t);
 996 
 997 /*
 998  * Get/set the pool the zone is currently bound to.
 999  */
1000 extern struct pool *zone_pool_get(zone_t *);
1001 extern void zone_pool_set(zone_t *, struct pool *);
1002 
1003 /*
1004  * Get/set the pset the zone is currently using.
1005  */
1006 extern psetid_t zone_pset_get(zone_t *);
1007 extern void zone_pset_set(zone_t *, psetid_t);
1008 
1009 /*
1010  * Get the number of cpus/online-cpus visible from the given zone.
1011  */
1012 extern int zone_ncpus_get(zone_t *);
1013 extern int zone_ncpus_online_get(zone_t *);
1014 
1015 /*
1016  * Returns true if the named pool/dataset is visible in the current zone.
1017  */
1018 extern int zone_dataset_visible(const char *, int *);
1019 extern int zone_dataset_visible_inzone(zone_t *, const char *, int *);
1020 
1021 /*
1022  * zone version of kadmin()
1023  */
1024 extern int zone_kadmin(int, int, const char *, cred_t *);
1025 extern void zone_shutdown_global(void);
1026 
1027 extern void mount_in_progress(zone_t *);
1028 extern void mount_completed(zone_t *);
1029 
1030 extern int zone_walk(int (*)(zone_t *, void *), void *);
1031 
1032 struct page;
1033 extern void zone_add_page(struct page *);
1034 extern void zone_rm_page(struct page *);
1035 extern void zone_pageout_stat(int, zone_pageout_op_t);
1036 extern void zone_get_physmem_data(int, pgcnt_t *, pgcnt_t *);
1037 
1038 /* Interfaces for page scanning */
1039 extern uint_t zone_num_over_cap;
1040 extern zone_persist_t zone_pdata[MAX_ZONES];
1041 
1042 extern rctl_hndl_t rc_zone_locked_mem;
1043 extern rctl_hndl_t rc_zone_max_swap;
1044 extern rctl_hndl_t rc_zone_phys_mem;
1045 extern rctl_hndl_t rc_zone_max_lofi;
1046 
1047 /* For publishing sysevents related to a particular zone */
1048 extern void zone_sysevent_publish(zone_t *, const char *, const char *,
1049     nvlist_t *);
1050 
1051 #endif  /* _KERNEL */
1052 
1053 #ifdef  __cplusplus
1054 }
1055 #endif
1056 
1057 #endif  /* _SYS_ZONE_H */