17c478bd9Sstevel@tonic-gate /* 27c478bd9Sstevel@tonic-gate * CDDL HEADER START 37c478bd9Sstevel@tonic-gate * 47c478bd9Sstevel@tonic-gate * The contents of this file are subject to the terms of the 545916cd2Sjpk * Common Development and Distribution License (the "License"). 645916cd2Sjpk * You may not use this file except in compliance with the License. 77c478bd9Sstevel@tonic-gate * 87c478bd9Sstevel@tonic-gate * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 97c478bd9Sstevel@tonic-gate * or http://www.opensolaris.org/os/licensing. 107c478bd9Sstevel@tonic-gate * See the License for the specific language governing permissions 117c478bd9Sstevel@tonic-gate * and limitations under the License. 127c478bd9Sstevel@tonic-gate * 137c478bd9Sstevel@tonic-gate * When distributing Covered Code, include this CDDL HEADER in each 147c478bd9Sstevel@tonic-gate * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 157c478bd9Sstevel@tonic-gate * If applicable, add the following below this CDDL HEADER, with the 167c478bd9Sstevel@tonic-gate * fields enclosed by brackets "[]" replaced with your own identifying 177c478bd9Sstevel@tonic-gate * information: Portions Copyright [yyyy] [name of copyright owner] 187c478bd9Sstevel@tonic-gate * 197c478bd9Sstevel@tonic-gate * CDDL HEADER END 207c478bd9Sstevel@tonic-gate */ 2197eda132Sraf 227c478bd9Sstevel@tonic-gate /* 23134a1f4eSCasper H.S. Dik * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. 24542a813cSJerry Jelinek * Copyright 2013, Joyent Inc. All rights reserved. 257c478bd9Sstevel@tonic-gate */ 267c478bd9Sstevel@tonic-gate 277c478bd9Sstevel@tonic-gate /* 287c478bd9Sstevel@tonic-gate * Zones 297c478bd9Sstevel@tonic-gate * 307c478bd9Sstevel@tonic-gate * A zone is a named collection of processes, namespace constraints, 317c478bd9Sstevel@tonic-gate * and other system resources which comprise a secure and manageable 327c478bd9Sstevel@tonic-gate * application containment facility. 337c478bd9Sstevel@tonic-gate * 347c478bd9Sstevel@tonic-gate * Zones (represented by the reference counted zone_t) are tracked in 357c478bd9Sstevel@tonic-gate * the kernel in the zonehash. Elsewhere in the kernel, Zone IDs 367c478bd9Sstevel@tonic-gate * (zoneid_t) are used to track zone association. Zone IDs are 377c478bd9Sstevel@tonic-gate * dynamically generated when the zone is created; if a persistent 387c478bd9Sstevel@tonic-gate * identifier is needed (core files, accounting logs, audit trail, 397c478bd9Sstevel@tonic-gate * etc.), the zone name should be used. 407c478bd9Sstevel@tonic-gate * 417c478bd9Sstevel@tonic-gate * 427c478bd9Sstevel@tonic-gate * Global Zone: 437c478bd9Sstevel@tonic-gate * 447c478bd9Sstevel@tonic-gate * The global zone (zoneid 0) is automatically associated with all 457c478bd9Sstevel@tonic-gate * system resources that have not been bound to a user-created zone. 467c478bd9Sstevel@tonic-gate * This means that even systems where zones are not in active use 477c478bd9Sstevel@tonic-gate * have a global zone, and all processes, mounts, etc. are 487c478bd9Sstevel@tonic-gate * associated with that zone. The global zone is generally 497c478bd9Sstevel@tonic-gate * unconstrained in terms of privileges and access, though the usual 507c478bd9Sstevel@tonic-gate * credential and privilege based restrictions apply. 517c478bd9Sstevel@tonic-gate * 527c478bd9Sstevel@tonic-gate * 537c478bd9Sstevel@tonic-gate * Zone States: 547c478bd9Sstevel@tonic-gate * 557c478bd9Sstevel@tonic-gate * The states in which a zone may be in and the transitions are as 567c478bd9Sstevel@tonic-gate * follows: 577c478bd9Sstevel@tonic-gate * 587c478bd9Sstevel@tonic-gate * ZONE_IS_UNINITIALIZED: primordial state for a zone. The partially 597c478bd9Sstevel@tonic-gate * initialized zone is added to the list of active zones on the system but 607c478bd9Sstevel@tonic-gate * isn't accessible. 617c478bd9Sstevel@tonic-gate * 62bd41d0a8Snordmark * ZONE_IS_INITIALIZED: Initialization complete except the ZSD callbacks are 63bd41d0a8Snordmark * not yet completed. Not possible to enter the zone, but attributes can 64bd41d0a8Snordmark * be retrieved. 65bd41d0a8Snordmark * 667c478bd9Sstevel@tonic-gate * ZONE_IS_READY: zsched (the kernel dummy process for a zone) is 677c478bd9Sstevel@tonic-gate * ready. The zone is made visible after the ZSD constructor callbacks are 687c478bd9Sstevel@tonic-gate * executed. A zone remains in this state until it transitions into 697c478bd9Sstevel@tonic-gate * the ZONE_IS_BOOTING state as a result of a call to zone_boot(). 707c478bd9Sstevel@tonic-gate * 717c478bd9Sstevel@tonic-gate * ZONE_IS_BOOTING: in this shortlived-state, zsched attempts to start 727c478bd9Sstevel@tonic-gate * init. Should that fail, the zone proceeds to the ZONE_IS_SHUTTING_DOWN 737c478bd9Sstevel@tonic-gate * state. 747c478bd9Sstevel@tonic-gate * 757c478bd9Sstevel@tonic-gate * ZONE_IS_RUNNING: The zone is open for business: zsched has 767c478bd9Sstevel@tonic-gate * successfully started init. A zone remains in this state until 777c478bd9Sstevel@tonic-gate * zone_shutdown() is called. 787c478bd9Sstevel@tonic-gate * 797c478bd9Sstevel@tonic-gate * ZONE_IS_SHUTTING_DOWN: zone_shutdown() has been called, the system is 807c478bd9Sstevel@tonic-gate * killing all processes running in the zone. The zone remains 817c478bd9Sstevel@tonic-gate * in this state until there are no more user processes running in the zone. 827c478bd9Sstevel@tonic-gate * zone_create(), zone_enter(), and zone_destroy() on this zone will fail. 837c478bd9Sstevel@tonic-gate * Since zone_shutdown() is restartable, it may be called successfully 847c478bd9Sstevel@tonic-gate * multiple times for the same zone_t. Setting of the zone's state to 857c478bd9Sstevel@tonic-gate * ZONE_IS_SHUTTING_DOWN is synchronized with mounts, so VOP_MOUNT() may check 867c478bd9Sstevel@tonic-gate * the zone's status without worrying about it being a moving target. 877c478bd9Sstevel@tonic-gate * 887c478bd9Sstevel@tonic-gate * ZONE_IS_EMPTY: zone_shutdown() has been called, and there 897c478bd9Sstevel@tonic-gate * are no more user processes in the zone. The zone remains in this 907c478bd9Sstevel@tonic-gate * state until there are no more kernel threads associated with the 917c478bd9Sstevel@tonic-gate * zone. zone_create(), zone_enter(), and zone_destroy() on this zone will 927c478bd9Sstevel@tonic-gate * fail. 937c478bd9Sstevel@tonic-gate * 947c478bd9Sstevel@tonic-gate * ZONE_IS_DOWN: All kernel threads doing work on behalf of the zone 957c478bd9Sstevel@tonic-gate * have exited. zone_shutdown() returns. Henceforth it is not possible to 967c478bd9Sstevel@tonic-gate * join the zone or create kernel threads therein. 977c478bd9Sstevel@tonic-gate * 987c478bd9Sstevel@tonic-gate * ZONE_IS_DYING: zone_destroy() has been called on the zone; zone 997c478bd9Sstevel@tonic-gate * remains in this state until zsched exits. Calls to zone_find_by_*() 1007c478bd9Sstevel@tonic-gate * return NULL from now on. 1017c478bd9Sstevel@tonic-gate * 1027c478bd9Sstevel@tonic-gate * ZONE_IS_DEAD: zsched has exited (zone_ntasks == 0). There are no 1037c478bd9Sstevel@tonic-gate * processes or threads doing work on behalf of the zone. The zone is 1047c478bd9Sstevel@tonic-gate * removed from the list of active zones. zone_destroy() returns, and 1057c478bd9Sstevel@tonic-gate * the zone can be recreated. 1067c478bd9Sstevel@tonic-gate * 1077c478bd9Sstevel@tonic-gate * ZONE_IS_FREE (internal state): zone_ref goes to 0, ZSD destructor 1087c478bd9Sstevel@tonic-gate * callbacks are executed, and all memory associated with the zone is 1097c478bd9Sstevel@tonic-gate * freed. 1107c478bd9Sstevel@tonic-gate * 1117c478bd9Sstevel@tonic-gate * Threads can wait for the zone to enter a requested state by using 1127c478bd9Sstevel@tonic-gate * zone_status_wait() or zone_status_timedwait() with the desired 1137c478bd9Sstevel@tonic-gate * state passed in as an argument. Zone state transitions are 1147c478bd9Sstevel@tonic-gate * uni-directional; it is not possible to move back to an earlier state. 1157c478bd9Sstevel@tonic-gate * 1167c478bd9Sstevel@tonic-gate * 1177c478bd9Sstevel@tonic-gate * Zone-Specific Data: 1187c478bd9Sstevel@tonic-gate * 1197c478bd9Sstevel@tonic-gate * Subsystems needing to maintain zone-specific data can store that 1207c478bd9Sstevel@tonic-gate * data using the ZSD mechanism. This provides a zone-specific data 1217c478bd9Sstevel@tonic-gate * store, similar to thread-specific data (see pthread_getspecific(3C) 1227c478bd9Sstevel@tonic-gate * or the TSD code in uts/common/disp/thread.c. Also, ZSD can be used 1237c478bd9Sstevel@tonic-gate * to register callbacks to be invoked when a zone is created, shut 1247c478bd9Sstevel@tonic-gate * down, or destroyed. This can be used to initialize zone-specific 1257c478bd9Sstevel@tonic-gate * data for new zones and to clean up when zones go away. 1267c478bd9Sstevel@tonic-gate * 1277c478bd9Sstevel@tonic-gate * 1287c478bd9Sstevel@tonic-gate * Data Structures: 1297c478bd9Sstevel@tonic-gate * 1307c478bd9Sstevel@tonic-gate * The per-zone structure (zone_t) is reference counted, and freed 1317c478bd9Sstevel@tonic-gate * when all references are released. zone_hold and zone_rele can be 1327c478bd9Sstevel@tonic-gate * used to adjust the reference count. In addition, reference counts 1337c478bd9Sstevel@tonic-gate * associated with the cred_t structure are tracked separately using 1347c478bd9Sstevel@tonic-gate * zone_cred_hold and zone_cred_rele. 1357c478bd9Sstevel@tonic-gate * 1367c478bd9Sstevel@tonic-gate * Pointers to active zone_t's are stored in two hash tables; one 1377c478bd9Sstevel@tonic-gate * for searching by id, the other for searching by name. Lookups 1387c478bd9Sstevel@tonic-gate * can be performed on either basis, using zone_find_by_id and 1397c478bd9Sstevel@tonic-gate * zone_find_by_name. Both return zone_t pointers with the zone 1407c478bd9Sstevel@tonic-gate * held, so zone_rele should be called when the pointer is no longer 1417c478bd9Sstevel@tonic-gate * needed. Zones can also be searched by path; zone_find_by_path 1427c478bd9Sstevel@tonic-gate * returns the zone with which a path name is associated (global 1437c478bd9Sstevel@tonic-gate * zone if the path is not within some other zone's file system 1447c478bd9Sstevel@tonic-gate * hierarchy). This currently requires iterating through each zone, 1457c478bd9Sstevel@tonic-gate * so it is slower than an id or name search via a hash table. 1467c478bd9Sstevel@tonic-gate * 1477c478bd9Sstevel@tonic-gate * 1487c478bd9Sstevel@tonic-gate * Locking: 1497c478bd9Sstevel@tonic-gate * 1507c478bd9Sstevel@tonic-gate * zonehash_lock: This is a top-level global lock used to protect the 1517c478bd9Sstevel@tonic-gate * zone hash tables and lists. Zones cannot be created or destroyed 1527c478bd9Sstevel@tonic-gate * while this lock is held. 1537c478bd9Sstevel@tonic-gate * zone_status_lock: This is a global lock protecting zone state. 1547c478bd9Sstevel@tonic-gate * Zones cannot change state while this lock is held. It also 1557c478bd9Sstevel@tonic-gate * protects the list of kernel threads associated with a zone. 1567c478bd9Sstevel@tonic-gate * zone_lock: This is a per-zone lock used to protect several fields of 1577c478bd9Sstevel@tonic-gate * the zone_t (see <sys/zone.h> for details). In addition, holding 1587c478bd9Sstevel@tonic-gate * this lock means that the zone cannot go away. 1590209230bSgjelinek * zone_nlwps_lock: This is a per-zone lock used to protect the fields 1600209230bSgjelinek * related to the zone.max-lwps rctl. 1610209230bSgjelinek * zone_mem_lock: This is a per-zone lock used to protect the fields 1620209230bSgjelinek * related to the zone.max-locked-memory and zone.max-swap rctls. 1630fbb751dSJohn Levon * zone_rctl_lock: This is a per-zone lock used to protect other rctls, 1640fbb751dSJohn Levon * currently just max_lofi 1657c478bd9Sstevel@tonic-gate * zsd_key_lock: This is a global lock protecting the key state for ZSD. 1667c478bd9Sstevel@tonic-gate * zone_deathrow_lock: This is a global lock protecting the "deathrow" 1677c478bd9Sstevel@tonic-gate * list (a list of zones in the ZONE_IS_DEAD state). 1687c478bd9Sstevel@tonic-gate * 1697c478bd9Sstevel@tonic-gate * Ordering requirements: 1707c478bd9Sstevel@tonic-gate * pool_lock --> cpu_lock --> zonehash_lock --> zone_status_lock --> 1717c478bd9Sstevel@tonic-gate * zone_lock --> zsd_key_lock --> pidlock --> p_lock 1727c478bd9Sstevel@tonic-gate * 1730209230bSgjelinek * When taking zone_mem_lock or zone_nlwps_lock, the lock ordering is: 1740209230bSgjelinek * zonehash_lock --> a_lock --> pidlock --> p_lock --> zone_mem_lock 175ff19e029SMenno Lageman * zonehash_lock --> a_lock --> pidlock --> p_lock --> zone_nlwps_lock 1760209230bSgjelinek * 1777c478bd9Sstevel@tonic-gate * Blocking memory allocations are permitted while holding any of the 1787c478bd9Sstevel@tonic-gate * zone locks. 1797c478bd9Sstevel@tonic-gate * 1807c478bd9Sstevel@tonic-gate * 1817c478bd9Sstevel@tonic-gate * System Call Interface: 1827c478bd9Sstevel@tonic-gate * 1837c478bd9Sstevel@tonic-gate * The zone subsystem can be managed and queried from user level with 1847c478bd9Sstevel@tonic-gate * the following system calls (all subcodes of the primary "zone" 1857c478bd9Sstevel@tonic-gate * system call): 1867c478bd9Sstevel@tonic-gate * - zone_create: creates a zone with selected attributes (name, 187fa9e4066Sahrens * root path, privileges, resource controls, ZFS datasets) 1887c478bd9Sstevel@tonic-gate * - zone_enter: allows the current process to enter a zone 1897c478bd9Sstevel@tonic-gate * - zone_getattr: reports attributes of a zone 1903f2f09c1Sdp * - zone_setattr: set attributes of a zone 1913f2f09c1Sdp * - zone_boot: set 'init' running for the zone 1927c478bd9Sstevel@tonic-gate * - zone_list: lists all zones active in the system 1937c478bd9Sstevel@tonic-gate * - zone_lookup: looks up zone id based on name 1947c478bd9Sstevel@tonic-gate * - zone_shutdown: initiates shutdown process (see states above) 1957c478bd9Sstevel@tonic-gate * - zone_destroy: completes shutdown process (see states above) 1967c478bd9Sstevel@tonic-gate * 1977c478bd9Sstevel@tonic-gate */ 1987c478bd9Sstevel@tonic-gate 1997c478bd9Sstevel@tonic-gate #include <sys/priv_impl.h> 2007c478bd9Sstevel@tonic-gate #include <sys/cred.h> 2017c478bd9Sstevel@tonic-gate #include <c2/audit.h> 2027c478bd9Sstevel@tonic-gate #include <sys/debug.h> 2037c478bd9Sstevel@tonic-gate #include <sys/file.h> 2047c478bd9Sstevel@tonic-gate #include <sys/kmem.h> 2050209230bSgjelinek #include <sys/kstat.h> 2067c478bd9Sstevel@tonic-gate #include <sys/mutex.h> 20745916cd2Sjpk #include <sys/note.h> 2087c478bd9Sstevel@tonic-gate #include <sys/pathname.h> 2097c478bd9Sstevel@tonic-gate #include <sys/proc.h> 2107c478bd9Sstevel@tonic-gate #include <sys/project.h> 211cf8f45c7Sdstaff #include <sys/sysevent.h> 2127c478bd9Sstevel@tonic-gate #include <sys/task.h> 2137c478bd9Sstevel@tonic-gate #include <sys/systm.h> 2147c478bd9Sstevel@tonic-gate #include <sys/types.h> 2157c478bd9Sstevel@tonic-gate #include <sys/utsname.h> 2167c478bd9Sstevel@tonic-gate #include <sys/vnode.h> 2177c478bd9Sstevel@tonic-gate #include <sys/vfs.h> 2187c478bd9Sstevel@tonic-gate #include <sys/systeminfo.h> 2197c478bd9Sstevel@tonic-gate #include <sys/policy.h> 2207c478bd9Sstevel@tonic-gate #include <sys/cred_impl.h> 2217c478bd9Sstevel@tonic-gate #include <sys/contract_impl.h> 2227c478bd9Sstevel@tonic-gate #include <sys/contract/process_impl.h> 2237c478bd9Sstevel@tonic-gate #include <sys/class.h> 2247c478bd9Sstevel@tonic-gate #include <sys/pool.h> 2257c478bd9Sstevel@tonic-gate #include <sys/pool_pset.h> 2267c478bd9Sstevel@tonic-gate #include <sys/pset.h> 227a19609f8Sjv #include <sys/strlog.h> 2287c478bd9Sstevel@tonic-gate #include <sys/sysmacros.h> 2297c478bd9Sstevel@tonic-gate #include <sys/callb.h> 2307c478bd9Sstevel@tonic-gate #include <sys/vmparam.h> 2317c478bd9Sstevel@tonic-gate #include <sys/corectl.h> 232824c205fSml #include <sys/ipc_impl.h> 233134a1f4eSCasper H.S. Dik #include <sys/klpd.h> 2347c478bd9Sstevel@tonic-gate 2357c478bd9Sstevel@tonic-gate #include <sys/door.h> 2367c478bd9Sstevel@tonic-gate #include <sys/cpuvar.h> 237bd41d0a8Snordmark #include <sys/sdt.h> 2387c478bd9Sstevel@tonic-gate 2397c478bd9Sstevel@tonic-gate #include <sys/uadmin.h> 2407c478bd9Sstevel@tonic-gate #include <sys/session.h> 2417c478bd9Sstevel@tonic-gate #include <sys/cmn_err.h> 2427c478bd9Sstevel@tonic-gate #include <sys/modhash.h> 2433f2f09c1Sdp #include <sys/sunddi.h> 2447c478bd9Sstevel@tonic-gate #include <sys/nvpair.h> 2457c478bd9Sstevel@tonic-gate #include <sys/rctl.h> 2467c478bd9Sstevel@tonic-gate #include <sys/fss.h> 2479acbbeafSnn #include <sys/brand.h> 2487c478bd9Sstevel@tonic-gate #include <sys/zone.h> 249f4b3ec61Sdh #include <net/if.h> 250c97ad5cdSakolb #include <sys/cpucaps.h> 2510209230bSgjelinek #include <vm/seg.h> 2522b24ab6bSSebastien Roy #include <sys/mac.h> 2532b24ab6bSSebastien Roy 254a19609f8Sjv /* 255a19609f8Sjv * This constant specifies the number of seconds that threads waiting for 256a19609f8Sjv * subsystems to release a zone's general-purpose references will wait before 257a19609f8Sjv * they log the zone's reference counts. The constant's value shouldn't 258a19609f8Sjv * be so small that reference counts are unnecessarily reported for zones 259a19609f8Sjv * whose references are slowly released. On the other hand, it shouldn't be so 260a19609f8Sjv * large that users reboot their systems out of frustration over hung zones 261a19609f8Sjv * before the system logs the zones' reference counts. 262a19609f8Sjv */ 263a19609f8Sjv #define ZONE_DESTROY_TIMEOUT_SECS 60 264a19609f8Sjv 2652b24ab6bSSebastien Roy /* List of data link IDs which are accessible from the zone */ 2662b24ab6bSSebastien Roy typedef struct zone_dl { 2672b24ab6bSSebastien Roy datalink_id_t zdl_id; 268550b6e40SSowmini Varadhan nvlist_t *zdl_net; 2692b24ab6bSSebastien Roy list_node_t zdl_linkage; 2702b24ab6bSSebastien Roy } zone_dl_t; 2710209230bSgjelinek 2727c478bd9Sstevel@tonic-gate /* 2737c478bd9Sstevel@tonic-gate * cv used to signal that all references to the zone have been released. This 2747c478bd9Sstevel@tonic-gate * needs to be global since there may be multiple waiters, and the first to 2757c478bd9Sstevel@tonic-gate * wake up will free the zone_t, hence we cannot use zone->zone_cv. 2767c478bd9Sstevel@tonic-gate */ 2777c478bd9Sstevel@tonic-gate static kcondvar_t zone_destroy_cv; 2787c478bd9Sstevel@tonic-gate /* 2797c478bd9Sstevel@tonic-gate * Lock used to serialize access to zone_cv. This could have been per-zone, 2807c478bd9Sstevel@tonic-gate * but then we'd need another lock for zone_destroy_cv, and why bother? 2817c478bd9Sstevel@tonic-gate */ 2827c478bd9Sstevel@tonic-gate static kmutex_t zone_status_lock; 2837c478bd9Sstevel@tonic-gate 2847c478bd9Sstevel@tonic-gate /* 2857c478bd9Sstevel@tonic-gate * ZSD-related global variables. 2867c478bd9Sstevel@tonic-gate */ 2877c478bd9Sstevel@tonic-gate static kmutex_t zsd_key_lock; /* protects the following two */ 2887c478bd9Sstevel@tonic-gate /* 2897c478bd9Sstevel@tonic-gate * The next caller of zone_key_create() will be assigned a key of ++zsd_keyval. 2907c478bd9Sstevel@tonic-gate */ 2917c478bd9Sstevel@tonic-gate static zone_key_t zsd_keyval = 0; 2927c478bd9Sstevel@tonic-gate /* 2937c478bd9Sstevel@tonic-gate * Global list of registered keys. We use this when a new zone is created. 2947c478bd9Sstevel@tonic-gate */ 2957c478bd9Sstevel@tonic-gate static list_t zsd_registered_keys; 2967c478bd9Sstevel@tonic-gate 2977c478bd9Sstevel@tonic-gate int zone_hash_size = 256; 29845916cd2Sjpk static mod_hash_t *zonehashbyname, *zonehashbyid, *zonehashbylabel; 2997c478bd9Sstevel@tonic-gate static kmutex_t zonehash_lock; 3007c478bd9Sstevel@tonic-gate static uint_t zonecount; 3017c478bd9Sstevel@tonic-gate static id_space_t *zoneid_space; 3027c478bd9Sstevel@tonic-gate 3037c478bd9Sstevel@tonic-gate /* 3047c478bd9Sstevel@tonic-gate * The global zone (aka zone0) is the all-seeing, all-knowing zone in which the 3057c478bd9Sstevel@tonic-gate * kernel proper runs, and which manages all other zones. 3067c478bd9Sstevel@tonic-gate * 3077c478bd9Sstevel@tonic-gate * Although not declared as static, the variable "zone0" should not be used 3087c478bd9Sstevel@tonic-gate * except for by code that needs to reference the global zone early on in boot, 3097c478bd9Sstevel@tonic-gate * before it is fully initialized. All other consumers should use 3107c478bd9Sstevel@tonic-gate * 'global_zone'. 3117c478bd9Sstevel@tonic-gate */ 3127c478bd9Sstevel@tonic-gate zone_t zone0; 3137c478bd9Sstevel@tonic-gate zone_t *global_zone = NULL; /* Set when the global zone is initialized */ 3147c478bd9Sstevel@tonic-gate 3157c478bd9Sstevel@tonic-gate /* 3167c478bd9Sstevel@tonic-gate * List of active zones, protected by zonehash_lock. 3177c478bd9Sstevel@tonic-gate */ 3187c478bd9Sstevel@tonic-gate static list_t zone_active; 3197c478bd9Sstevel@tonic-gate 3207c478bd9Sstevel@tonic-gate /* 3217c478bd9Sstevel@tonic-gate * List of destroyed zones that still have outstanding cred references. 3227c478bd9Sstevel@tonic-gate * Used for debugging. Uses a separate lock to avoid lock ordering 3237c478bd9Sstevel@tonic-gate * problems in zone_free. 3247c478bd9Sstevel@tonic-gate */ 3257c478bd9Sstevel@tonic-gate static list_t zone_deathrow; 3267c478bd9Sstevel@tonic-gate static kmutex_t zone_deathrow_lock; 3277c478bd9Sstevel@tonic-gate 3287c478bd9Sstevel@tonic-gate /* number of zones is limited by virtual interface limit in IP */ 3297c478bd9Sstevel@tonic-gate uint_t maxzones = 8192; 3307c478bd9Sstevel@tonic-gate 331cf8f45c7Sdstaff /* Event channel to sent zone state change notifications */ 332cf8f45c7Sdstaff evchan_t *zone_event_chan; 333cf8f45c7Sdstaff 334cf8f45c7Sdstaff /* 335cf8f45c7Sdstaff * This table holds the mapping from kernel zone states to 336cf8f45c7Sdstaff * states visible in the state notification API. 337cf8f45c7Sdstaff * The idea is that we only expose "obvious" states and 338cf8f45c7Sdstaff * do not expose states which are just implementation details. 339cf8f45c7Sdstaff */ 340cf8f45c7Sdstaff const char *zone_status_table[] = { 341cf8f45c7Sdstaff ZONE_EVENT_UNINITIALIZED, /* uninitialized */ 342bd41d0a8Snordmark ZONE_EVENT_INITIALIZED, /* initialized */ 343cf8f45c7Sdstaff ZONE_EVENT_READY, /* ready */ 344cf8f45c7Sdstaff ZONE_EVENT_READY, /* booting */ 345cf8f45c7Sdstaff ZONE_EVENT_RUNNING, /* running */ 346cf8f45c7Sdstaff ZONE_EVENT_SHUTTING_DOWN, /* shutting_down */ 347cf8f45c7Sdstaff ZONE_EVENT_SHUTTING_DOWN, /* empty */ 348cf8f45c7Sdstaff ZONE_EVENT_SHUTTING_DOWN, /* down */ 349cf8f45c7Sdstaff ZONE_EVENT_SHUTTING_DOWN, /* dying */ 350cf8f45c7Sdstaff ZONE_EVENT_UNINITIALIZED, /* dead */ 351cf8f45c7Sdstaff }; 352cf8f45c7Sdstaff 353a19609f8Sjv /* 354a19609f8Sjv * This array contains the names of the subsystems listed in zone_ref_subsys_t 355a19609f8Sjv * (see sys/zone.h). 356a19609f8Sjv */ 357a19609f8Sjv static char *zone_ref_subsys_names[] = { 358a19609f8Sjv "NFS", /* ZONE_REF_NFS */ 359a19609f8Sjv "NFSv4", /* ZONE_REF_NFSV4 */ 360a19609f8Sjv "SMBFS", /* ZONE_REF_SMBFS */ 361a19609f8Sjv "MNTFS", /* ZONE_REF_MNTFS */ 362a19609f8Sjv "LOFI", /* ZONE_REF_LOFI */ 363a19609f8Sjv "VFS", /* ZONE_REF_VFS */ 364a19609f8Sjv "IPC" /* ZONE_REF_IPC */ 365a19609f8Sjv }; 366a19609f8Sjv 3677c478bd9Sstevel@tonic-gate /* 3687c478bd9Sstevel@tonic-gate * This isn't static so lint doesn't complain. 3697c478bd9Sstevel@tonic-gate */ 3707c478bd9Sstevel@tonic-gate rctl_hndl_t rc_zone_cpu_shares; 371c6939658Ssl rctl_hndl_t rc_zone_locked_mem; 3720209230bSgjelinek rctl_hndl_t rc_zone_max_swap; 3730fbb751dSJohn Levon rctl_hndl_t rc_zone_max_lofi; 374c97ad5cdSakolb rctl_hndl_t rc_zone_cpu_cap; 3757c478bd9Sstevel@tonic-gate rctl_hndl_t rc_zone_nlwps; 376ff19e029SMenno Lageman rctl_hndl_t rc_zone_nprocs; 377824c205fSml rctl_hndl_t rc_zone_shmmax; 378824c205fSml rctl_hndl_t rc_zone_shmmni; 379824c205fSml rctl_hndl_t rc_zone_semmni; 380824c205fSml rctl_hndl_t rc_zone_msgmni; 3817c478bd9Sstevel@tonic-gate 3823f2f09c1Sdp const char * const zone_default_initname = "/sbin/init"; 38345916cd2Sjpk static char * const zone_prefix = "/zone/"; 3847c478bd9Sstevel@tonic-gate static int zone_shutdown(zoneid_t zoneid); 3852b24ab6bSSebastien Roy static int zone_add_datalink(zoneid_t, datalink_id_t); 3862b24ab6bSSebastien Roy static int zone_remove_datalink(zoneid_t, datalink_id_t); 3872b24ab6bSSebastien Roy static int zone_list_datalink(zoneid_t, int *, datalink_id_t *); 388550b6e40SSowmini Varadhan static int zone_set_network(zoneid_t, zone_net_data_t *); 389550b6e40SSowmini Varadhan static int zone_get_network(zoneid_t, zone_net_data_t *); 3907c478bd9Sstevel@tonic-gate 391bd41d0a8Snordmark typedef boolean_t zsd_applyfn_t(kmutex_t *, boolean_t, zone_t *, zone_key_t); 392bd41d0a8Snordmark 393bd41d0a8Snordmark static void zsd_apply_all_zones(zsd_applyfn_t *, zone_key_t); 394bd41d0a8Snordmark static void zsd_apply_all_keys(zsd_applyfn_t *, zone_t *); 395bd41d0a8Snordmark static boolean_t zsd_apply_create(kmutex_t *, boolean_t, zone_t *, zone_key_t); 396bd41d0a8Snordmark static boolean_t zsd_apply_shutdown(kmutex_t *, boolean_t, zone_t *, 397bd41d0a8Snordmark zone_key_t); 398bd41d0a8Snordmark static boolean_t zsd_apply_destroy(kmutex_t *, boolean_t, zone_t *, zone_key_t); 399bd41d0a8Snordmark static boolean_t zsd_wait_for_creator(zone_t *, struct zsd_entry *, 400bd41d0a8Snordmark kmutex_t *); 401bd41d0a8Snordmark static boolean_t zsd_wait_for_inprogress(zone_t *, struct zsd_entry *, 402bd41d0a8Snordmark kmutex_t *); 403bd41d0a8Snordmark 404821c4a97Sdp /* 405821c4a97Sdp * Bump this number when you alter the zone syscall interfaces; this is 406821c4a97Sdp * because we need to have support for previous API versions in libc 407821c4a97Sdp * to support patching; libc calls into the kernel to determine this number. 408821c4a97Sdp * 409821c4a97Sdp * Version 1 of the API is the version originally shipped with Solaris 10 410821c4a97Sdp * Version 2 alters the zone_create system call in order to support more 411821c4a97Sdp * arguments by moving the args into a structure; and to do better 412821c4a97Sdp * error reporting when zone_create() fails. 413821c4a97Sdp * Version 3 alters the zone_create system call in order to support the 414821c4a97Sdp * import of ZFS datasets to zones. 41545916cd2Sjpk * Version 4 alters the zone_create system call in order to support 41645916cd2Sjpk * Trusted Extensions. 4173f2f09c1Sdp * Version 5 alters the zone_boot system call, and converts its old 4183f2f09c1Sdp * bootargs parameter to be set by the zone_setattr API instead. 419f4b3ec61Sdh * Version 6 adds the flag argument to zone_create. 420821c4a97Sdp */ 421f4b3ec61Sdh static const int ZONE_SYSCALL_API_VERSION = 6; 422821c4a97Sdp 4237c478bd9Sstevel@tonic-gate /* 4247c478bd9Sstevel@tonic-gate * Certain filesystems (such as NFS and autofs) need to know which zone 4257c478bd9Sstevel@tonic-gate * the mount is being placed in. Because of this, we need to be able to 426*5fd5c689SJerry Jelinek * ensure that a zone isn't in the process of being created/destroyed such 427*5fd5c689SJerry Jelinek * that nfs_mount() thinks it is in the global/NGZ zone, while by the time 428*5fd5c689SJerry Jelinek * it gets added the list of mounted zones, it ends up on the wrong zone's 429*5fd5c689SJerry Jelinek * mount list. Since a zone can't reside on an NFS file system, we don't 430*5fd5c689SJerry Jelinek * have to worry about the zonepath itself. 4317c478bd9Sstevel@tonic-gate * 4327c478bd9Sstevel@tonic-gate * The following functions: block_mounts()/resume_mounts() and 4337c478bd9Sstevel@tonic-gate * mount_in_progress()/mount_completed() are used by zones and the VFS 434*5fd5c689SJerry Jelinek * layer (respectively) to synchronize zone state transitions and new 435*5fd5c689SJerry Jelinek * mounts within a zone. This syncronization is on a per-zone basis, so 436*5fd5c689SJerry Jelinek * activity for one zone will not interfere with activity for another zone. 4377c478bd9Sstevel@tonic-gate * 4387c478bd9Sstevel@tonic-gate * The semantics are like a reader-reader lock such that there may 439*5fd5c689SJerry Jelinek * either be multiple mounts (or zone state transitions, if that weren't 4407c478bd9Sstevel@tonic-gate * serialized by zonehash_lock) in progress at the same time, but not 4417c478bd9Sstevel@tonic-gate * both. 4427c478bd9Sstevel@tonic-gate * 4437c478bd9Sstevel@tonic-gate * We use cv's so the user can ctrl-C out of the operation if it's 4447c478bd9Sstevel@tonic-gate * taking too long. 4457c478bd9Sstevel@tonic-gate * 4467c478bd9Sstevel@tonic-gate * The semantics are such that there is unfair bias towards the 447*5fd5c689SJerry Jelinek * "current" operation. This means that zone halt may starve if 448*5fd5c689SJerry Jelinek * there is a rapid succession of new mounts coming in to the zone. 4497c478bd9Sstevel@tonic-gate */ 4507c478bd9Sstevel@tonic-gate /* 4517c478bd9Sstevel@tonic-gate * Prevent new mounts from progressing to the point of calling 4527c478bd9Sstevel@tonic-gate * VFS_MOUNT(). If there are already mounts in this "region", wait for 4537c478bd9Sstevel@tonic-gate * them to complete. 4547c478bd9Sstevel@tonic-gate */ 4557c478bd9Sstevel@tonic-gate static int 456*5fd5c689SJerry Jelinek block_mounts(zone_t *zp) 4577c478bd9Sstevel@tonic-gate { 4587c478bd9Sstevel@tonic-gate int retval = 0; 4597c478bd9Sstevel@tonic-gate 4607c478bd9Sstevel@tonic-gate /* 4617c478bd9Sstevel@tonic-gate * Since it may block for a long time, block_mounts() shouldn't be 4627c478bd9Sstevel@tonic-gate * called with zonehash_lock held. 4637c478bd9Sstevel@tonic-gate */ 4647c478bd9Sstevel@tonic-gate ASSERT(MUTEX_NOT_HELD(&zonehash_lock)); 465*5fd5c689SJerry Jelinek mutex_enter(&zp->zone_mount_lock); 466*5fd5c689SJerry Jelinek while (zp->zone_mounts_in_progress > 0) { 467*5fd5c689SJerry Jelinek if (cv_wait_sig(&zp->zone_mount_cv, &zp->zone_mount_lock) == 0) 4687c478bd9Sstevel@tonic-gate goto signaled; 4697c478bd9Sstevel@tonic-gate } 4707c478bd9Sstevel@tonic-gate /* 4717c478bd9Sstevel@tonic-gate * A negative value of mounts_in_progress indicates that mounts 472*5fd5c689SJerry Jelinek * have been blocked by (-mounts_in_progress) different callers 473*5fd5c689SJerry Jelinek * (remotely possible if two threads enter zone_shutdown at the same 474*5fd5c689SJerry Jelinek * time). 4757c478bd9Sstevel@tonic-gate */ 476*5fd5c689SJerry Jelinek zp->zone_mounts_in_progress--; 4777c478bd9Sstevel@tonic-gate retval = 1; 4787c478bd9Sstevel@tonic-gate signaled: 479*5fd5c689SJerry Jelinek mutex_exit(&zp->zone_mount_lock); 4807c478bd9Sstevel@tonic-gate return (retval); 4817c478bd9Sstevel@tonic-gate } 4827c478bd9Sstevel@tonic-gate 4837c478bd9Sstevel@tonic-gate /* 4847c478bd9Sstevel@tonic-gate * The VFS layer may progress with new mounts as far as we're concerned. 4857c478bd9Sstevel@tonic-gate * Allow them to progress if we were the last obstacle. 4867c478bd9Sstevel@tonic-gate */ 4877c478bd9Sstevel@tonic-gate static void 488*5fd5c689SJerry Jelinek resume_mounts(zone_t *zp) 4897c478bd9Sstevel@tonic-gate { 490*5fd5c689SJerry Jelinek mutex_enter(&zp->zone_mount_lock); 491*5fd5c689SJerry Jelinek if (++zp->zone_mounts_in_progress == 0) 492*5fd5c689SJerry Jelinek cv_broadcast(&zp->zone_mount_cv); 493*5fd5c689SJerry Jelinek mutex_exit(&zp->zone_mount_lock); 4947c478bd9Sstevel@tonic-gate } 4957c478bd9Sstevel@tonic-gate 4967c478bd9Sstevel@tonic-gate /* 497*5fd5c689SJerry Jelinek * The VFS layer is busy with a mount; this zone should wait until all 498*5fd5c689SJerry Jelinek * of its mounts are completed to progress. 4997c478bd9Sstevel@tonic-gate */ 5007c478bd9Sstevel@tonic-gate void 501*5fd5c689SJerry Jelinek mount_in_progress(zone_t *zp) 5027c478bd9Sstevel@tonic-gate { 503*5fd5c689SJerry Jelinek mutex_enter(&zp->zone_mount_lock); 504*5fd5c689SJerry Jelinek while (zp->zone_mounts_in_progress < 0) 505*5fd5c689SJerry Jelinek cv_wait(&zp->zone_mount_cv, &zp->zone_mount_lock); 506*5fd5c689SJerry Jelinek zp->zone_mounts_in_progress++; 507*5fd5c689SJerry Jelinek mutex_exit(&zp->zone_mount_lock); 5087c478bd9Sstevel@tonic-gate } 5097c478bd9Sstevel@tonic-gate 5107c478bd9Sstevel@tonic-gate /* 5117c478bd9Sstevel@tonic-gate * VFS is done with one mount; wake up any waiting block_mounts() 5127c478bd9Sstevel@tonic-gate * callers if this is the last mount. 5137c478bd9Sstevel@tonic-gate */ 5147c478bd9Sstevel@tonic-gate void 515*5fd5c689SJerry Jelinek mount_completed(zone_t *zp) 5167c478bd9Sstevel@tonic-gate { 517*5fd5c689SJerry Jelinek mutex_enter(&zp->zone_mount_lock); 518*5fd5c689SJerry Jelinek if (--zp->zone_mounts_in_progress == 0) 519*5fd5c689SJerry Jelinek cv_broadcast(&zp->zone_mount_cv); 520*5fd5c689SJerry Jelinek mutex_exit(&zp->zone_mount_lock); 5217c478bd9Sstevel@tonic-gate } 5227c478bd9Sstevel@tonic-gate 5237c478bd9Sstevel@tonic-gate /* 5247c478bd9Sstevel@tonic-gate * ZSD routines. 5257c478bd9Sstevel@tonic-gate * 5267c478bd9Sstevel@tonic-gate * Zone Specific Data (ZSD) is modeled after Thread Specific Data as 5277c478bd9Sstevel@tonic-gate * defined by the pthread_key_create() and related interfaces. 5287c478bd9Sstevel@tonic-gate * 5297c478bd9Sstevel@tonic-gate * Kernel subsystems may register one or more data items and/or 5307c478bd9Sstevel@tonic-gate * callbacks to be executed when a zone is created, shutdown, or 5317c478bd9Sstevel@tonic-gate * destroyed. 5327c478bd9Sstevel@tonic-gate * 5337c478bd9Sstevel@tonic-gate * Unlike the thread counterpart, destructor callbacks will be executed 5347c478bd9Sstevel@tonic-gate * even if the data pointer is NULL and/or there are no constructor 5357c478bd9Sstevel@tonic-gate * callbacks, so it is the responsibility of such callbacks to check for 5367c478bd9Sstevel@tonic-gate * NULL data values if necessary. 5377c478bd9Sstevel@tonic-gate * 5387c478bd9Sstevel@tonic-gate * The locking strategy and overall picture is as follows: 5397c478bd9Sstevel@tonic-gate * 5407c478bd9Sstevel@tonic-gate * When someone calls zone_key_create(), a template ZSD entry is added to the 541bd41d0a8Snordmark * global list "zsd_registered_keys", protected by zsd_key_lock. While 542bd41d0a8Snordmark * holding that lock all the existing zones are marked as 543bd41d0a8Snordmark * ZSD_CREATE_NEEDED and a copy of the ZSD entry added to the per-zone 544bd41d0a8Snordmark * zone_zsd list (protected by zone_lock). The global list is updated first 545bd41d0a8Snordmark * (under zone_key_lock) to make sure that newly created zones use the 546bd41d0a8Snordmark * most recent list of keys. Then under zonehash_lock we walk the zones 547bd41d0a8Snordmark * and mark them. Similar locking is used in zone_key_delete(). 5487c478bd9Sstevel@tonic-gate * 549bd41d0a8Snordmark * The actual create, shutdown, and destroy callbacks are done without 550bd41d0a8Snordmark * holding any lock. And zsd_flags are used to ensure that the operations 551bd41d0a8Snordmark * completed so that when zone_key_create (and zone_create) is done, as well as 552bd41d0a8Snordmark * zone_key_delete (and zone_destroy) is done, all the necessary callbacks 553bd41d0a8Snordmark * are completed. 5547c478bd9Sstevel@tonic-gate * 5557c478bd9Sstevel@tonic-gate * When new zones are created constructor callbacks for all registered ZSD 556bd41d0a8Snordmark * entries will be called. That also uses the above two phases of marking 557bd41d0a8Snordmark * what needs to be done, and then running the callbacks without holding 558bd41d0a8Snordmark * any locks. 5597c478bd9Sstevel@tonic-gate * 5607c478bd9Sstevel@tonic-gate * The framework does not provide any locking around zone_getspecific() and 5617c478bd9Sstevel@tonic-gate * zone_setspecific() apart from that needed for internal consistency, so 5627c478bd9Sstevel@tonic-gate * callers interested in atomic "test-and-set" semantics will need to provide 5637c478bd9Sstevel@tonic-gate * their own locking. 5647c478bd9Sstevel@tonic-gate */ 5657c478bd9Sstevel@tonic-gate 566bd41d0a8Snordmark /* 567bd41d0a8Snordmark * Helper function to find the zsd_entry associated with the key in the 568bd41d0a8Snordmark * given list. 569bd41d0a8Snordmark */ 570bd41d0a8Snordmark static struct zsd_entry * 571bd41d0a8Snordmark zsd_find(list_t *l, zone_key_t key) 572bd41d0a8Snordmark { 573bd41d0a8Snordmark struct zsd_entry *zsd; 5747c478bd9Sstevel@tonic-gate 575bd41d0a8Snordmark for (zsd = list_head(l); zsd != NULL; zsd = list_next(l, zsd)) { 576bd41d0a8Snordmark if (zsd->zsd_key == key) { 577bd41d0a8Snordmark return (zsd); 5787c478bd9Sstevel@tonic-gate } 5797c478bd9Sstevel@tonic-gate } 580bd41d0a8Snordmark return (NULL); 5817c478bd9Sstevel@tonic-gate } 5827c478bd9Sstevel@tonic-gate 5837c478bd9Sstevel@tonic-gate /* 5847c478bd9Sstevel@tonic-gate * Helper function to find the zsd_entry associated with the key in the 585bd41d0a8Snordmark * given list. Move it to the front of the list. 5867c478bd9Sstevel@tonic-gate */ 5877c478bd9Sstevel@tonic-gate static struct zsd_entry * 588bd41d0a8Snordmark zsd_find_mru(list_t *l, zone_key_t key) 5897c478bd9Sstevel@tonic-gate { 5907c478bd9Sstevel@tonic-gate struct zsd_entry *zsd; 5917c478bd9Sstevel@tonic-gate 5927c478bd9Sstevel@tonic-gate for (zsd = list_head(l); zsd != NULL; zsd = list_next(l, zsd)) { 5937c478bd9Sstevel@tonic-gate if (zsd->zsd_key == key) { 5947c478bd9Sstevel@tonic-gate /* 5957c478bd9Sstevel@tonic-gate * Move to head of list to keep list in MRU order. 5967c478bd9Sstevel@tonic-gate */ 5977c478bd9Sstevel@tonic-gate if (zsd != list_head(l)) { 5987c478bd9Sstevel@tonic-gate list_remove(l, zsd); 5997c478bd9Sstevel@tonic-gate list_insert_head(l, zsd); 6007c478bd9Sstevel@tonic-gate } 6017c478bd9Sstevel@tonic-gate return (zsd); 6027c478bd9Sstevel@tonic-gate } 6037c478bd9Sstevel@tonic-gate } 6047c478bd9Sstevel@tonic-gate return (NULL); 6057c478bd9Sstevel@tonic-gate } 6067c478bd9Sstevel@tonic-gate 607bd41d0a8Snordmark void 608bd41d0a8Snordmark zone_key_create(zone_key_t *keyp, void *(*create)(zoneid_t), 609bd41d0a8Snordmark void (*shutdown)(zoneid_t, void *), void (*destroy)(zoneid_t, void *)) 610bd41d0a8Snordmark { 611bd41d0a8Snordmark struct zsd_entry *zsdp; 612bd41d0a8Snordmark struct zsd_entry *t; 613bd41d0a8Snordmark struct zone *zone; 614bd41d0a8Snordmark zone_key_t key; 615bd41d0a8Snordmark 616bd41d0a8Snordmark zsdp = kmem_zalloc(sizeof (*zsdp), KM_SLEEP); 617bd41d0a8Snordmark zsdp->zsd_data = NULL; 618bd41d0a8Snordmark zsdp->zsd_create = create; 619bd41d0a8Snordmark zsdp->zsd_shutdown = shutdown; 620bd41d0a8Snordmark zsdp->zsd_destroy = destroy; 621bd41d0a8Snordmark 622bd41d0a8Snordmark /* 623bd41d0a8Snordmark * Insert in global list of callbacks. Makes future zone creations 624bd41d0a8Snordmark * see it. 625bd41d0a8Snordmark */ 626bd41d0a8Snordmark mutex_enter(&zsd_key_lock); 627fe16170aSPramod Batni key = zsdp->zsd_key = ++zsd_keyval; 628bd41d0a8Snordmark ASSERT(zsd_keyval != 0); 629bd41d0a8Snordmark list_insert_tail(&zsd_registered_keys, zsdp); 630bd41d0a8Snordmark mutex_exit(&zsd_key_lock); 631bd41d0a8Snordmark 632bd41d0a8Snordmark /* 633bd41d0a8Snordmark * Insert for all existing zones and mark them as needing 634bd41d0a8Snordmark * a create callback. 635bd41d0a8Snordmark */ 636bd41d0a8Snordmark mutex_enter(&zonehash_lock); /* stop the world */ 637bd41d0a8Snordmark for (zone = list_head(&zone_active); zone != NULL; 638bd41d0a8Snordmark zone = list_next(&zone_active, zone)) { 639bd41d0a8Snordmark zone_status_t status; 640bd41d0a8Snordmark 641bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 642bd41d0a8Snordmark 643bd41d0a8Snordmark /* Skip zones that are on the way down or not yet up */ 644bd41d0a8Snordmark status = zone_status_get(zone); 645bd41d0a8Snordmark if (status >= ZONE_IS_DOWN || 646bd41d0a8Snordmark status == ZONE_IS_UNINITIALIZED) { 647bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 648bd41d0a8Snordmark continue; 649bd41d0a8Snordmark } 650bd41d0a8Snordmark 651bd41d0a8Snordmark t = zsd_find_mru(&zone->zone_zsd, key); 652bd41d0a8Snordmark if (t != NULL) { 653bd41d0a8Snordmark /* 654bd41d0a8Snordmark * A zsd_configure already inserted it after 655bd41d0a8Snordmark * we dropped zsd_key_lock above. 656bd41d0a8Snordmark */ 657bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 658bd41d0a8Snordmark continue; 659bd41d0a8Snordmark } 660bd41d0a8Snordmark t = kmem_zalloc(sizeof (*t), KM_SLEEP); 661bd41d0a8Snordmark t->zsd_key = key; 662bd41d0a8Snordmark t->zsd_create = create; 663bd41d0a8Snordmark t->zsd_shutdown = shutdown; 664bd41d0a8Snordmark t->zsd_destroy = destroy; 665bd41d0a8Snordmark if (create != NULL) { 666bd41d0a8Snordmark t->zsd_flags = ZSD_CREATE_NEEDED; 667bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__needed, 668bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 669bd41d0a8Snordmark } 670bd41d0a8Snordmark list_insert_tail(&zone->zone_zsd, t); 671bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 672bd41d0a8Snordmark } 673bd41d0a8Snordmark mutex_exit(&zonehash_lock); 674bd41d0a8Snordmark 675bd41d0a8Snordmark if (create != NULL) { 676bd41d0a8Snordmark /* Now call the create callback for this key */ 677bd41d0a8Snordmark zsd_apply_all_zones(zsd_apply_create, key); 678bd41d0a8Snordmark } 679fe16170aSPramod Batni /* 680835ee219SRobert Harris * It is safe for consumers to use the key now, make it 681835ee219SRobert Harris * globally visible. Specifically zone_getspecific() will 682835ee219SRobert Harris * always successfully return the zone specific data associated 683835ee219SRobert Harris * with the key. 684835ee219SRobert Harris */ 685fe16170aSPramod Batni *keyp = key; 686fe16170aSPramod Batni 687bd41d0a8Snordmark } 688bd41d0a8Snordmark 6897c478bd9Sstevel@tonic-gate /* 6907c478bd9Sstevel@tonic-gate * Function called when a module is being unloaded, or otherwise wishes 6917c478bd9Sstevel@tonic-gate * to unregister its ZSD key and callbacks. 692bd41d0a8Snordmark * 693bd41d0a8Snordmark * Remove from the global list and determine the functions that need to 694bd41d0a8Snordmark * be called under a global lock. Then call the functions without 695bd41d0a8Snordmark * holding any locks. Finally free up the zone_zsd entries. (The apply 696bd41d0a8Snordmark * functions need to access the zone_zsd entries to find zsd_data etc.) 6977c478bd9Sstevel@tonic-gate */ 6987c478bd9Sstevel@tonic-gate int 6997c478bd9Sstevel@tonic-gate zone_key_delete(zone_key_t key) 7007c478bd9Sstevel@tonic-gate { 7017c478bd9Sstevel@tonic-gate struct zsd_entry *zsdp = NULL; 7027c478bd9Sstevel@tonic-gate zone_t *zone; 7037c478bd9Sstevel@tonic-gate 7047c478bd9Sstevel@tonic-gate mutex_enter(&zsd_key_lock); 705bd41d0a8Snordmark zsdp = zsd_find_mru(&zsd_registered_keys, key); 706bd41d0a8Snordmark if (zsdp == NULL) { 707bd41d0a8Snordmark mutex_exit(&zsd_key_lock); 708bd41d0a8Snordmark return (-1); 709bd41d0a8Snordmark } 7107c478bd9Sstevel@tonic-gate list_remove(&zsd_registered_keys, zsdp); 7117c478bd9Sstevel@tonic-gate mutex_exit(&zsd_key_lock); 7127c478bd9Sstevel@tonic-gate 713bd41d0a8Snordmark mutex_enter(&zonehash_lock); 7147c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 7157c478bd9Sstevel@tonic-gate zone = list_next(&zone_active, zone)) { 7167c478bd9Sstevel@tonic-gate struct zsd_entry *del; 717bd41d0a8Snordmark 718bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 719bd41d0a8Snordmark del = zsd_find_mru(&zone->zone_zsd, key); 720bd41d0a8Snordmark if (del == NULL) { 721bd41d0a8Snordmark /* 722bd41d0a8Snordmark * Somebody else got here first e.g the zone going 723bd41d0a8Snordmark * away. 724bd41d0a8Snordmark */ 725bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 726bd41d0a8Snordmark continue; 727bd41d0a8Snordmark } 728bd41d0a8Snordmark ASSERT(del->zsd_shutdown == zsdp->zsd_shutdown); 729bd41d0a8Snordmark ASSERT(del->zsd_destroy == zsdp->zsd_destroy); 730bd41d0a8Snordmark if (del->zsd_shutdown != NULL && 731bd41d0a8Snordmark (del->zsd_flags & ZSD_SHUTDOWN_ALL) == 0) { 732bd41d0a8Snordmark del->zsd_flags |= ZSD_SHUTDOWN_NEEDED; 733bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__needed, 734bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 735bd41d0a8Snordmark } 736bd41d0a8Snordmark if (del->zsd_destroy != NULL && 737bd41d0a8Snordmark (del->zsd_flags & ZSD_DESTROY_ALL) == 0) { 738bd41d0a8Snordmark del->zsd_flags |= ZSD_DESTROY_NEEDED; 739bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__needed, 740bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 7417c478bd9Sstevel@tonic-gate } 7427c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 7437c478bd9Sstevel@tonic-gate } 7447c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 7457c478bd9Sstevel@tonic-gate kmem_free(zsdp, sizeof (*zsdp)); 7467c478bd9Sstevel@tonic-gate 747bd41d0a8Snordmark /* Now call the shutdown and destroy callback for this key */ 748bd41d0a8Snordmark zsd_apply_all_zones(zsd_apply_shutdown, key); 749bd41d0a8Snordmark zsd_apply_all_zones(zsd_apply_destroy, key); 750bd41d0a8Snordmark 751bd41d0a8Snordmark /* Now we can free up the zsdp structures in each zone */ 752bd41d0a8Snordmark mutex_enter(&zonehash_lock); 7537c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 754bd41d0a8Snordmark zone = list_next(&zone_active, zone)) { 755bd41d0a8Snordmark struct zsd_entry *del; 756bd41d0a8Snordmark 757bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 758bd41d0a8Snordmark del = zsd_find(&zone->zone_zsd, key); 759bd41d0a8Snordmark if (del != NULL) { 760bd41d0a8Snordmark list_remove(&zone->zone_zsd, del); 761bd41d0a8Snordmark ASSERT(!(del->zsd_flags & ZSD_ALL_INPROGRESS)); 762bd41d0a8Snordmark kmem_free(del, sizeof (*del)); 763bd41d0a8Snordmark } 7647c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 765bd41d0a8Snordmark } 7667c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 767bd41d0a8Snordmark 768bd41d0a8Snordmark return (0); 7697c478bd9Sstevel@tonic-gate } 7707c478bd9Sstevel@tonic-gate 7717c478bd9Sstevel@tonic-gate /* 7727c478bd9Sstevel@tonic-gate * ZSD counterpart of pthread_setspecific(). 773bd41d0a8Snordmark * 774bd41d0a8Snordmark * Since all zsd callbacks, including those with no create function, 775bd41d0a8Snordmark * have an entry in zone_zsd, if the key is registered it is part of 776bd41d0a8Snordmark * the zone_zsd list. 777bd41d0a8Snordmark * Return an error if the key wasn't registerd. 7787c478bd9Sstevel@tonic-gate */ 7797c478bd9Sstevel@tonic-gate int 7807c478bd9Sstevel@tonic-gate zone_setspecific(zone_key_t key, zone_t *zone, const void *data) 7817c478bd9Sstevel@tonic-gate { 7827c478bd9Sstevel@tonic-gate struct zsd_entry *t; 7837c478bd9Sstevel@tonic-gate 7847c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 785bd41d0a8Snordmark t = zsd_find_mru(&zone->zone_zsd, key); 7867c478bd9Sstevel@tonic-gate if (t != NULL) { 7877c478bd9Sstevel@tonic-gate /* 7887c478bd9Sstevel@tonic-gate * Replace old value with new 7897c478bd9Sstevel@tonic-gate */ 7907c478bd9Sstevel@tonic-gate t->zsd_data = (void *)data; 7917c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 7927c478bd9Sstevel@tonic-gate return (0); 7937c478bd9Sstevel@tonic-gate } 7947c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 795bd41d0a8Snordmark return (-1); 7967c478bd9Sstevel@tonic-gate } 7977c478bd9Sstevel@tonic-gate 7987c478bd9Sstevel@tonic-gate /* 7997c478bd9Sstevel@tonic-gate * ZSD counterpart of pthread_getspecific(). 8007c478bd9Sstevel@tonic-gate */ 8017c478bd9Sstevel@tonic-gate void * 8027c478bd9Sstevel@tonic-gate zone_getspecific(zone_key_t key, zone_t *zone) 8037c478bd9Sstevel@tonic-gate { 8047c478bd9Sstevel@tonic-gate struct zsd_entry *t; 8057c478bd9Sstevel@tonic-gate void *data; 8067c478bd9Sstevel@tonic-gate 8077c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 808bd41d0a8Snordmark t = zsd_find_mru(&zone->zone_zsd, key); 8097c478bd9Sstevel@tonic-gate data = (t == NULL ? NULL : t->zsd_data); 8107c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 8117c478bd9Sstevel@tonic-gate return (data); 8127c478bd9Sstevel@tonic-gate } 8137c478bd9Sstevel@tonic-gate 8147c478bd9Sstevel@tonic-gate /* 8157c478bd9Sstevel@tonic-gate * Function used to initialize a zone's list of ZSD callbacks and data 8167c478bd9Sstevel@tonic-gate * when the zone is being created. The callbacks are initialized from 817bd41d0a8Snordmark * the template list (zsd_registered_keys). The constructor callback is 818bd41d0a8Snordmark * executed later (once the zone exists and with locks dropped). 8197c478bd9Sstevel@tonic-gate */ 8207c478bd9Sstevel@tonic-gate static void 8217c478bd9Sstevel@tonic-gate zone_zsd_configure(zone_t *zone) 8227c478bd9Sstevel@tonic-gate { 8237c478bd9Sstevel@tonic-gate struct zsd_entry *zsdp; 8247c478bd9Sstevel@tonic-gate struct zsd_entry *t; 8257c478bd9Sstevel@tonic-gate 8267c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 8277c478bd9Sstevel@tonic-gate ASSERT(list_head(&zone->zone_zsd) == NULL); 828bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 8297c478bd9Sstevel@tonic-gate mutex_enter(&zsd_key_lock); 8307c478bd9Sstevel@tonic-gate for (zsdp = list_head(&zsd_registered_keys); zsdp != NULL; 8317c478bd9Sstevel@tonic-gate zsdp = list_next(&zsd_registered_keys, zsdp)) { 832bd41d0a8Snordmark /* 833bd41d0a8Snordmark * Since this zone is ZONE_IS_UNCONFIGURED, zone_key_create 834bd41d0a8Snordmark * should not have added anything to it. 835bd41d0a8Snordmark */ 836bd41d0a8Snordmark ASSERT(zsd_find(&zone->zone_zsd, zsdp->zsd_key) == NULL); 837bd41d0a8Snordmark 838bd41d0a8Snordmark t = kmem_zalloc(sizeof (*t), KM_SLEEP); 839bd41d0a8Snordmark t->zsd_key = zsdp->zsd_key; 840bd41d0a8Snordmark t->zsd_create = zsdp->zsd_create; 841bd41d0a8Snordmark t->zsd_shutdown = zsdp->zsd_shutdown; 842bd41d0a8Snordmark t->zsd_destroy = zsdp->zsd_destroy; 8437c478bd9Sstevel@tonic-gate if (zsdp->zsd_create != NULL) { 844bd41d0a8Snordmark t->zsd_flags = ZSD_CREATE_NEEDED; 845bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__needed, 846bd41d0a8Snordmark zone_t *, zone, zone_key_t, zsdp->zsd_key); 8477c478bd9Sstevel@tonic-gate } 848bd41d0a8Snordmark list_insert_tail(&zone->zone_zsd, t); 8497c478bd9Sstevel@tonic-gate } 8507c478bd9Sstevel@tonic-gate mutex_exit(&zsd_key_lock); 851bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 8527c478bd9Sstevel@tonic-gate } 8537c478bd9Sstevel@tonic-gate 8547c478bd9Sstevel@tonic-gate enum zsd_callback_type { ZSD_CREATE, ZSD_SHUTDOWN, ZSD_DESTROY }; 8557c478bd9Sstevel@tonic-gate 8567c478bd9Sstevel@tonic-gate /* 8577c478bd9Sstevel@tonic-gate * Helper function to execute shutdown or destructor callbacks. 8587c478bd9Sstevel@tonic-gate */ 8597c478bd9Sstevel@tonic-gate static void 8607c478bd9Sstevel@tonic-gate zone_zsd_callbacks(zone_t *zone, enum zsd_callback_type ct) 8617c478bd9Sstevel@tonic-gate { 8627c478bd9Sstevel@tonic-gate struct zsd_entry *t; 8637c478bd9Sstevel@tonic-gate 8647c478bd9Sstevel@tonic-gate ASSERT(ct == ZSD_SHUTDOWN || ct == ZSD_DESTROY); 8657c478bd9Sstevel@tonic-gate ASSERT(ct != ZSD_SHUTDOWN || zone_status_get(zone) >= ZONE_IS_EMPTY); 8667c478bd9Sstevel@tonic-gate ASSERT(ct != ZSD_DESTROY || zone_status_get(zone) >= ZONE_IS_DOWN); 8677c478bd9Sstevel@tonic-gate 8687c478bd9Sstevel@tonic-gate /* 869bd41d0a8Snordmark * Run the callback solely based on what is registered for the zone 870bd41d0a8Snordmark * in zone_zsd. The global list can change independently of this 871bd41d0a8Snordmark * as keys are registered and unregistered and we don't register new 872bd41d0a8Snordmark * callbacks for a zone that is in the process of going away. 8737c478bd9Sstevel@tonic-gate */ 874bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 875bd41d0a8Snordmark for (t = list_head(&zone->zone_zsd); t != NULL; 876bd41d0a8Snordmark t = list_next(&zone->zone_zsd, t)) { 877bd41d0a8Snordmark zone_key_t key = t->zsd_key; 8787c478bd9Sstevel@tonic-gate 8797c478bd9Sstevel@tonic-gate /* Skip if no callbacks registered */ 880bd41d0a8Snordmark 881bd41d0a8Snordmark if (ct == ZSD_SHUTDOWN) { 882bd41d0a8Snordmark if (t->zsd_shutdown != NULL && 883bd41d0a8Snordmark (t->zsd_flags & ZSD_SHUTDOWN_ALL) == 0) { 884bd41d0a8Snordmark t->zsd_flags |= ZSD_SHUTDOWN_NEEDED; 885bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__needed, 886bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 8877c478bd9Sstevel@tonic-gate } 8887c478bd9Sstevel@tonic-gate } else { 889bd41d0a8Snordmark if (t->zsd_destroy != NULL && 890bd41d0a8Snordmark (t->zsd_flags & ZSD_DESTROY_ALL) == 0) { 891bd41d0a8Snordmark t->zsd_flags |= ZSD_DESTROY_NEEDED; 892bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__needed, 893bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 8947c478bd9Sstevel@tonic-gate } 8957c478bd9Sstevel@tonic-gate } 8967c478bd9Sstevel@tonic-gate } 897bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 898bd41d0a8Snordmark 899bd41d0a8Snordmark /* Now call the shutdown and destroy callback for this key */ 900bd41d0a8Snordmark zsd_apply_all_keys(zsd_apply_shutdown, zone); 901bd41d0a8Snordmark zsd_apply_all_keys(zsd_apply_destroy, zone); 902bd41d0a8Snordmark 9037c478bd9Sstevel@tonic-gate } 9047c478bd9Sstevel@tonic-gate 9057c478bd9Sstevel@tonic-gate /* 9067c478bd9Sstevel@tonic-gate * Called when the zone is going away; free ZSD-related memory, and 9077c478bd9Sstevel@tonic-gate * destroy the zone_zsd list. 9087c478bd9Sstevel@tonic-gate */ 9097c478bd9Sstevel@tonic-gate static void 9107c478bd9Sstevel@tonic-gate zone_free_zsd(zone_t *zone) 9117c478bd9Sstevel@tonic-gate { 9127c478bd9Sstevel@tonic-gate struct zsd_entry *t, *next; 9137c478bd9Sstevel@tonic-gate 9147c478bd9Sstevel@tonic-gate /* 9157c478bd9Sstevel@tonic-gate * Free all the zsd_entry's we had on this zone. 9167c478bd9Sstevel@tonic-gate */ 917bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 9187c478bd9Sstevel@tonic-gate for (t = list_head(&zone->zone_zsd); t != NULL; t = next) { 9197c478bd9Sstevel@tonic-gate next = list_next(&zone->zone_zsd, t); 9207c478bd9Sstevel@tonic-gate list_remove(&zone->zone_zsd, t); 921bd41d0a8Snordmark ASSERT(!(t->zsd_flags & ZSD_ALL_INPROGRESS)); 9227c478bd9Sstevel@tonic-gate kmem_free(t, sizeof (*t)); 9237c478bd9Sstevel@tonic-gate } 9247c478bd9Sstevel@tonic-gate list_destroy(&zone->zone_zsd); 925bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 926bd41d0a8Snordmark 927bd41d0a8Snordmark } 928bd41d0a8Snordmark 929bd41d0a8Snordmark /* 930bd41d0a8Snordmark * Apply a function to all zones for particular key value. 931bd41d0a8Snordmark * 932bd41d0a8Snordmark * The applyfn has to drop zonehash_lock if it does some work, and 933bd41d0a8Snordmark * then reacquire it before it returns. 934bd41d0a8Snordmark * When the lock is dropped we don't follow list_next even 935bd41d0a8Snordmark * if it is possible to do so without any hazards. This is 936bd41d0a8Snordmark * because we want the design to allow for the list of zones 937bd41d0a8Snordmark * to change in any arbitrary way during the time the 938bd41d0a8Snordmark * lock was dropped. 939bd41d0a8Snordmark * 940bd41d0a8Snordmark * It is safe to restart the loop at list_head since the applyfn 941bd41d0a8Snordmark * changes the zsd_flags as it does work, so a subsequent 942bd41d0a8Snordmark * pass through will have no effect in applyfn, hence the loop will terminate 943bd41d0a8Snordmark * in at worst O(N^2). 944bd41d0a8Snordmark */ 945bd41d0a8Snordmark static void 946bd41d0a8Snordmark zsd_apply_all_zones(zsd_applyfn_t *applyfn, zone_key_t key) 947bd41d0a8Snordmark { 948bd41d0a8Snordmark zone_t *zone; 949bd41d0a8Snordmark 950bd41d0a8Snordmark mutex_enter(&zonehash_lock); 951bd41d0a8Snordmark zone = list_head(&zone_active); 952bd41d0a8Snordmark while (zone != NULL) { 953bd41d0a8Snordmark if ((applyfn)(&zonehash_lock, B_FALSE, zone, key)) { 954bd41d0a8Snordmark /* Lock dropped - restart at head */ 955bd41d0a8Snordmark zone = list_head(&zone_active); 956bd41d0a8Snordmark } else { 957bd41d0a8Snordmark zone = list_next(&zone_active, zone); 958bd41d0a8Snordmark } 959bd41d0a8Snordmark } 960bd41d0a8Snordmark mutex_exit(&zonehash_lock); 961bd41d0a8Snordmark } 962bd41d0a8Snordmark 963bd41d0a8Snordmark /* 964bd41d0a8Snordmark * Apply a function to all keys for a particular zone. 965bd41d0a8Snordmark * 966bd41d0a8Snordmark * The applyfn has to drop zonehash_lock if it does some work, and 967bd41d0a8Snordmark * then reacquire it before it returns. 968bd41d0a8Snordmark * When the lock is dropped we don't follow list_next even 969bd41d0a8Snordmark * if it is possible to do so without any hazards. This is 970bd41d0a8Snordmark * because we want the design to allow for the list of zsd callbacks 971bd41d0a8Snordmark * to change in any arbitrary way during the time the 972bd41d0a8Snordmark * lock was dropped. 973bd41d0a8Snordmark * 974bd41d0a8Snordmark * It is safe to restart the loop at list_head since the applyfn 975bd41d0a8Snordmark * changes the zsd_flags as it does work, so a subsequent 976bd41d0a8Snordmark * pass through will have no effect in applyfn, hence the loop will terminate 977bd41d0a8Snordmark * in at worst O(N^2). 978bd41d0a8Snordmark */ 979bd41d0a8Snordmark static void 980bd41d0a8Snordmark zsd_apply_all_keys(zsd_applyfn_t *applyfn, zone_t *zone) 981bd41d0a8Snordmark { 982bd41d0a8Snordmark struct zsd_entry *t; 983bd41d0a8Snordmark 984bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 985bd41d0a8Snordmark t = list_head(&zone->zone_zsd); 986bd41d0a8Snordmark while (t != NULL) { 987bd41d0a8Snordmark if ((applyfn)(NULL, B_TRUE, zone, t->zsd_key)) { 988bd41d0a8Snordmark /* Lock dropped - restart at head */ 989bd41d0a8Snordmark t = list_head(&zone->zone_zsd); 990bd41d0a8Snordmark } else { 991bd41d0a8Snordmark t = list_next(&zone->zone_zsd, t); 992bd41d0a8Snordmark } 993bd41d0a8Snordmark } 994bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 995bd41d0a8Snordmark } 996bd41d0a8Snordmark 997bd41d0a8Snordmark /* 998bd41d0a8Snordmark * Call the create function for the zone and key if CREATE_NEEDED 999bd41d0a8Snordmark * is set. 1000bd41d0a8Snordmark * If some other thread gets here first and sets CREATE_INPROGRESS, then 1001bd41d0a8Snordmark * we wait for that thread to complete so that we can ensure that 1002bd41d0a8Snordmark * all the callbacks are done when we've looped over all zones/keys. 1003bd41d0a8Snordmark * 1004bd41d0a8Snordmark * When we call the create function, we drop the global held by the 1005bd41d0a8Snordmark * caller, and return true to tell the caller it needs to re-evalute the 1006bd41d0a8Snordmark * state. 1007bd41d0a8Snordmark * If the caller holds zone_lock then zone_lock_held is set, and zone_lock 1008bd41d0a8Snordmark * remains held on exit. 1009bd41d0a8Snordmark */ 1010bd41d0a8Snordmark static boolean_t 1011bd41d0a8Snordmark zsd_apply_create(kmutex_t *lockp, boolean_t zone_lock_held, 1012bd41d0a8Snordmark zone_t *zone, zone_key_t key) 1013bd41d0a8Snordmark { 1014bd41d0a8Snordmark void *result; 1015bd41d0a8Snordmark struct zsd_entry *t; 1016bd41d0a8Snordmark boolean_t dropped; 1017bd41d0a8Snordmark 1018bd41d0a8Snordmark if (lockp != NULL) { 1019bd41d0a8Snordmark ASSERT(MUTEX_HELD(lockp)); 1020bd41d0a8Snordmark } 1021bd41d0a8Snordmark if (zone_lock_held) { 1022bd41d0a8Snordmark ASSERT(MUTEX_HELD(&zone->zone_lock)); 1023bd41d0a8Snordmark } else { 1024bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1025bd41d0a8Snordmark } 1026bd41d0a8Snordmark 1027bd41d0a8Snordmark t = zsd_find(&zone->zone_zsd, key); 1028bd41d0a8Snordmark if (t == NULL) { 1029bd41d0a8Snordmark /* 1030bd41d0a8Snordmark * Somebody else got here first e.g the zone going 1031bd41d0a8Snordmark * away. 1032bd41d0a8Snordmark */ 1033bd41d0a8Snordmark if (!zone_lock_held) 1034bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1035bd41d0a8Snordmark return (B_FALSE); 1036bd41d0a8Snordmark } 1037bd41d0a8Snordmark dropped = B_FALSE; 1038bd41d0a8Snordmark if (zsd_wait_for_inprogress(zone, t, lockp)) 1039bd41d0a8Snordmark dropped = B_TRUE; 1040bd41d0a8Snordmark 1041bd41d0a8Snordmark if (t->zsd_flags & ZSD_CREATE_NEEDED) { 1042bd41d0a8Snordmark t->zsd_flags &= ~ZSD_CREATE_NEEDED; 1043bd41d0a8Snordmark t->zsd_flags |= ZSD_CREATE_INPROGRESS; 1044bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__inprogress, 1045bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1046bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1047bd41d0a8Snordmark if (lockp != NULL) 1048bd41d0a8Snordmark mutex_exit(lockp); 1049bd41d0a8Snordmark 1050bd41d0a8Snordmark dropped = B_TRUE; 1051bd41d0a8Snordmark ASSERT(t->zsd_create != NULL); 1052bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__start, 1053bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1054bd41d0a8Snordmark 1055bd41d0a8Snordmark result = (*t->zsd_create)(zone->zone_id); 1056bd41d0a8Snordmark 1057bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__end, 1058bd41d0a8Snordmark zone_t *, zone, voidn *, result); 1059bd41d0a8Snordmark 1060bd41d0a8Snordmark ASSERT(result != NULL); 1061bd41d0a8Snordmark if (lockp != NULL) 1062bd41d0a8Snordmark mutex_enter(lockp); 1063bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1064bd41d0a8Snordmark t->zsd_data = result; 1065bd41d0a8Snordmark t->zsd_flags &= ~ZSD_CREATE_INPROGRESS; 1066bd41d0a8Snordmark t->zsd_flags |= ZSD_CREATE_COMPLETED; 1067bd41d0a8Snordmark cv_broadcast(&t->zsd_cv); 1068bd41d0a8Snordmark DTRACE_PROBE2(zsd__create__completed, 1069bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1070bd41d0a8Snordmark } 1071bd41d0a8Snordmark if (!zone_lock_held) 1072bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1073bd41d0a8Snordmark return (dropped); 1074bd41d0a8Snordmark } 1075bd41d0a8Snordmark 1076bd41d0a8Snordmark /* 1077bd41d0a8Snordmark * Call the shutdown function for the zone and key if SHUTDOWN_NEEDED 1078bd41d0a8Snordmark * is set. 1079bd41d0a8Snordmark * If some other thread gets here first and sets *_INPROGRESS, then 1080bd41d0a8Snordmark * we wait for that thread to complete so that we can ensure that 1081bd41d0a8Snordmark * all the callbacks are done when we've looped over all zones/keys. 1082bd41d0a8Snordmark * 1083bd41d0a8Snordmark * When we call the shutdown function, we drop the global held by the 1084bd41d0a8Snordmark * caller, and return true to tell the caller it needs to re-evalute the 1085bd41d0a8Snordmark * state. 1086bd41d0a8Snordmark * If the caller holds zone_lock then zone_lock_held is set, and zone_lock 1087bd41d0a8Snordmark * remains held on exit. 1088bd41d0a8Snordmark */ 1089bd41d0a8Snordmark static boolean_t 1090bd41d0a8Snordmark zsd_apply_shutdown(kmutex_t *lockp, boolean_t zone_lock_held, 1091bd41d0a8Snordmark zone_t *zone, zone_key_t key) 1092bd41d0a8Snordmark { 1093bd41d0a8Snordmark struct zsd_entry *t; 1094bd41d0a8Snordmark void *data; 1095bd41d0a8Snordmark boolean_t dropped; 1096bd41d0a8Snordmark 1097bd41d0a8Snordmark if (lockp != NULL) { 1098bd41d0a8Snordmark ASSERT(MUTEX_HELD(lockp)); 1099bd41d0a8Snordmark } 1100bd41d0a8Snordmark if (zone_lock_held) { 1101bd41d0a8Snordmark ASSERT(MUTEX_HELD(&zone->zone_lock)); 1102bd41d0a8Snordmark } else { 1103bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1104bd41d0a8Snordmark } 1105bd41d0a8Snordmark 1106bd41d0a8Snordmark t = zsd_find(&zone->zone_zsd, key); 1107bd41d0a8Snordmark if (t == NULL) { 1108bd41d0a8Snordmark /* 1109bd41d0a8Snordmark * Somebody else got here first e.g the zone going 1110bd41d0a8Snordmark * away. 1111bd41d0a8Snordmark */ 1112bd41d0a8Snordmark if (!zone_lock_held) 1113bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1114bd41d0a8Snordmark return (B_FALSE); 1115bd41d0a8Snordmark } 1116bd41d0a8Snordmark dropped = B_FALSE; 1117bd41d0a8Snordmark if (zsd_wait_for_creator(zone, t, lockp)) 1118bd41d0a8Snordmark dropped = B_TRUE; 1119bd41d0a8Snordmark 1120bd41d0a8Snordmark if (zsd_wait_for_inprogress(zone, t, lockp)) 1121bd41d0a8Snordmark dropped = B_TRUE; 1122bd41d0a8Snordmark 1123bd41d0a8Snordmark if (t->zsd_flags & ZSD_SHUTDOWN_NEEDED) { 1124bd41d0a8Snordmark t->zsd_flags &= ~ZSD_SHUTDOWN_NEEDED; 1125bd41d0a8Snordmark t->zsd_flags |= ZSD_SHUTDOWN_INPROGRESS; 1126bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__inprogress, 1127bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1128bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1129bd41d0a8Snordmark if (lockp != NULL) 1130bd41d0a8Snordmark mutex_exit(lockp); 1131bd41d0a8Snordmark dropped = B_TRUE; 1132bd41d0a8Snordmark 1133bd41d0a8Snordmark ASSERT(t->zsd_shutdown != NULL); 1134bd41d0a8Snordmark data = t->zsd_data; 1135bd41d0a8Snordmark 1136bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__start, 1137bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1138bd41d0a8Snordmark 1139bd41d0a8Snordmark (t->zsd_shutdown)(zone->zone_id, data); 1140bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__end, 1141bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1142bd41d0a8Snordmark 1143bd41d0a8Snordmark if (lockp != NULL) 1144bd41d0a8Snordmark mutex_enter(lockp); 1145bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1146bd41d0a8Snordmark t->zsd_flags &= ~ZSD_SHUTDOWN_INPROGRESS; 1147bd41d0a8Snordmark t->zsd_flags |= ZSD_SHUTDOWN_COMPLETED; 1148bd41d0a8Snordmark cv_broadcast(&t->zsd_cv); 1149bd41d0a8Snordmark DTRACE_PROBE2(zsd__shutdown__completed, 1150bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1151bd41d0a8Snordmark } 1152bd41d0a8Snordmark if (!zone_lock_held) 1153bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1154bd41d0a8Snordmark return (dropped); 1155bd41d0a8Snordmark } 1156bd41d0a8Snordmark 1157bd41d0a8Snordmark /* 1158bd41d0a8Snordmark * Call the destroy function for the zone and key if DESTROY_NEEDED 1159bd41d0a8Snordmark * is set. 1160bd41d0a8Snordmark * If some other thread gets here first and sets *_INPROGRESS, then 1161bd41d0a8Snordmark * we wait for that thread to complete so that we can ensure that 1162bd41d0a8Snordmark * all the callbacks are done when we've looped over all zones/keys. 1163bd41d0a8Snordmark * 1164bd41d0a8Snordmark * When we call the destroy function, we drop the global held by the 1165bd41d0a8Snordmark * caller, and return true to tell the caller it needs to re-evalute the 1166bd41d0a8Snordmark * state. 1167bd41d0a8Snordmark * If the caller holds zone_lock then zone_lock_held is set, and zone_lock 1168bd41d0a8Snordmark * remains held on exit. 1169bd41d0a8Snordmark */ 1170bd41d0a8Snordmark static boolean_t 1171bd41d0a8Snordmark zsd_apply_destroy(kmutex_t *lockp, boolean_t zone_lock_held, 1172bd41d0a8Snordmark zone_t *zone, zone_key_t key) 1173bd41d0a8Snordmark { 1174bd41d0a8Snordmark struct zsd_entry *t; 1175bd41d0a8Snordmark void *data; 1176bd41d0a8Snordmark boolean_t dropped; 1177bd41d0a8Snordmark 1178bd41d0a8Snordmark if (lockp != NULL) { 1179bd41d0a8Snordmark ASSERT(MUTEX_HELD(lockp)); 1180bd41d0a8Snordmark } 1181bd41d0a8Snordmark if (zone_lock_held) { 1182bd41d0a8Snordmark ASSERT(MUTEX_HELD(&zone->zone_lock)); 1183bd41d0a8Snordmark } else { 1184bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1185bd41d0a8Snordmark } 1186bd41d0a8Snordmark 1187bd41d0a8Snordmark t = zsd_find(&zone->zone_zsd, key); 1188bd41d0a8Snordmark if (t == NULL) { 1189bd41d0a8Snordmark /* 1190bd41d0a8Snordmark * Somebody else got here first e.g the zone going 1191bd41d0a8Snordmark * away. 1192bd41d0a8Snordmark */ 1193bd41d0a8Snordmark if (!zone_lock_held) 1194bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1195bd41d0a8Snordmark return (B_FALSE); 1196bd41d0a8Snordmark } 1197bd41d0a8Snordmark dropped = B_FALSE; 1198bd41d0a8Snordmark if (zsd_wait_for_creator(zone, t, lockp)) 1199bd41d0a8Snordmark dropped = B_TRUE; 1200bd41d0a8Snordmark 1201bd41d0a8Snordmark if (zsd_wait_for_inprogress(zone, t, lockp)) 1202bd41d0a8Snordmark dropped = B_TRUE; 1203bd41d0a8Snordmark 1204bd41d0a8Snordmark if (t->zsd_flags & ZSD_DESTROY_NEEDED) { 1205bd41d0a8Snordmark t->zsd_flags &= ~ZSD_DESTROY_NEEDED; 1206bd41d0a8Snordmark t->zsd_flags |= ZSD_DESTROY_INPROGRESS; 1207bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__inprogress, 1208bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1209bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1210bd41d0a8Snordmark if (lockp != NULL) 1211bd41d0a8Snordmark mutex_exit(lockp); 1212bd41d0a8Snordmark dropped = B_TRUE; 1213bd41d0a8Snordmark 1214bd41d0a8Snordmark ASSERT(t->zsd_destroy != NULL); 1215bd41d0a8Snordmark data = t->zsd_data; 1216bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__start, 1217bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1218bd41d0a8Snordmark 1219bd41d0a8Snordmark (t->zsd_destroy)(zone->zone_id, data); 1220bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__end, 1221bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1222bd41d0a8Snordmark 1223bd41d0a8Snordmark if (lockp != NULL) 1224bd41d0a8Snordmark mutex_enter(lockp); 1225bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1226bd41d0a8Snordmark t->zsd_data = NULL; 1227bd41d0a8Snordmark t->zsd_flags &= ~ZSD_DESTROY_INPROGRESS; 1228bd41d0a8Snordmark t->zsd_flags |= ZSD_DESTROY_COMPLETED; 1229bd41d0a8Snordmark cv_broadcast(&t->zsd_cv); 1230bd41d0a8Snordmark DTRACE_PROBE2(zsd__destroy__completed, 1231bd41d0a8Snordmark zone_t *, zone, zone_key_t, key); 1232bd41d0a8Snordmark } 1233bd41d0a8Snordmark if (!zone_lock_held) 1234bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1235bd41d0a8Snordmark return (dropped); 1236bd41d0a8Snordmark } 1237bd41d0a8Snordmark 1238bd41d0a8Snordmark /* 1239bd41d0a8Snordmark * Wait for any CREATE_NEEDED flag to be cleared. 1240bd41d0a8Snordmark * Returns true if lockp was temporarily dropped while waiting. 1241bd41d0a8Snordmark */ 1242bd41d0a8Snordmark static boolean_t 1243bd41d0a8Snordmark zsd_wait_for_creator(zone_t *zone, struct zsd_entry *t, kmutex_t *lockp) 1244bd41d0a8Snordmark { 1245bd41d0a8Snordmark boolean_t dropped = B_FALSE; 1246bd41d0a8Snordmark 1247bd41d0a8Snordmark while (t->zsd_flags & ZSD_CREATE_NEEDED) { 1248bd41d0a8Snordmark DTRACE_PROBE2(zsd__wait__for__creator, 1249bd41d0a8Snordmark zone_t *, zone, struct zsd_entry *, t); 1250bd41d0a8Snordmark if (lockp != NULL) { 1251bd41d0a8Snordmark dropped = B_TRUE; 1252bd41d0a8Snordmark mutex_exit(lockp); 1253bd41d0a8Snordmark } 1254bd41d0a8Snordmark cv_wait(&t->zsd_cv, &zone->zone_lock); 1255bd41d0a8Snordmark if (lockp != NULL) { 1256bd41d0a8Snordmark /* First drop zone_lock to preserve order */ 1257bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1258bd41d0a8Snordmark mutex_enter(lockp); 1259bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1260bd41d0a8Snordmark } 1261bd41d0a8Snordmark } 1262bd41d0a8Snordmark return (dropped); 1263bd41d0a8Snordmark } 1264bd41d0a8Snordmark 1265bd41d0a8Snordmark /* 1266bd41d0a8Snordmark * Wait for any INPROGRESS flag to be cleared. 1267bd41d0a8Snordmark * Returns true if lockp was temporarily dropped while waiting. 1268bd41d0a8Snordmark */ 1269bd41d0a8Snordmark static boolean_t 1270bd41d0a8Snordmark zsd_wait_for_inprogress(zone_t *zone, struct zsd_entry *t, kmutex_t *lockp) 1271bd41d0a8Snordmark { 1272bd41d0a8Snordmark boolean_t dropped = B_FALSE; 1273bd41d0a8Snordmark 1274bd41d0a8Snordmark while (t->zsd_flags & ZSD_ALL_INPROGRESS) { 1275bd41d0a8Snordmark DTRACE_PROBE2(zsd__wait__for__inprogress, 1276bd41d0a8Snordmark zone_t *, zone, struct zsd_entry *, t); 1277bd41d0a8Snordmark if (lockp != NULL) { 1278bd41d0a8Snordmark dropped = B_TRUE; 1279bd41d0a8Snordmark mutex_exit(lockp); 1280bd41d0a8Snordmark } 1281bd41d0a8Snordmark cv_wait(&t->zsd_cv, &zone->zone_lock); 1282bd41d0a8Snordmark if (lockp != NULL) { 1283bd41d0a8Snordmark /* First drop zone_lock to preserve order */ 1284bd41d0a8Snordmark mutex_exit(&zone->zone_lock); 1285bd41d0a8Snordmark mutex_enter(lockp); 1286bd41d0a8Snordmark mutex_enter(&zone->zone_lock); 1287bd41d0a8Snordmark } 1288bd41d0a8Snordmark } 1289bd41d0a8Snordmark return (dropped); 12907c478bd9Sstevel@tonic-gate } 12917c478bd9Sstevel@tonic-gate 1292fa9e4066Sahrens /* 1293fa9e4066Sahrens * Frees memory associated with the zone dataset list. 1294fa9e4066Sahrens */ 1295fa9e4066Sahrens static void 1296fa9e4066Sahrens zone_free_datasets(zone_t *zone) 1297fa9e4066Sahrens { 1298fa9e4066Sahrens zone_dataset_t *t, *next; 1299fa9e4066Sahrens 1300fa9e4066Sahrens for (t = list_head(&zone->zone_datasets); t != NULL; t = next) { 1301fa9e4066Sahrens next = list_next(&zone->zone_datasets, t); 1302fa9e4066Sahrens list_remove(&zone->zone_datasets, t); 1303fa9e4066Sahrens kmem_free(t->zd_dataset, strlen(t->zd_dataset) + 1); 1304fa9e4066Sahrens kmem_free(t, sizeof (*t)); 1305fa9e4066Sahrens } 1306fa9e4066Sahrens list_destroy(&zone->zone_datasets); 1307fa9e4066Sahrens } 1308fa9e4066Sahrens 13097c478bd9Sstevel@tonic-gate /* 13107c478bd9Sstevel@tonic-gate * zone.cpu-shares resource control support. 13117c478bd9Sstevel@tonic-gate */ 13127c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 13137c478bd9Sstevel@tonic-gate static rctl_qty_t 13147c478bd9Sstevel@tonic-gate zone_cpu_shares_usage(rctl_t *rctl, struct proc *p) 13157c478bd9Sstevel@tonic-gate { 13167c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&p->p_lock)); 13177c478bd9Sstevel@tonic-gate return (p->p_zone->zone_shares); 13187c478bd9Sstevel@tonic-gate } 13197c478bd9Sstevel@tonic-gate 13207c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 13217c478bd9Sstevel@tonic-gate static int 13227c478bd9Sstevel@tonic-gate zone_cpu_shares_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, 13237c478bd9Sstevel@tonic-gate rctl_qty_t nv) 13247c478bd9Sstevel@tonic-gate { 13257c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&p->p_lock)); 13267c478bd9Sstevel@tonic-gate ASSERT(e->rcep_t == RCENTITY_ZONE); 13277c478bd9Sstevel@tonic-gate if (e->rcep_p.zone == NULL) 13287c478bd9Sstevel@tonic-gate return (0); 13297c478bd9Sstevel@tonic-gate 13307c478bd9Sstevel@tonic-gate e->rcep_p.zone->zone_shares = nv; 13317c478bd9Sstevel@tonic-gate return (0); 13327c478bd9Sstevel@tonic-gate } 13337c478bd9Sstevel@tonic-gate 13347c478bd9Sstevel@tonic-gate static rctl_ops_t zone_cpu_shares_ops = { 13357c478bd9Sstevel@tonic-gate rcop_no_action, 13367c478bd9Sstevel@tonic-gate zone_cpu_shares_usage, 13377c478bd9Sstevel@tonic-gate zone_cpu_shares_set, 13387c478bd9Sstevel@tonic-gate rcop_no_test 13397c478bd9Sstevel@tonic-gate }; 13407c478bd9Sstevel@tonic-gate 1341c97ad5cdSakolb /* 1342c97ad5cdSakolb * zone.cpu-cap resource control support. 1343c97ad5cdSakolb */ 1344c97ad5cdSakolb /*ARGSUSED*/ 1345c97ad5cdSakolb static rctl_qty_t 1346c97ad5cdSakolb zone_cpu_cap_get(rctl_t *rctl, struct proc *p) 1347c97ad5cdSakolb { 1348c97ad5cdSakolb ASSERT(MUTEX_HELD(&p->p_lock)); 1349c97ad5cdSakolb return (cpucaps_zone_get(p->p_zone)); 1350c97ad5cdSakolb } 1351c97ad5cdSakolb 1352c97ad5cdSakolb /*ARGSUSED*/ 1353c97ad5cdSakolb static int 1354c97ad5cdSakolb zone_cpu_cap_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, 1355c97ad5cdSakolb rctl_qty_t nv) 1356c97ad5cdSakolb { 1357c97ad5cdSakolb zone_t *zone = e->rcep_p.zone; 1358c97ad5cdSakolb 1359c97ad5cdSakolb ASSERT(MUTEX_HELD(&p->p_lock)); 1360c97ad5cdSakolb ASSERT(e->rcep_t == RCENTITY_ZONE); 1361c97ad5cdSakolb 1362c97ad5cdSakolb if (zone == NULL) 1363c97ad5cdSakolb return (0); 1364c97ad5cdSakolb 1365c97ad5cdSakolb /* 1366c97ad5cdSakolb * set cap to the new value. 1367c97ad5cdSakolb */ 1368c97ad5cdSakolb return (cpucaps_zone_set(zone, nv)); 1369c97ad5cdSakolb } 1370c97ad5cdSakolb 1371c97ad5cdSakolb static rctl_ops_t zone_cpu_cap_ops = { 1372c97ad5cdSakolb rcop_no_action, 1373c97ad5cdSakolb zone_cpu_cap_get, 1374c97ad5cdSakolb zone_cpu_cap_set, 1375c97ad5cdSakolb rcop_no_test 1376c97ad5cdSakolb }; 1377c97ad5cdSakolb 13787c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 13797c478bd9Sstevel@tonic-gate static rctl_qty_t 13807c478bd9Sstevel@tonic-gate zone_lwps_usage(rctl_t *r, proc_t *p) 13817c478bd9Sstevel@tonic-gate { 13827c478bd9Sstevel@tonic-gate rctl_qty_t nlwps; 13837c478bd9Sstevel@tonic-gate zone_t *zone = p->p_zone; 13847c478bd9Sstevel@tonic-gate 13857c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&p->p_lock)); 13867c478bd9Sstevel@tonic-gate 13877c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_nlwps_lock); 13887c478bd9Sstevel@tonic-gate nlwps = zone->zone_nlwps; 13897c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_nlwps_lock); 13907c478bd9Sstevel@tonic-gate 13917c478bd9Sstevel@tonic-gate return (nlwps); 13927c478bd9Sstevel@tonic-gate } 13937c478bd9Sstevel@tonic-gate 13947c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 13957c478bd9Sstevel@tonic-gate static int 13967c478bd9Sstevel@tonic-gate zone_lwps_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rcntl, 13977c478bd9Sstevel@tonic-gate rctl_qty_t incr, uint_t flags) 13987c478bd9Sstevel@tonic-gate { 13997c478bd9Sstevel@tonic-gate rctl_qty_t nlwps; 14007c478bd9Sstevel@tonic-gate 14017c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&p->p_lock)); 14027c478bd9Sstevel@tonic-gate ASSERT(e->rcep_t == RCENTITY_ZONE); 14037c478bd9Sstevel@tonic-gate if (e->rcep_p.zone == NULL) 14047c478bd9Sstevel@tonic-gate return (0); 14057c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&(e->rcep_p.zone->zone_nlwps_lock))); 14067c478bd9Sstevel@tonic-gate nlwps = e->rcep_p.zone->zone_nlwps; 14077c478bd9Sstevel@tonic-gate 14087c478bd9Sstevel@tonic-gate if (nlwps + incr > rcntl->rcv_value) 14097c478bd9Sstevel@tonic-gate return (1); 14107c478bd9Sstevel@tonic-gate 14117c478bd9Sstevel@tonic-gate return (0); 14127c478bd9Sstevel@tonic-gate } 14137c478bd9Sstevel@tonic-gate 14147c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 14157c478bd9Sstevel@tonic-gate static int 1416c6939658Ssl zone_lwps_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, rctl_qty_t nv) 1417c6939658Ssl { 14187c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&p->p_lock)); 14197c478bd9Sstevel@tonic-gate ASSERT(e->rcep_t == RCENTITY_ZONE); 14207c478bd9Sstevel@tonic-gate if (e->rcep_p.zone == NULL) 14217c478bd9Sstevel@tonic-gate return (0); 14227c478bd9Sstevel@tonic-gate e->rcep_p.zone->zone_nlwps_ctl = nv; 14237c478bd9Sstevel@tonic-gate return (0); 14247c478bd9Sstevel@tonic-gate } 14257c478bd9Sstevel@tonic-gate 14267c478bd9Sstevel@tonic-gate static rctl_ops_t zone_lwps_ops = { 14277c478bd9Sstevel@tonic-gate rcop_no_action, 14287c478bd9Sstevel@tonic-gate zone_lwps_usage, 14297c478bd9Sstevel@tonic-gate zone_lwps_set, 14307c478bd9Sstevel@tonic-gate zone_lwps_test, 14317c478bd9Sstevel@tonic-gate }; 14327c478bd9Sstevel@tonic-gate 1433ff19e029SMenno Lageman /*ARGSUSED*/ 1434ff19e029SMenno Lageman static rctl_qty_t 1435ff19e029SMenno Lageman zone_procs_usage(rctl_t *r, proc_t *p) 1436ff19e029SMenno Lageman { 1437ff19e029SMenno Lageman rctl_qty_t nprocs; 1438ff19e029SMenno Lageman zone_t *zone = p->p_zone; 1439ff19e029SMenno Lageman 1440ff19e029SMenno Lageman ASSERT(MUTEX_HELD(&p->p_lock)); 1441ff19e029SMenno Lageman 1442ff19e029SMenno Lageman mutex_enter(&zone->zone_nlwps_lock); 1443ff19e029SMenno Lageman nprocs = zone->zone_nprocs; 1444ff19e029SMenno Lageman mutex_exit(&zone->zone_nlwps_lock); 1445ff19e029SMenno Lageman 1446ff19e029SMenno Lageman return (nprocs); 1447ff19e029SMenno Lageman } 1448ff19e029SMenno Lageman 1449ff19e029SMenno Lageman /*ARGSUSED*/ 1450ff19e029SMenno Lageman static int 1451ff19e029SMenno Lageman zone_procs_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rcntl, 1452ff19e029SMenno Lageman rctl_qty_t incr, uint_t flags) 1453ff19e029SMenno Lageman { 1454ff19e029SMenno Lageman rctl_qty_t nprocs; 1455ff19e029SMenno Lageman 1456ff19e029SMenno Lageman ASSERT(MUTEX_HELD(&p->p_lock)); 1457ff19e029SMenno Lageman ASSERT(e->rcep_t == RCENTITY_ZONE); 1458ff19e029SMenno Lageman if (e->rcep_p.zone == NULL) 1459ff19e029SMenno Lageman return (0); 1460ff19e029SMenno Lageman ASSERT(MUTEX_HELD(&(e->rcep_p.zone->zone_nlwps_lock))); 1461ff19e029SMenno Lageman nprocs = e->rcep_p.zone->zone_nprocs; 1462ff19e029SMenno Lageman 1463ff19e029SMenno Lageman if (nprocs + incr > rcntl->rcv_value) 1464ff19e029SMenno Lageman return (1); 1465ff19e029SMenno Lageman 1466ff19e029SMenno Lageman return (0); 1467ff19e029SMenno Lageman } 1468ff19e029SMenno Lageman 1469ff19e029SMenno Lageman /*ARGSUSED*/ 1470ff19e029SMenno Lageman static int 1471ff19e029SMenno Lageman zone_procs_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, rctl_qty_t nv) 1472ff19e029SMenno Lageman { 1473ff19e029SMenno Lageman ASSERT(MUTEX_HELD(&p->p_lock)); 1474ff19e029SMenno Lageman ASSERT(e->rcep_t == RCENTITY_ZONE); 1475ff19e029SMenno Lageman if (e->rcep_p.zone == NULL) 1476ff19e029SMenno Lageman return (0); 1477ff19e029SMenno Lageman e->rcep_p.zone->zone_nprocs_ctl = nv; 1478ff19e029SMenno Lageman return (0); 1479ff19e029SMenno Lageman } 1480ff19e029SMenno Lageman 1481ff19e029SMenno Lageman static rctl_ops_t zone_procs_ops = { 1482ff19e029SMenno Lageman rcop_no_action, 1483ff19e029SMenno Lageman zone_procs_usage, 1484ff19e029SMenno Lageman zone_procs_set, 1485ff19e029SMenno Lageman zone_procs_test, 1486ff19e029SMenno Lageman }; 1487ff19e029SMenno Lageman 1488824c205fSml /*ARGSUSED*/ 1489824c205fSml static int 1490824c205fSml zone_shmmax_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rval, 1491824c205fSml rctl_qty_t incr, uint_t flags) 1492824c205fSml { 1493824c205fSml rctl_qty_t v; 1494824c205fSml ASSERT(MUTEX_HELD(&p->p_lock)); 1495824c205fSml ASSERT(e->rcep_t == RCENTITY_ZONE); 1496824c205fSml v = e->rcep_p.zone->zone_shmmax + incr; 1497824c205fSml if (v > rval->rcv_value) 1498824c205fSml return (1); 1499824c205fSml return (0); 1500824c205fSml } 1501824c205fSml 1502824c205fSml static rctl_ops_t zone_shmmax_ops = { 1503824c205fSml rcop_no_action, 1504824c205fSml rcop_no_usage, 1505824c205fSml rcop_no_set, 1506824c205fSml zone_shmmax_test 1507824c205fSml }; 1508824c205fSml 1509824c205fSml /*ARGSUSED*/ 1510824c205fSml static int 1511824c205fSml zone_shmmni_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rval, 1512824c205fSml rctl_qty_t incr, uint_t flags) 1513824c205fSml { 1514824c205fSml rctl_qty_t v; 1515824c205fSml ASSERT(MUTEX_HELD(&p->p_lock)); 1516824c205fSml ASSERT(e->rcep_t == RCENTITY_ZONE); 1517824c205fSml v = e->rcep_p.zone->zone_ipc.ipcq_shmmni + incr; 1518824c205fSml if (v > rval->rcv_value) 1519824c205fSml return (1); 1520824c205fSml return (0); 1521824c205fSml } 1522824c205fSml 1523824c205fSml static rctl_ops_t zone_shmmni_ops = { 1524824c205fSml rcop_no_action, 1525824c205fSml rcop_no_usage, 1526824c205fSml rcop_no_set, 1527824c205fSml zone_shmmni_test 1528824c205fSml }; 1529824c205fSml 1530824c205fSml /*ARGSUSED*/ 1531824c205fSml static int 1532824c205fSml zone_semmni_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rval, 1533824c205fSml rctl_qty_t incr, uint_t flags) 1534824c205fSml { 1535824c205fSml rctl_qty_t v; 1536824c205fSml ASSERT(MUTEX_HELD(&p->p_lock)); 1537824c205fSml ASSERT(e->rcep_t == RCENTITY_ZONE); 1538824c205fSml v = e->rcep_p.zone->zone_ipc.ipcq_semmni + incr; 1539824c205fSml if (v > rval->rcv_value) 1540824c205fSml return (1); 1541824c205fSml return (0); 1542824c205fSml } 1543824c205fSml 1544824c205fSml static rctl_ops_t zone_semmni_ops = { 1545824c205fSml rcop_no_action, 1546824c205fSml rcop_no_usage, 1547824c205fSml rcop_no_set, 1548824c205fSml zone_semmni_test 1549824c205fSml }; 1550824c205fSml 1551824c205fSml /*ARGSUSED*/ 1552824c205fSml static int 1553824c205fSml zone_msgmni_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, rctl_val_t *rval, 1554824c205fSml rctl_qty_t incr, uint_t flags) 1555824c205fSml { 1556824c205fSml rctl_qty_t v; 1557824c205fSml ASSERT(MUTEX_HELD(&p->p_lock)); 1558824c205fSml ASSERT(e->rcep_t == RCENTITY_ZONE); 1559824c205fSml v = e->rcep_p.zone->zone_ipc.ipcq_msgmni + incr; 1560824c205fSml if (v > rval->rcv_value) 1561824c205fSml return (1); 1562824c205fSml return (0); 1563824c205fSml } 1564824c205fSml 1565824c205fSml static rctl_ops_t zone_msgmni_ops = { 1566824c205fSml rcop_no_action, 1567824c205fSml rcop_no_usage, 1568824c205fSml rcop_no_set, 1569824c205fSml zone_msgmni_test 1570824c205fSml }; 1571824c205fSml 1572c6939658Ssl /*ARGSUSED*/ 1573c6939658Ssl static rctl_qty_t 1574c6939658Ssl zone_locked_mem_usage(rctl_t *rctl, struct proc *p) 1575c6939658Ssl { 1576c6939658Ssl rctl_qty_t q; 1577c6939658Ssl ASSERT(MUTEX_HELD(&p->p_lock)); 15780209230bSgjelinek mutex_enter(&p->p_zone->zone_mem_lock); 1579c6939658Ssl q = p->p_zone->zone_locked_mem; 15800209230bSgjelinek mutex_exit(&p->p_zone->zone_mem_lock); 1581c6939658Ssl return (q); 1582c6939658Ssl } 1583c6939658Ssl 1584c6939658Ssl /*ARGSUSED*/ 1585c6939658Ssl static int 1586c6939658Ssl zone_locked_mem_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, 1587c6939658Ssl rctl_val_t *rcntl, rctl_qty_t incr, uint_t flags) 1588c6939658Ssl { 1589c6939658Ssl rctl_qty_t q; 15900209230bSgjelinek zone_t *z; 15910209230bSgjelinek 15920209230bSgjelinek z = e->rcep_p.zone; 1593c6939658Ssl ASSERT(MUTEX_HELD(&p->p_lock)); 15940209230bSgjelinek ASSERT(MUTEX_HELD(&z->zone_mem_lock)); 15950209230bSgjelinek q = z->zone_locked_mem; 1596c6939658Ssl if (q + incr > rcntl->rcv_value) 1597c6939658Ssl return (1); 1598c6939658Ssl return (0); 1599c6939658Ssl } 1600c6939658Ssl 1601c6939658Ssl /*ARGSUSED*/ 1602c6939658Ssl static int 1603c6939658Ssl zone_locked_mem_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, 1604c6939658Ssl rctl_qty_t nv) 1605c6939658Ssl { 1606c6939658Ssl ASSERT(MUTEX_HELD(&p->p_lock)); 1607c6939658Ssl ASSERT(e->rcep_t == RCENTITY_ZONE); 1608c6939658Ssl if (e->rcep_p.zone == NULL) 1609c6939658Ssl return (0); 1610c6939658Ssl e->rcep_p.zone->zone_locked_mem_ctl = nv; 1611c6939658Ssl return (0); 1612c6939658Ssl } 1613c6939658Ssl 1614c6939658Ssl static rctl_ops_t zone_locked_mem_ops = { 1615c6939658Ssl rcop_no_action, 1616c6939658Ssl zone_locked_mem_usage, 1617c6939658Ssl zone_locked_mem_set, 1618c6939658Ssl zone_locked_mem_test 1619c6939658Ssl }; 1620824c205fSml 16210209230bSgjelinek /*ARGSUSED*/ 16220209230bSgjelinek static rctl_qty_t 16230209230bSgjelinek zone_max_swap_usage(rctl_t *rctl, struct proc *p) 16240209230bSgjelinek { 16250209230bSgjelinek rctl_qty_t q; 16260209230bSgjelinek zone_t *z = p->p_zone; 16270209230bSgjelinek 16280209230bSgjelinek ASSERT(MUTEX_HELD(&p->p_lock)); 16290209230bSgjelinek mutex_enter(&z->zone_mem_lock); 16300209230bSgjelinek q = z->zone_max_swap; 16310209230bSgjelinek mutex_exit(&z->zone_mem_lock); 16320209230bSgjelinek return (q); 16330209230bSgjelinek } 16340209230bSgjelinek 16350209230bSgjelinek /*ARGSUSED*/ 16360209230bSgjelinek static int 16370209230bSgjelinek zone_max_swap_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, 16380209230bSgjelinek rctl_val_t *rcntl, rctl_qty_t incr, uint_t flags) 16390209230bSgjelinek { 16400209230bSgjelinek rctl_qty_t q; 16410209230bSgjelinek zone_t *z; 16420209230bSgjelinek 16430209230bSgjelinek z = e->rcep_p.zone; 16440209230bSgjelinek ASSERT(MUTEX_HELD(&p->p_lock)); 16450209230bSgjelinek ASSERT(MUTEX_HELD(&z->zone_mem_lock)); 16460209230bSgjelinek q = z->zone_max_swap; 16470209230bSgjelinek if (q + incr > rcntl->rcv_value) 16480209230bSgjelinek return (1); 16490209230bSgjelinek return (0); 16500209230bSgjelinek } 16510209230bSgjelinek 16520209230bSgjelinek /*ARGSUSED*/ 16530209230bSgjelinek static int 16540209230bSgjelinek zone_max_swap_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, 16550209230bSgjelinek rctl_qty_t nv) 16560209230bSgjelinek { 16570209230bSgjelinek ASSERT(MUTEX_HELD(&p->p_lock)); 16580209230bSgjelinek ASSERT(e->rcep_t == RCENTITY_ZONE); 16590209230bSgjelinek if (e->rcep_p.zone == NULL) 16600209230bSgjelinek return (0); 16610209230bSgjelinek e->rcep_p.zone->zone_max_swap_ctl = nv; 16620209230bSgjelinek return (0); 16630209230bSgjelinek } 16640209230bSgjelinek 16650209230bSgjelinek static rctl_ops_t zone_max_swap_ops = { 16660209230bSgjelinek rcop_no_action, 16670209230bSgjelinek zone_max_swap_usage, 16680209230bSgjelinek zone_max_swap_set, 16690209230bSgjelinek zone_max_swap_test 16700209230bSgjelinek }; 16710209230bSgjelinek 16720fbb751dSJohn Levon /*ARGSUSED*/ 16730fbb751dSJohn Levon static rctl_qty_t 16740fbb751dSJohn Levon zone_max_lofi_usage(rctl_t *rctl, struct proc *p) 16750fbb751dSJohn Levon { 16760fbb751dSJohn Levon rctl_qty_t q; 16770fbb751dSJohn Levon zone_t *z = p->p_zone; 16780fbb751dSJohn Levon 16790fbb751dSJohn Levon ASSERT(MUTEX_HELD(&p->p_lock)); 16800fbb751dSJohn Levon mutex_enter(&z->zone_rctl_lock); 16810fbb751dSJohn Levon q = z->zone_max_lofi; 16820fbb751dSJohn Levon mutex_exit(&z->zone_rctl_lock); 16830fbb751dSJohn Levon return (q); 16840fbb751dSJohn Levon } 16850fbb751dSJohn Levon 16860fbb751dSJohn Levon /*ARGSUSED*/ 16870fbb751dSJohn Levon static int 16880fbb751dSJohn Levon zone_max_lofi_test(rctl_t *r, proc_t *p, rctl_entity_p_t *e, 16890fbb751dSJohn Levon rctl_val_t *rcntl, rctl_qty_t incr, uint_t flags) 16900fbb751dSJohn Levon { 16910fbb751dSJohn Levon rctl_qty_t q; 16920fbb751dSJohn Levon zone_t *z; 16930fbb751dSJohn Levon 16940fbb751dSJohn Levon z = e->rcep_p.zone; 16950fbb751dSJohn Levon ASSERT(MUTEX_HELD(&p->p_lock)); 16960fbb751dSJohn Levon ASSERT(MUTEX_HELD(&z->zone_rctl_lock)); 16970fbb751dSJohn Levon q = z->zone_max_lofi; 16980fbb751dSJohn Levon if (q + incr > rcntl->rcv_value) 16990fbb751dSJohn Levon return (1); 17000fbb751dSJohn Levon return (0); 17010fbb751dSJohn Levon } 17020fbb751dSJohn Levon 17030fbb751dSJohn Levon /*ARGSUSED*/ 17040fbb751dSJohn Levon static int 17050fbb751dSJohn Levon zone_max_lofi_set(rctl_t *rctl, struct proc *p, rctl_entity_p_t *e, 17060fbb751dSJohn Levon rctl_qty_t nv) 17070fbb751dSJohn Levon { 17080fbb751dSJohn Levon ASSERT(MUTEX_HELD(&p->p_lock)); 17090fbb751dSJohn Levon ASSERT(e->rcep_t == RCENTITY_ZONE); 17100fbb751dSJohn Levon if (e->rcep_p.zone == NULL) 17110fbb751dSJohn Levon return (0); 17120fbb751dSJohn Levon e->rcep_p.zone->zone_max_lofi_ctl = nv; 17130fbb751dSJohn Levon return (0); 17140fbb751dSJohn Levon } 17150fbb751dSJohn Levon 17160fbb751dSJohn Levon static rctl_ops_t zone_max_lofi_ops = { 17170fbb751dSJohn Levon rcop_no_action, 17180fbb751dSJohn Levon zone_max_lofi_usage, 17190fbb751dSJohn Levon zone_max_lofi_set, 17200fbb751dSJohn Levon zone_max_lofi_test 17210fbb751dSJohn Levon }; 17220fbb751dSJohn Levon 17237c478bd9Sstevel@tonic-gate /* 17247c478bd9Sstevel@tonic-gate * Helper function to brand the zone with a unique ID. 17257c478bd9Sstevel@tonic-gate */ 17267c478bd9Sstevel@tonic-gate static void 17277c478bd9Sstevel@tonic-gate zone_uniqid(zone_t *zone) 17287c478bd9Sstevel@tonic-gate { 17297c478bd9Sstevel@tonic-gate static uint64_t uniqid = 0; 17307c478bd9Sstevel@tonic-gate 17317c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 17327c478bd9Sstevel@tonic-gate zone->zone_uniqid = uniqid++; 17337c478bd9Sstevel@tonic-gate } 17347c478bd9Sstevel@tonic-gate 17357c478bd9Sstevel@tonic-gate /* 17367c478bd9Sstevel@tonic-gate * Returns a held pointer to the "kcred" for the specified zone. 17377c478bd9Sstevel@tonic-gate */ 17387c478bd9Sstevel@tonic-gate struct cred * 17397c478bd9Sstevel@tonic-gate zone_get_kcred(zoneid_t zoneid) 17407c478bd9Sstevel@tonic-gate { 17417c478bd9Sstevel@tonic-gate zone_t *zone; 17427c478bd9Sstevel@tonic-gate cred_t *cr; 17437c478bd9Sstevel@tonic-gate 17447c478bd9Sstevel@tonic-gate if ((zone = zone_find_by_id(zoneid)) == NULL) 17457c478bd9Sstevel@tonic-gate return (NULL); 17467c478bd9Sstevel@tonic-gate cr = zone->zone_kcred; 17477c478bd9Sstevel@tonic-gate crhold(cr); 17487c478bd9Sstevel@tonic-gate zone_rele(zone); 17497c478bd9Sstevel@tonic-gate return (cr); 17507c478bd9Sstevel@tonic-gate } 17517c478bd9Sstevel@tonic-gate 17520209230bSgjelinek static int 17530209230bSgjelinek zone_lockedmem_kstat_update(kstat_t *ksp, int rw) 17540209230bSgjelinek { 17550209230bSgjelinek zone_t *zone = ksp->ks_private; 17560209230bSgjelinek zone_kstat_t *zk = ksp->ks_data; 17570209230bSgjelinek 17580209230bSgjelinek if (rw == KSTAT_WRITE) 17590209230bSgjelinek return (EACCES); 17600209230bSgjelinek 17610209230bSgjelinek zk->zk_usage.value.ui64 = zone->zone_locked_mem; 17620209230bSgjelinek zk->zk_value.value.ui64 = zone->zone_locked_mem_ctl; 17630209230bSgjelinek return (0); 17640209230bSgjelinek } 17650209230bSgjelinek 1766ff19e029SMenno Lageman static int 1767ff19e029SMenno Lageman zone_nprocs_kstat_update(kstat_t *ksp, int rw) 1768ff19e029SMenno Lageman { 1769ff19e029SMenno Lageman zone_t *zone = ksp->ks_private; 1770ff19e029SMenno Lageman zone_kstat_t *zk = ksp->ks_data; 1771ff19e029SMenno Lageman 1772ff19e029SMenno Lageman if (rw == KSTAT_WRITE) 1773ff19e029SMenno Lageman return (EACCES); 1774ff19e029SMenno Lageman 1775ff19e029SMenno Lageman zk->zk_usage.value.ui64 = zone->zone_nprocs; 1776ff19e029SMenno Lageman zk->zk_value.value.ui64 = zone->zone_nprocs_ctl; 1777ff19e029SMenno Lageman return (0); 1778ff19e029SMenno Lageman } 1779ff19e029SMenno Lageman 17800209230bSgjelinek static int 17810209230bSgjelinek zone_swapresv_kstat_update(kstat_t *ksp, int rw) 17820209230bSgjelinek { 17830209230bSgjelinek zone_t *zone = ksp->ks_private; 17840209230bSgjelinek zone_kstat_t *zk = ksp->ks_data; 17850209230bSgjelinek 17860209230bSgjelinek if (rw == KSTAT_WRITE) 17870209230bSgjelinek return (EACCES); 17880209230bSgjelinek 17890209230bSgjelinek zk->zk_usage.value.ui64 = zone->zone_max_swap; 17900209230bSgjelinek zk->zk_value.value.ui64 = zone->zone_max_swap_ctl; 17910209230bSgjelinek return (0); 17920209230bSgjelinek } 17930209230bSgjelinek 1794ff19e029SMenno Lageman static kstat_t * 1795ff19e029SMenno Lageman zone_kstat_create_common(zone_t *zone, char *name, 1796ff19e029SMenno Lageman int (*updatefunc) (kstat_t *, int)) 17970209230bSgjelinek { 17980209230bSgjelinek kstat_t *ksp; 17990209230bSgjelinek zone_kstat_t *zk; 18000209230bSgjelinek 1801ff19e029SMenno Lageman ksp = rctl_kstat_create_zone(zone, name, KSTAT_TYPE_NAMED, 18020209230bSgjelinek sizeof (zone_kstat_t) / sizeof (kstat_named_t), 18030209230bSgjelinek KSTAT_FLAG_VIRTUAL); 18040209230bSgjelinek 18050209230bSgjelinek if (ksp == NULL) 1806ff19e029SMenno Lageman return (NULL); 18070209230bSgjelinek 18080209230bSgjelinek zk = ksp->ks_data = kmem_alloc(sizeof (zone_kstat_t), KM_SLEEP); 18090209230bSgjelinek ksp->ks_data_size += strlen(zone->zone_name) + 1; 18100209230bSgjelinek kstat_named_init(&zk->zk_zonename, "zonename", KSTAT_DATA_STRING); 18110209230bSgjelinek kstat_named_setstr(&zk->zk_zonename, zone->zone_name); 18120209230bSgjelinek kstat_named_init(&zk->zk_usage, "usage", KSTAT_DATA_UINT64); 18130209230bSgjelinek kstat_named_init(&zk->zk_value, "value", KSTAT_DATA_UINT64); 1814ff19e029SMenno Lageman ksp->ks_update = updatefunc; 18150209230bSgjelinek ksp->ks_private = zone; 18160209230bSgjelinek kstat_install(ksp); 1817ff19e029SMenno Lageman return (ksp); 1818ff19e029SMenno Lageman } 18190209230bSgjelinek 1820542a813cSJerry Jelinek static int 1821542a813cSJerry Jelinek zone_misc_kstat_update(kstat_t *ksp, int rw) 1822542a813cSJerry Jelinek { 1823542a813cSJerry Jelinek zone_t *zone = ksp->ks_private; 1824542a813cSJerry Jelinek zone_misc_kstat_t *zmp = ksp->ks_data; 1825542a813cSJerry Jelinek hrtime_t tmp; 1826542a813cSJerry Jelinek 1827542a813cSJerry Jelinek if (rw == KSTAT_WRITE) 1828542a813cSJerry Jelinek return (EACCES); 1829542a813cSJerry Jelinek 1830542a813cSJerry Jelinek tmp = zone->zone_utime; 1831542a813cSJerry Jelinek scalehrtime(&tmp); 1832542a813cSJerry Jelinek zmp->zm_utime.value.ui64 = tmp; 1833542a813cSJerry Jelinek tmp = zone->zone_stime; 1834542a813cSJerry Jelinek scalehrtime(&tmp); 1835542a813cSJerry Jelinek zmp->zm_stime.value.ui64 = tmp; 1836542a813cSJerry Jelinek tmp = zone->zone_wtime; 1837542a813cSJerry Jelinek scalehrtime(&tmp); 1838542a813cSJerry Jelinek zmp->zm_wtime.value.ui64 = tmp; 1839542a813cSJerry Jelinek 184081d43577SJerry Jelinek zmp->zm_avenrun1.value.ui32 = zone->zone_avenrun[0]; 184181d43577SJerry Jelinek zmp->zm_avenrun5.value.ui32 = zone->zone_avenrun[1]; 184281d43577SJerry Jelinek zmp->zm_avenrun15.value.ui32 = zone->zone_avenrun[2]; 184381d43577SJerry Jelinek 18442dc692e0SJerry Jelinek zmp->zm_ffcap.value.ui32 = zone->zone_ffcap; 18452dc692e0SJerry Jelinek zmp->zm_ffnoproc.value.ui32 = zone->zone_ffnoproc; 18462dc692e0SJerry Jelinek zmp->zm_ffnomem.value.ui32 = zone->zone_ffnomem; 18472dc692e0SJerry Jelinek zmp->zm_ffmisc.value.ui32 = zone->zone_ffmisc; 18482dc692e0SJerry Jelinek 1849542a813cSJerry Jelinek return (0); 1850542a813cSJerry Jelinek } 1851542a813cSJerry Jelinek 1852542a813cSJerry Jelinek static kstat_t * 1853542a813cSJerry Jelinek zone_misc_kstat_create(zone_t *zone) 1854542a813cSJerry Jelinek { 1855542a813cSJerry Jelinek kstat_t *ksp; 1856542a813cSJerry Jelinek zone_misc_kstat_t *zmp; 1857542a813cSJerry Jelinek 1858542a813cSJerry Jelinek if ((ksp = kstat_create_zone("zones", zone->zone_id, 1859542a813cSJerry Jelinek zone->zone_name, "zone_misc", KSTAT_TYPE_NAMED, 1860542a813cSJerry Jelinek sizeof (zone_misc_kstat_t) / sizeof (kstat_named_t), 1861542a813cSJerry Jelinek KSTAT_FLAG_VIRTUAL, zone->zone_id)) == NULL) 1862542a813cSJerry Jelinek return (NULL); 1863542a813cSJerry Jelinek 1864542a813cSJerry Jelinek if (zone->zone_id != GLOBAL_ZONEID) 1865542a813cSJerry Jelinek kstat_zone_add(ksp, GLOBAL_ZONEID); 1866542a813cSJerry Jelinek 1867542a813cSJerry Jelinek zmp = ksp->ks_data = kmem_zalloc(sizeof (zone_misc_kstat_t), KM_SLEEP); 1868542a813cSJerry Jelinek ksp->ks_data_size += strlen(zone->zone_name) + 1; 1869542a813cSJerry Jelinek ksp->ks_lock = &zone->zone_misc_lock; 1870542a813cSJerry Jelinek zone->zone_misc_stats = zmp; 1871542a813cSJerry Jelinek 1872542a813cSJerry Jelinek /* The kstat "name" field is not large enough for a full zonename */ 1873542a813cSJerry Jelinek kstat_named_init(&zmp->zm_zonename, "zonename", KSTAT_DATA_STRING); 1874542a813cSJerry Jelinek kstat_named_setstr(&zmp->zm_zonename, zone->zone_name); 1875542a813cSJerry Jelinek kstat_named_init(&zmp->zm_utime, "nsec_user", KSTAT_DATA_UINT64); 1876542a813cSJerry Jelinek kstat_named_init(&zmp->zm_stime, "nsec_sys", KSTAT_DATA_UINT64); 1877542a813cSJerry Jelinek kstat_named_init(&zmp->zm_wtime, "nsec_waitrq", KSTAT_DATA_UINT64); 187881d43577SJerry Jelinek kstat_named_init(&zmp->zm_avenrun1, "avenrun_1min", KSTAT_DATA_UINT32); 187981d43577SJerry Jelinek kstat_named_init(&zmp->zm_avenrun5, "avenrun_5min", KSTAT_DATA_UINT32); 188081d43577SJerry Jelinek kstat_named_init(&zmp->zm_avenrun15, "avenrun_15min", 188181d43577SJerry Jelinek KSTAT_DATA_UINT32); 18822dc692e0SJerry Jelinek kstat_named_init(&zmp->zm_ffcap, "forkfail_cap", KSTAT_DATA_UINT32); 18832dc692e0SJerry Jelinek kstat_named_init(&zmp->zm_ffnoproc, "forkfail_noproc", 18842dc692e0SJerry Jelinek KSTAT_DATA_UINT32); 18852dc692e0SJerry Jelinek kstat_named_init(&zmp->zm_ffnomem, "forkfail_nomem", KSTAT_DATA_UINT32); 18862dc692e0SJerry Jelinek kstat_named_init(&zmp->zm_ffmisc, "forkfail_misc", KSTAT_DATA_UINT32); 18872dc692e0SJerry Jelinek 1888542a813cSJerry Jelinek 1889542a813cSJerry Jelinek ksp->ks_update = zone_misc_kstat_update; 1890542a813cSJerry Jelinek ksp->ks_private = zone; 1891542a813cSJerry Jelinek 1892542a813cSJerry Jelinek kstat_install(ksp); 1893542a813cSJerry Jelinek return (ksp); 1894542a813cSJerry Jelinek } 1895542a813cSJerry Jelinek 1896ff19e029SMenno Lageman static void 1897ff19e029SMenno Lageman zone_kstat_create(zone_t *zone) 1898ff19e029SMenno Lageman { 1899ff19e029SMenno Lageman zone->zone_lockedmem_kstat = zone_kstat_create_common(zone, 1900ff19e029SMenno Lageman "lockedmem", zone_lockedmem_kstat_update); 1901ff19e029SMenno Lageman zone->zone_swapresv_kstat = zone_kstat_create_common(zone, 1902ff19e029SMenno Lageman "swapresv", zone_swapresv_kstat_update); 1903ff19e029SMenno Lageman zone->zone_nprocs_kstat = zone_kstat_create_common(zone, 1904ff19e029SMenno Lageman "nprocs", zone_nprocs_kstat_update); 1905542a813cSJerry Jelinek 1906542a813cSJerry Jelinek if ((zone->zone_misc_ksp = zone_misc_kstat_create(zone)) == NULL) { 1907542a813cSJerry Jelinek zone->zone_misc_stats = kmem_zalloc( 1908542a813cSJerry Jelinek sizeof (zone_misc_kstat_t), KM_SLEEP); 1909542a813cSJerry Jelinek } 19100209230bSgjelinek } 19110209230bSgjelinek 19120209230bSgjelinek static void 1913542a813cSJerry Jelinek zone_kstat_delete_common(kstat_t **pkstat, size_t datasz) 19140209230bSgjelinek { 19150209230bSgjelinek void *data; 19160209230bSgjelinek 1917ff19e029SMenno Lageman if (*pkstat != NULL) { 1918ff19e029SMenno Lageman data = (*pkstat)->ks_data; 1919ff19e029SMenno Lageman kstat_delete(*pkstat); 1920542a813cSJerry Jelinek kmem_free(data, datasz); 1921ff19e029SMenno Lageman *pkstat = NULL; 19220209230bSgjelinek } 19230209230bSgjelinek } 19240209230bSgjelinek 1925ff19e029SMenno Lageman static void 1926ff19e029SMenno Lageman zone_kstat_delete(zone_t *zone) 1927ff19e029SMenno Lageman { 1928542a813cSJerry Jelinek zone_kstat_delete_common(&zone->zone_lockedmem_kstat, 1929542a813cSJerry Jelinek sizeof (zone_kstat_t)); 1930542a813cSJerry Jelinek zone_kstat_delete_common(&zone->zone_swapresv_kstat, 1931542a813cSJerry Jelinek sizeof (zone_kstat_t)); 1932542a813cSJerry Jelinek zone_kstat_delete_common(&zone->zone_nprocs_kstat, 1933542a813cSJerry Jelinek sizeof (zone_kstat_t)); 1934542a813cSJerry Jelinek zone_kstat_delete_common(&zone->zone_misc_ksp, 1935542a813cSJerry Jelinek sizeof (zone_misc_kstat_t)); 1936ff19e029SMenno Lageman } 1937ff19e029SMenno Lageman 19387c478bd9Sstevel@tonic-gate /* 19397c478bd9Sstevel@tonic-gate * Called very early on in boot to initialize the ZSD list so that 19407c478bd9Sstevel@tonic-gate * zone_key_create() can be called before zone_init(). It also initializes 19417c478bd9Sstevel@tonic-gate * portions of zone0 which may be used before zone_init() is called. The 19427c478bd9Sstevel@tonic-gate * variable "global_zone" will be set when zone0 is fully initialized by 19437c478bd9Sstevel@tonic-gate * zone_init(). 19447c478bd9Sstevel@tonic-gate */ 19457c478bd9Sstevel@tonic-gate void 19467c478bd9Sstevel@tonic-gate zone_zsd_init(void) 19477c478bd9Sstevel@tonic-gate { 19487c478bd9Sstevel@tonic-gate mutex_init(&zonehash_lock, NULL, MUTEX_DEFAULT, NULL); 19497c478bd9Sstevel@tonic-gate mutex_init(&zsd_key_lock, NULL, MUTEX_DEFAULT, NULL); 19507c478bd9Sstevel@tonic-gate list_create(&zsd_registered_keys, sizeof (struct zsd_entry), 19517c478bd9Sstevel@tonic-gate offsetof(struct zsd_entry, zsd_linkage)); 19527c478bd9Sstevel@tonic-gate list_create(&zone_active, sizeof (zone_t), 19537c478bd9Sstevel@tonic-gate offsetof(zone_t, zone_linkage)); 19547c478bd9Sstevel@tonic-gate list_create(&zone_deathrow, sizeof (zone_t), 19557c478bd9Sstevel@tonic-gate offsetof(zone_t, zone_linkage)); 19567c478bd9Sstevel@tonic-gate 19577c478bd9Sstevel@tonic-gate mutex_init(&zone0.zone_lock, NULL, MUTEX_DEFAULT, NULL); 19587c478bd9Sstevel@tonic-gate mutex_init(&zone0.zone_nlwps_lock, NULL, MUTEX_DEFAULT, NULL); 19590209230bSgjelinek mutex_init(&zone0.zone_mem_lock, NULL, MUTEX_DEFAULT, NULL); 19607c478bd9Sstevel@tonic-gate zone0.zone_shares = 1; 19610209230bSgjelinek zone0.zone_nlwps = 0; 19627c478bd9Sstevel@tonic-gate zone0.zone_nlwps_ctl = INT_MAX; 1963ff19e029SMenno Lageman zone0.zone_nprocs = 0; 1964ff19e029SMenno Lageman zone0.zone_nprocs_ctl = INT_MAX; 19650209230bSgjelinek zone0.zone_locked_mem = 0; 19660209230bSgjelinek zone0.zone_locked_mem_ctl = UINT64_MAX; 19670209230bSgjelinek ASSERT(zone0.zone_max_swap == 0); 19680209230bSgjelinek zone0.zone_max_swap_ctl = UINT64_MAX; 19690fbb751dSJohn Levon zone0.zone_max_lofi = 0; 19700fbb751dSJohn Levon zone0.zone_max_lofi_ctl = UINT64_MAX; 1971824c205fSml zone0.zone_shmmax = 0; 1972824c205fSml zone0.zone_ipc.ipcq_shmmni = 0; 1973824c205fSml zone0.zone_ipc.ipcq_semmni = 0; 1974824c205fSml zone0.zone_ipc.ipcq_msgmni = 0; 19757c478bd9Sstevel@tonic-gate zone0.zone_name = GLOBAL_ZONENAME; 19767c478bd9Sstevel@tonic-gate zone0.zone_nodename = utsname.nodename; 19777c478bd9Sstevel@tonic-gate zone0.zone_domain = srpc_domain; 19785679c89fSjv zone0.zone_hostid = HW_INVALID_HOSTID; 19790fbb751dSJohn Levon zone0.zone_fs_allowed = NULL; 19807c478bd9Sstevel@tonic-gate zone0.zone_ref = 1; 19817c478bd9Sstevel@tonic-gate zone0.zone_id = GLOBAL_ZONEID; 19827c478bd9Sstevel@tonic-gate zone0.zone_status = ZONE_IS_RUNNING; 19837c478bd9Sstevel@tonic-gate zone0.zone_rootpath = "/"; 19847c478bd9Sstevel@tonic-gate zone0.zone_rootpathlen = 2; 19857c478bd9Sstevel@tonic-gate zone0.zone_psetid = ZONE_PS_INVAL; 19867c478bd9Sstevel@tonic-gate zone0.zone_ncpus = 0; 19877c478bd9Sstevel@tonic-gate zone0.zone_ncpus_online = 0; 19887c478bd9Sstevel@tonic-gate zone0.zone_proc_initpid = 1; 19893f2f09c1Sdp zone0.zone_initname = initname; 19900209230bSgjelinek zone0.zone_lockedmem_kstat = NULL; 19910209230bSgjelinek zone0.zone_swapresv_kstat = NULL; 1992ff19e029SMenno Lageman zone0.zone_nprocs_kstat = NULL; 1993542a813cSJerry Jelinek 1994542a813cSJerry Jelinek zone0.zone_stime = 0; 1995542a813cSJerry Jelinek zone0.zone_utime = 0; 1996542a813cSJerry Jelinek zone0.zone_wtime = 0; 1997542a813cSJerry Jelinek 1998a19609f8Sjv list_create(&zone0.zone_ref_list, sizeof (zone_ref_t), 1999a19609f8Sjv offsetof(zone_ref_t, zref_linkage)); 20007c478bd9Sstevel@tonic-gate list_create(&zone0.zone_zsd, sizeof (struct zsd_entry), 20017c478bd9Sstevel@tonic-gate offsetof(struct zsd_entry, zsd_linkage)); 20027c478bd9Sstevel@tonic-gate list_insert_head(&zone_active, &zone0); 20037c478bd9Sstevel@tonic-gate 20047c478bd9Sstevel@tonic-gate /* 20057c478bd9Sstevel@tonic-gate * The root filesystem is not mounted yet, so zone_rootvp cannot be set 20067c478bd9Sstevel@tonic-gate * to anything meaningful. It is assigned to be 'rootdir' in 20077c478bd9Sstevel@tonic-gate * vfs_mountroot(). 20087c478bd9Sstevel@tonic-gate */ 20097c478bd9Sstevel@tonic-gate zone0.zone_rootvp = NULL; 20107c478bd9Sstevel@tonic-gate zone0.zone_vfslist = NULL; 20113f2f09c1Sdp zone0.zone_bootargs = initargs; 20127c478bd9Sstevel@tonic-gate zone0.zone_privset = kmem_alloc(sizeof (priv_set_t), KM_SLEEP); 20137c478bd9Sstevel@tonic-gate /* 20147c478bd9Sstevel@tonic-gate * The global zone has all privileges 20157c478bd9Sstevel@tonic-gate */ 20167c478bd9Sstevel@tonic-gate priv_fillset(zone0.zone_privset); 20177c478bd9Sstevel@tonic-gate /* 20187c478bd9Sstevel@tonic-gate * Add p0 to the global zone 20197c478bd9Sstevel@tonic-gate */ 20207c478bd9Sstevel@tonic-gate zone0.zone_zsched = &p0; 20217c478bd9Sstevel@tonic-gate p0.p_zone = &zone0; 20227c478bd9Sstevel@tonic-gate } 20237c478bd9Sstevel@tonic-gate 202445916cd2Sjpk /* 202545916cd2Sjpk * Compute a hash value based on the contents of the label and the DOI. The 202645916cd2Sjpk * hash algorithm is somewhat arbitrary, but is based on the observation that 202745916cd2Sjpk * humans will likely pick labels that differ by amounts that work out to be 202845916cd2Sjpk * multiples of the number of hash chains, and thus stirring in some primes 202945916cd2Sjpk * should help. 203045916cd2Sjpk */ 203145916cd2Sjpk static uint_t 203245916cd2Sjpk hash_bylabel(void *hdata, mod_hash_key_t key) 203345916cd2Sjpk { 203445916cd2Sjpk const ts_label_t *lab = (ts_label_t *)key; 203545916cd2Sjpk const uint32_t *up, *ue; 203645916cd2Sjpk uint_t hash; 203745916cd2Sjpk int i; 203845916cd2Sjpk 203945916cd2Sjpk _NOTE(ARGUNUSED(hdata)); 204045916cd2Sjpk 204145916cd2Sjpk hash = lab->tsl_doi + (lab->tsl_doi << 1); 204245916cd2Sjpk /* we depend on alignment of label, but not representation */ 204345916cd2Sjpk up = (const uint32_t *)&lab->tsl_label; 204445916cd2Sjpk ue = up + sizeof (lab->tsl_label) / sizeof (*up); 204545916cd2Sjpk i = 1; 204645916cd2Sjpk while (up < ue) { 204745916cd2Sjpk /* using 2^n + 1, 1 <= n <= 16 as source of many primes */ 204845916cd2Sjpk hash += *up + (*up << ((i % 16) + 1)); 204945916cd2Sjpk up++; 205045916cd2Sjpk i++; 205145916cd2Sjpk } 205245916cd2Sjpk return (hash); 205345916cd2Sjpk } 205445916cd2Sjpk 205545916cd2Sjpk /* 205645916cd2Sjpk * All that mod_hash cares about here is zero (equal) versus non-zero (not 205745916cd2Sjpk * equal). This may need to be changed if less than / greater than is ever 205845916cd2Sjpk * needed. 205945916cd2Sjpk */ 206045916cd2Sjpk static int 206145916cd2Sjpk hash_labelkey_cmp(mod_hash_key_t key1, mod_hash_key_t key2) 206245916cd2Sjpk { 206345916cd2Sjpk ts_label_t *lab1 = (ts_label_t *)key1; 206445916cd2Sjpk ts_label_t *lab2 = (ts_label_t *)key2; 206545916cd2Sjpk 206645916cd2Sjpk return (label_equal(lab1, lab2) ? 0 : 1); 206745916cd2Sjpk } 206845916cd2Sjpk 20697c478bd9Sstevel@tonic-gate /* 20707c478bd9Sstevel@tonic-gate * Called by main() to initialize the zones framework. 20717c478bd9Sstevel@tonic-gate */ 20727c478bd9Sstevel@tonic-gate void 20737c478bd9Sstevel@tonic-gate zone_init(void) 20747c478bd9Sstevel@tonic-gate { 20757c478bd9Sstevel@tonic-gate rctl_dict_entry_t *rde; 20767c478bd9Sstevel@tonic-gate rctl_val_t *dval; 20777c478bd9Sstevel@tonic-gate rctl_set_t *set; 20787c478bd9Sstevel@tonic-gate rctl_alloc_gp_t *gp; 20797c478bd9Sstevel@tonic-gate rctl_entity_p_t e; 2080cf8f45c7Sdstaff int res; 20817c478bd9Sstevel@tonic-gate 20827c478bd9Sstevel@tonic-gate ASSERT(curproc == &p0); 20837c478bd9Sstevel@tonic-gate 20847c478bd9Sstevel@tonic-gate /* 20857c478bd9Sstevel@tonic-gate * Create ID space for zone IDs. ID 0 is reserved for the 20867c478bd9Sstevel@tonic-gate * global zone. 20877c478bd9Sstevel@tonic-gate */ 20887c478bd9Sstevel@tonic-gate zoneid_space = id_space_create("zoneid_space", 1, MAX_ZONEID); 20897c478bd9Sstevel@tonic-gate 20907c478bd9Sstevel@tonic-gate /* 20917c478bd9Sstevel@tonic-gate * Initialize generic zone resource controls, if any. 20927c478bd9Sstevel@tonic-gate */ 20937c478bd9Sstevel@tonic-gate rc_zone_cpu_shares = rctl_register("zone.cpu-shares", 20947c478bd9Sstevel@tonic-gate RCENTITY_ZONE, RCTL_GLOBAL_SIGNAL_NEVER | RCTL_GLOBAL_DENY_NEVER | 209519f92332Sml RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_COUNT | RCTL_GLOBAL_SYSLOG_NEVER, 2096c97ad5cdSakolb FSS_MAXSHARES, FSS_MAXSHARES, &zone_cpu_shares_ops); 2097c97ad5cdSakolb 2098c97ad5cdSakolb rc_zone_cpu_cap = rctl_register("zone.cpu-cap", 2099c97ad5cdSakolb RCENTITY_ZONE, RCTL_GLOBAL_SIGNAL_NEVER | RCTL_GLOBAL_DENY_ALWAYS | 2100c97ad5cdSakolb RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_COUNT |RCTL_GLOBAL_SYSLOG_NEVER | 2101c97ad5cdSakolb RCTL_GLOBAL_INFINITE, 2102c97ad5cdSakolb MAXCAP, MAXCAP, &zone_cpu_cap_ops); 21037c478bd9Sstevel@tonic-gate 21047c478bd9Sstevel@tonic-gate rc_zone_nlwps = rctl_register("zone.max-lwps", RCENTITY_ZONE, 21057c478bd9Sstevel@tonic-gate RCTL_GLOBAL_NOACTION | RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_COUNT, 21067c478bd9Sstevel@tonic-gate INT_MAX, INT_MAX, &zone_lwps_ops); 2107ff19e029SMenno Lageman 2108ff19e029SMenno Lageman rc_zone_nprocs = rctl_register("zone.max-processes", RCENTITY_ZONE, 2109ff19e029SMenno Lageman RCTL_GLOBAL_NOACTION | RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_COUNT, 2110ff19e029SMenno Lageman INT_MAX, INT_MAX, &zone_procs_ops); 2111ff19e029SMenno Lageman 2112824c205fSml /* 2113824c205fSml * System V IPC resource controls 2114824c205fSml */ 2115824c205fSml rc_zone_msgmni = rctl_register("zone.max-msg-ids", 2116824c205fSml RCENTITY_ZONE, RCTL_GLOBAL_DENY_ALWAYS | RCTL_GLOBAL_NOBASIC | 2117824c205fSml RCTL_GLOBAL_COUNT, IPC_IDS_MAX, IPC_IDS_MAX, &zone_msgmni_ops); 2118824c205fSml 2119824c205fSml rc_zone_semmni = rctl_register("zone.max-sem-ids", 2120824c205fSml RCENTITY_ZONE, RCTL_GLOBAL_DENY_ALWAYS | RCTL_GLOBAL_NOBASIC | 2121824c205fSml RCTL_GLOBAL_COUNT, IPC_IDS_MAX, IPC_IDS_MAX, &zone_semmni_ops); 2122824c205fSml 2123824c205fSml rc_zone_shmmni = rctl_register("zone.max-shm-ids", 2124824c205fSml RCENTITY_ZONE, RCTL_GLOBAL_DENY_ALWAYS | RCTL_GLOBAL_NOBASIC | 2125824c205fSml RCTL_GLOBAL_COUNT, IPC_IDS_MAX, IPC_IDS_MAX, &zone_shmmni_ops); 2126824c205fSml 2127824c205fSml rc_zone_shmmax = rctl_register("zone.max-shm-memory", 2128824c205fSml RCENTITY_ZONE, RCTL_GLOBAL_DENY_ALWAYS | RCTL_GLOBAL_NOBASIC | 2129824c205fSml RCTL_GLOBAL_BYTES, UINT64_MAX, UINT64_MAX, &zone_shmmax_ops); 2130824c205fSml 21317c478bd9Sstevel@tonic-gate /* 21327c478bd9Sstevel@tonic-gate * Create a rctl_val with PRIVILEGED, NOACTION, value = 1. Then attach 21337c478bd9Sstevel@tonic-gate * this at the head of the rctl_dict_entry for ``zone.cpu-shares''. 21347c478bd9Sstevel@tonic-gate */ 21357c478bd9Sstevel@tonic-gate dval = kmem_cache_alloc(rctl_val_cache, KM_SLEEP); 21367c478bd9Sstevel@tonic-gate bzero(dval, sizeof (rctl_val_t)); 21377c478bd9Sstevel@tonic-gate dval->rcv_value = 1; 21387c478bd9Sstevel@tonic-gate dval->rcv_privilege = RCPRIV_PRIVILEGED; 21397c478bd9Sstevel@tonic-gate dval->rcv_flagaction = RCTL_LOCAL_NOACTION; 21407c478bd9Sstevel@tonic-gate dval->rcv_action_recip_pid = -1; 21417c478bd9Sstevel@tonic-gate 21427c478bd9Sstevel@tonic-gate rde = rctl_dict_lookup("zone.cpu-shares"); 21437c478bd9Sstevel@tonic-gate (void) rctl_val_list_insert(&rde->rcd_default_value, dval); 21447c478bd9Sstevel@tonic-gate 2145c6939658Ssl rc_zone_locked_mem = rctl_register("zone.max-locked-memory", 2146c6939658Ssl RCENTITY_ZONE, RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_BYTES | 2147c6939658Ssl RCTL_GLOBAL_DENY_ALWAYS, UINT64_MAX, UINT64_MAX, 2148c6939658Ssl &zone_locked_mem_ops); 21490209230bSgjelinek 21500209230bSgjelinek rc_zone_max_swap = rctl_register("zone.max-swap", 21510209230bSgjelinek RCENTITY_ZONE, RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_BYTES | 21520209230bSgjelinek RCTL_GLOBAL_DENY_ALWAYS, UINT64_MAX, UINT64_MAX, 21530209230bSgjelinek &zone_max_swap_ops); 21540209230bSgjelinek 21550fbb751dSJohn Levon rc_zone_max_lofi = rctl_register("zone.max-lofi", 21560fbb751dSJohn Levon RCENTITY_ZONE, RCTL_GLOBAL_NOBASIC | RCTL_GLOBAL_COUNT | 21570fbb751dSJohn Levon RCTL_GLOBAL_DENY_ALWAYS, UINT64_MAX, UINT64_MAX, 21580fbb751dSJohn Levon &zone_max_lofi_ops); 21590fbb751dSJohn Levon 21607c478bd9Sstevel@tonic-gate /* 21617c478bd9Sstevel@tonic-gate * Initialize the ``global zone''. 21627c478bd9Sstevel@tonic-gate */ 21637c478bd9Sstevel@tonic-gate set = rctl_set_create(); 21647c478bd9Sstevel@tonic-gate gp = rctl_set_init_prealloc(RCENTITY_ZONE); 21657c478bd9Sstevel@tonic-gate mutex_enter(&p0.p_lock); 21667c478bd9Sstevel@tonic-gate e.rcep_p.zone = &zone0; 21677c478bd9Sstevel@tonic-gate e.rcep_t = RCENTITY_ZONE; 21687c478bd9Sstevel@tonic-gate zone0.zone_rctls = rctl_set_init(RCENTITY_ZONE, &p0, &e, set, 21697c478bd9Sstevel@tonic-gate gp); 21707c478bd9Sstevel@tonic-gate 21717c478bd9Sstevel@tonic-gate zone0.zone_nlwps = p0.p_lwpcnt; 2172ff19e029SMenno Lageman zone0.zone_nprocs = 1; 21737c478bd9Sstevel@tonic-gate zone0.zone_ntasks = 1; 21747c478bd9Sstevel@tonic-gate mutex_exit(&p0.p_lock); 21759acbbeafSnn zone0.zone_restart_init = B_TRUE; 21769acbbeafSnn zone0.zone_brand = &native_brand; 21777c478bd9Sstevel@tonic-gate rctl_prealloc_destroy(gp); 21787c478bd9Sstevel@tonic-gate /* 21790209230bSgjelinek * pool_default hasn't been initialized yet, so we let pool_init() 21800209230bSgjelinek * take care of making sure the global zone is in the default pool. 21810209230bSgjelinek */ 21820209230bSgjelinek 21830209230bSgjelinek /* 21840209230bSgjelinek * Initialize global zone kstats 21857c478bd9Sstevel@tonic-gate */ 21860209230bSgjelinek zone_kstat_create(&zone0); 218745916cd2Sjpk 218845916cd2Sjpk /* 218945916cd2Sjpk * Initialize zone label. 219045916cd2Sjpk * mlp are initialized when tnzonecfg is loaded. 219145916cd2Sjpk */ 219245916cd2Sjpk zone0.zone_slabel = l_admin_low; 219345916cd2Sjpk rw_init(&zone0.zone_mlps.mlpl_rwlock, NULL, RW_DEFAULT, NULL); 219445916cd2Sjpk label_hold(l_admin_low); 219545916cd2Sjpk 2196835ee219SRobert Harris /* 2197835ee219SRobert Harris * Initialise the lock for the database structure used by mntfs. 2198835ee219SRobert Harris */ 2199835ee219SRobert Harris rw_init(&zone0.zone_mntfs_db_lock, NULL, RW_DEFAULT, NULL); 2200835ee219SRobert Harris 22017c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 22027c478bd9Sstevel@tonic-gate zone_uniqid(&zone0); 22037c478bd9Sstevel@tonic-gate ASSERT(zone0.zone_uniqid == GLOBAL_ZONEUNIQID); 220445916cd2Sjpk 22057c478bd9Sstevel@tonic-gate zonehashbyid = mod_hash_create_idhash("zone_by_id", zone_hash_size, 22067c478bd9Sstevel@tonic-gate mod_hash_null_valdtor); 22077c478bd9Sstevel@tonic-gate zonehashbyname = mod_hash_create_strhash("zone_by_name", 22087c478bd9Sstevel@tonic-gate zone_hash_size, mod_hash_null_valdtor); 220945916cd2Sjpk /* 221045916cd2Sjpk * maintain zonehashbylabel only for labeled systems 221145916cd2Sjpk */ 221245916cd2Sjpk if (is_system_labeled()) 221345916cd2Sjpk zonehashbylabel = mod_hash_create_extended("zone_by_label", 221445916cd2Sjpk zone_hash_size, mod_hash_null_keydtor, 221545916cd2Sjpk mod_hash_null_valdtor, hash_bylabel, NULL, 221645916cd2Sjpk hash_labelkey_cmp, KM_SLEEP); 22177c478bd9Sstevel@tonic-gate zonecount = 1; 22187c478bd9Sstevel@tonic-gate 22197c478bd9Sstevel@tonic-gate (void) mod_hash_insert(zonehashbyid, (mod_hash_key_t)GLOBAL_ZONEID, 22207c478bd9Sstevel@tonic-gate (mod_hash_val_t)&zone0); 22217c478bd9Sstevel@tonic-gate (void) mod_hash_insert(zonehashbyname, (mod_hash_key_t)zone0.zone_name, 22227c478bd9Sstevel@tonic-gate (mod_hash_val_t)&zone0); 222348451833Scarlsonj if (is_system_labeled()) { 222448451833Scarlsonj zone0.zone_flags |= ZF_HASHED_LABEL; 222545916cd2Sjpk (void) mod_hash_insert(zonehashbylabel, 222645916cd2Sjpk (mod_hash_key_t)zone0.zone_slabel, (mod_hash_val_t)&zone0); 222748451833Scarlsonj } 222845916cd2Sjpk mutex_exit(&zonehash_lock); 222945916cd2Sjpk 22307c478bd9Sstevel@tonic-gate /* 22317c478bd9Sstevel@tonic-gate * We avoid setting zone_kcred until now, since kcred is initialized 22327c478bd9Sstevel@tonic-gate * sometime after zone_zsd_init() and before zone_init(). 22337c478bd9Sstevel@tonic-gate */ 22347c478bd9Sstevel@tonic-gate zone0.zone_kcred = kcred; 22357c478bd9Sstevel@tonic-gate /* 22367c478bd9Sstevel@tonic-gate * The global zone is fully initialized (except for zone_rootvp which 22377c478bd9Sstevel@tonic-gate * will be set when the root filesystem is mounted). 22387c478bd9Sstevel@tonic-gate */ 22397c478bd9Sstevel@tonic-gate global_zone = &zone0; 2240cf8f45c7Sdstaff 2241cf8f45c7Sdstaff /* 2242cf8f45c7Sdstaff * Setup an event channel to send zone status change notifications on 2243cf8f45c7Sdstaff */ 2244cf8f45c7Sdstaff res = sysevent_evc_bind(ZONE_EVENT_CHANNEL, &zone_event_chan, 2245cf8f45c7Sdstaff EVCH_CREAT); 2246cf8f45c7Sdstaff 2247cf8f45c7Sdstaff if (res) 2248cf8f45c7Sdstaff panic("Sysevent_evc_bind failed during zone setup.\n"); 22490209230bSgjelinek 22507c478bd9Sstevel@tonic-gate } 22517c478bd9Sstevel@tonic-gate 22527c478bd9Sstevel@tonic-gate static void 22537c478bd9Sstevel@tonic-gate zone_free(zone_t *zone) 22547c478bd9Sstevel@tonic-gate { 22557c478bd9Sstevel@tonic-gate ASSERT(zone != global_zone); 22567c478bd9Sstevel@tonic-gate ASSERT(zone->zone_ntasks == 0); 22577c478bd9Sstevel@tonic-gate ASSERT(zone->zone_nlwps == 0); 2258ff19e029SMenno Lageman ASSERT(zone->zone_nprocs == 0); 22597c478bd9Sstevel@tonic-gate ASSERT(zone->zone_cred_ref == 0); 22607c478bd9Sstevel@tonic-gate ASSERT(zone->zone_kcred == NULL); 22617c478bd9Sstevel@tonic-gate ASSERT(zone_status_get(zone) == ZONE_IS_DEAD || 22627c478bd9Sstevel@tonic-gate zone_status_get(zone) == ZONE_IS_UNINITIALIZED); 2263a19609f8Sjv ASSERT(list_is_empty(&zone->zone_ref_list)); 22647c478bd9Sstevel@tonic-gate 2265c97ad5cdSakolb /* 2266c97ad5cdSakolb * Remove any zone caps. 2267c97ad5cdSakolb */ 2268c97ad5cdSakolb cpucaps_zone_remove(zone); 2269c97ad5cdSakolb 2270c97ad5cdSakolb ASSERT(zone->zone_cpucap == NULL); 2271c97ad5cdSakolb 22727c478bd9Sstevel@tonic-gate /* remove from deathrow list */ 22737c478bd9Sstevel@tonic-gate if (zone_status_get(zone) == ZONE_IS_DEAD) { 22747c478bd9Sstevel@tonic-gate ASSERT(zone->zone_ref == 0); 22757c478bd9Sstevel@tonic-gate mutex_enter(&zone_deathrow_lock); 22767c478bd9Sstevel@tonic-gate list_remove(&zone_deathrow, zone); 22777c478bd9Sstevel@tonic-gate mutex_exit(&zone_deathrow_lock); 22787c478bd9Sstevel@tonic-gate } 22797c478bd9Sstevel@tonic-gate 2280a19609f8Sjv list_destroy(&zone->zone_ref_list); 22817c478bd9Sstevel@tonic-gate zone_free_zsd(zone); 2282fa9e4066Sahrens zone_free_datasets(zone); 22832b24ab6bSSebastien Roy list_destroy(&zone->zone_dl_list); 22847c478bd9Sstevel@tonic-gate 22857c478bd9Sstevel@tonic-gate if (zone->zone_rootvp != NULL) 22867c478bd9Sstevel@tonic-gate VN_RELE(zone->zone_rootvp); 22877c478bd9Sstevel@tonic-gate if (zone->zone_rootpath) 22887c478bd9Sstevel@tonic-gate kmem_free(zone->zone_rootpath, zone->zone_rootpathlen); 22897c478bd9Sstevel@tonic-gate if (zone->zone_name != NULL) 22907c478bd9Sstevel@tonic-gate kmem_free(zone->zone_name, ZONENAME_MAX); 229145916cd2Sjpk if (zone->zone_slabel != NULL) 229245916cd2Sjpk label_rele(zone->zone_slabel); 22937c478bd9Sstevel@tonic-gate if (zone->zone_nodename != NULL) 22947c478bd9Sstevel@tonic-gate kmem_free(zone->zone_nodename, _SYS_NMLN); 22957c478bd9Sstevel@tonic-gate if (zone->zone_domain != NULL) 22967c478bd9Sstevel@tonic-gate kmem_free(zone->zone_domain, _SYS_NMLN); 22977c478bd9Sstevel@tonic-gate if (zone->zone_privset != NULL) 22987c478bd9Sstevel@tonic-gate kmem_free(zone->zone_privset, sizeof (priv_set_t)); 22997c478bd9Sstevel@tonic-gate if (zone->zone_rctls != NULL) 23007c478bd9Sstevel@tonic-gate rctl_set_free(zone->zone_rctls); 23017c478bd9Sstevel@tonic-gate if (zone->zone_bootargs != NULL) 23020fbb751dSJohn Levon strfree(zone->zone_bootargs); 23033f2f09c1Sdp if (zone->zone_initname != NULL) 23040fbb751dSJohn Levon strfree(zone->zone_initname); 23050fbb751dSJohn Levon if (zone->zone_fs_allowed != NULL) 23060fbb751dSJohn Levon strfree(zone->zone_fs_allowed); 2307134a1f4eSCasper H.S. Dik if (zone->zone_pfexecd != NULL) 2308134a1f4eSCasper H.S. Dik klpd_freelist(&zone->zone_pfexecd); 23097c478bd9Sstevel@tonic-gate id_free(zoneid_space, zone->zone_id); 23107c478bd9Sstevel@tonic-gate mutex_destroy(&zone->zone_lock); 23117c478bd9Sstevel@tonic-gate cv_destroy(&zone->zone_cv); 231245916cd2Sjpk rw_destroy(&zone->zone_mlps.mlpl_rwlock); 2313835ee219SRobert Harris rw_destroy(&zone->zone_mntfs_db_lock); 23147c478bd9Sstevel@tonic-gate kmem_free(zone, sizeof (zone_t)); 23157c478bd9Sstevel@tonic-gate } 23167c478bd9Sstevel@tonic-gate 23177c478bd9Sstevel@tonic-gate /* 23187c478bd9Sstevel@tonic-gate * See block comment at the top of this file for information about zone 23197c478bd9Sstevel@tonic-gate * status values. 23207c478bd9Sstevel@tonic-gate */ 23217c478bd9Sstevel@tonic-gate /* 23227c478bd9Sstevel@tonic-gate * Convenience function for setting zone status. 23237c478bd9Sstevel@tonic-gate */ 23247c478bd9Sstevel@tonic-gate static void 23257c478bd9Sstevel@tonic-gate zone_status_set(zone_t *zone, zone_status_t status) 23267c478bd9Sstevel@tonic-gate { 2327cf8f45c7Sdstaff 2328cf8f45c7Sdstaff nvlist_t *nvl = NULL; 23297c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zone_status_lock)); 23307c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE && 23317c478bd9Sstevel@tonic-gate status >= zone_status_get(zone)); 2332cf8f45c7Sdstaff 2333cf8f45c7Sdstaff if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_SLEEP) || 2334cf8f45c7Sdstaff nvlist_add_string(nvl, ZONE_CB_NAME, zone->zone_name) || 2335cf8f45c7Sdstaff nvlist_add_string(nvl, ZONE_CB_NEWSTATE, 23363f2f09c1Sdp zone_status_table[status]) || 2337cf8f45c7Sdstaff nvlist_add_string(nvl, ZONE_CB_OLDSTATE, 23383f2f09c1Sdp zone_status_table[zone->zone_status]) || 2339cf8f45c7Sdstaff nvlist_add_int32(nvl, ZONE_CB_ZONEID, zone->zone_id) || 2340cf8f45c7Sdstaff nvlist_add_uint64(nvl, ZONE_CB_TIMESTAMP, (uint64_t)gethrtime()) || 2341cf8f45c7Sdstaff sysevent_evc_publish(zone_event_chan, ZONE_EVENT_STATUS_CLASS, 23423f2f09c1Sdp ZONE_EVENT_STATUS_SUBCLASS, "sun.com", "kernel", nvl, EVCH_SLEEP)) { 2343cf8f45c7Sdstaff #ifdef DEBUG 2344cf8f45c7Sdstaff (void) printf( 2345cf8f45c7Sdstaff "Failed to allocate and send zone state change event.\n"); 2346cf8f45c7Sdstaff #endif 2347cf8f45c7Sdstaff } 2348cf8f45c7Sdstaff nvlist_free(nvl); 2349cf8f45c7Sdstaff 23507c478bd9Sstevel@tonic-gate zone->zone_status = status; 2351cf8f45c7Sdstaff 23527c478bd9Sstevel@tonic-gate cv_broadcast(&zone->zone_cv); 23537c478bd9Sstevel@tonic-gate } 23547c478bd9Sstevel@tonic-gate 23557c478bd9Sstevel@tonic-gate /* 23567c478bd9Sstevel@tonic-gate * Public function to retrieve the zone status. The zone status may 23577c478bd9Sstevel@tonic-gate * change after it is retrieved. 23587c478bd9Sstevel@tonic-gate */ 23597c478bd9Sstevel@tonic-gate zone_status_t 23607c478bd9Sstevel@tonic-gate zone_status_get(zone_t *zone) 23617c478bd9Sstevel@tonic-gate { 23627c478bd9Sstevel@tonic-gate return (zone->zone_status); 23637c478bd9Sstevel@tonic-gate } 23647c478bd9Sstevel@tonic-gate 23657c478bd9Sstevel@tonic-gate static int 23667c478bd9Sstevel@tonic-gate zone_set_bootargs(zone_t *zone, const char *zone_bootargs) 23677c478bd9Sstevel@tonic-gate { 23680fbb751dSJohn Levon char *buf = kmem_zalloc(BOOTARGS_MAX, KM_SLEEP); 23693f2f09c1Sdp int err = 0; 23703f2f09c1Sdp 23713f2f09c1Sdp ASSERT(zone != global_zone); 23720fbb751dSJohn Levon if ((err = copyinstr(zone_bootargs, buf, BOOTARGS_MAX, NULL)) != 0) 23733f2f09c1Sdp goto done; /* EFAULT or ENAMETOOLONG */ 23743f2f09c1Sdp 23753f2f09c1Sdp if (zone->zone_bootargs != NULL) 23760fbb751dSJohn Levon strfree(zone->zone_bootargs); 23773f2f09c1Sdp 23780fbb751dSJohn Levon zone->zone_bootargs = strdup(buf); 23793f2f09c1Sdp 23803f2f09c1Sdp done: 23810fbb751dSJohn Levon kmem_free(buf, BOOTARGS_MAX); 23823f2f09c1Sdp return (err); 23833f2f09c1Sdp } 23843f2f09c1Sdp 238559f2ff5cSedp static int 238659f2ff5cSedp zone_set_brand(zone_t *zone, const char *brand) 238759f2ff5cSedp { 238859f2ff5cSedp struct brand_attr *attrp; 238959f2ff5cSedp brand_t *bp; 239059f2ff5cSedp 239159f2ff5cSedp attrp = kmem_alloc(sizeof (struct brand_attr), KM_SLEEP); 239259f2ff5cSedp if (copyin(brand, attrp, sizeof (struct brand_attr)) != 0) { 239359f2ff5cSedp kmem_free(attrp, sizeof (struct brand_attr)); 239459f2ff5cSedp return (EFAULT); 239559f2ff5cSedp } 239659f2ff5cSedp 239759f2ff5cSedp bp = brand_register_zone(attrp); 239859f2ff5cSedp kmem_free(attrp, sizeof (struct brand_attr)); 239959f2ff5cSedp if (bp == NULL) 240059f2ff5cSedp return (EINVAL); 240159f2ff5cSedp 240259f2ff5cSedp /* 240359f2ff5cSedp * This is the only place where a zone can change it's brand. 240459f2ff5cSedp * We already need to hold zone_status_lock to check the zone 240559f2ff5cSedp * status, so we'll just use that lock to serialize zone 240659f2ff5cSedp * branding requests as well. 240759f2ff5cSedp */ 240859f2ff5cSedp mutex_enter(&zone_status_lock); 240959f2ff5cSedp 241059f2ff5cSedp /* Re-Branding is not allowed and the zone can't be booted yet */ 241159f2ff5cSedp if ((ZONE_IS_BRANDED(zone)) || 241259f2ff5cSedp (zone_status_get(zone) >= ZONE_IS_BOOTING)) { 241359f2ff5cSedp mutex_exit(&zone_status_lock); 241459f2ff5cSedp brand_unregister_zone(bp); 241559f2ff5cSedp return (EINVAL); 241659f2ff5cSedp } 241759f2ff5cSedp 2418319378d9Seh /* set up the brand specific data */ 241959f2ff5cSedp zone->zone_brand = bp; 2420319378d9Seh ZBROP(zone)->b_init_brand_data(zone); 2421319378d9Seh 242259f2ff5cSedp mutex_exit(&zone_status_lock); 242359f2ff5cSedp return (0); 242459f2ff5cSedp } 242559f2ff5cSedp 24260fbb751dSJohn Levon static int 24270fbb751dSJohn Levon zone_set_fs_allowed(zone_t *zone, const char *zone_fs_allowed) 24280fbb751dSJohn Levon { 24290fbb751dSJohn Levon char *buf = kmem_zalloc(ZONE_FS_ALLOWED_MAX, KM_SLEEP); 24300fbb751dSJohn Levon int err = 0; 24310fbb751dSJohn Levon 24320fbb751dSJohn Levon ASSERT(zone != global_zone); 24330fbb751dSJohn Levon if ((err = copyinstr(zone_fs_allowed, buf, 24340fbb751dSJohn Levon ZONE_FS_ALLOWED_MAX, NULL)) != 0) 24350fbb751dSJohn Levon goto done; 24360fbb751dSJohn Levon 24370fbb751dSJohn Levon if (zone->zone_fs_allowed != NULL) 24380fbb751dSJohn Levon strfree(zone->zone_fs_allowed); 24390fbb751dSJohn Levon 24400fbb751dSJohn Levon zone->zone_fs_allowed = strdup(buf); 24410fbb751dSJohn Levon 24420fbb751dSJohn Levon done: 24430fbb751dSJohn Levon kmem_free(buf, ZONE_FS_ALLOWED_MAX); 24440fbb751dSJohn Levon return (err); 24450fbb751dSJohn Levon } 24460fbb751dSJohn Levon 24473f2f09c1Sdp static int 24483f2f09c1Sdp zone_set_initname(zone_t *zone, const char *zone_initname) 24493f2f09c1Sdp { 24503f2f09c1Sdp char initname[INITNAME_SZ]; 24517c478bd9Sstevel@tonic-gate size_t len; 24523f2f09c1Sdp int err = 0; 24537c478bd9Sstevel@tonic-gate 24543f2f09c1Sdp ASSERT(zone != global_zone); 24553f2f09c1Sdp if ((err = copyinstr(zone_initname, initname, INITNAME_SZ, &len)) != 0) 24567c478bd9Sstevel@tonic-gate return (err); /* EFAULT or ENAMETOOLONG */ 24577c478bd9Sstevel@tonic-gate 24583f2f09c1Sdp if (zone->zone_initname != NULL) 24590fbb751dSJohn Levon strfree(zone->zone_initname); 24603f2f09c1Sdp 24613f2f09c1Sdp zone->zone_initname = kmem_alloc(strlen(initname) + 1, KM_SLEEP); 24623f2f09c1Sdp (void) strcpy(zone->zone_initname, initname); 24637c478bd9Sstevel@tonic-gate return (0); 24647c478bd9Sstevel@tonic-gate } 24657c478bd9Sstevel@tonic-gate 24660209230bSgjelinek static int 24670209230bSgjelinek zone_set_phys_mcap(zone_t *zone, const uint64_t *zone_mcap) 24680209230bSgjelinek { 24690209230bSgjelinek uint64_t mcap; 24700209230bSgjelinek int err = 0; 24710209230bSgjelinek 24720209230bSgjelinek if ((err = copyin(zone_mcap, &mcap, sizeof (uint64_t))) == 0) 24730209230bSgjelinek zone->zone_phys_mcap = mcap; 24740209230bSgjelinek 24750209230bSgjelinek return (err); 24760209230bSgjelinek } 24770209230bSgjelinek 24780209230bSgjelinek static int 24790209230bSgjelinek zone_set_sched_class(zone_t *zone, const char *new_class) 24800209230bSgjelinek { 24810209230bSgjelinek char sched_class[PC_CLNMSZ]; 24820209230bSgjelinek id_t classid; 24830209230bSgjelinek int err; 24840209230bSgjelinek 24850209230bSgjelinek ASSERT(zone != global_zone); 24860209230bSgjelinek if ((err = copyinstr(new_class, sched_class, PC_CLNMSZ, NULL)) != 0) 24870209230bSgjelinek return (err); /* EFAULT or ENAMETOOLONG */ 24880209230bSgjelinek 248935a5a358SJonathan Adams if (getcid(sched_class, &classid) != 0 || CLASS_KERNEL(classid)) 24900209230bSgjelinek return (set_errno(EINVAL)); 24910209230bSgjelinek zone->zone_defaultcid = classid; 24920209230bSgjelinek ASSERT(zone->zone_defaultcid > 0 && 24930209230bSgjelinek zone->zone_defaultcid < loaded_classes); 24940209230bSgjelinek 24950209230bSgjelinek return (0); 24960209230bSgjelinek } 24970209230bSgjelinek 24987c478bd9Sstevel@tonic-gate /* 24997c478bd9Sstevel@tonic-gate * Block indefinitely waiting for (zone_status >= status) 25007c478bd9Sstevel@tonic-gate */ 25017c478bd9Sstevel@tonic-gate void 25027c478bd9Sstevel@tonic-gate zone_status_wait(zone_t *zone, zone_status_t status) 25037c478bd9Sstevel@tonic-gate { 25047c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE); 25057c478bd9Sstevel@tonic-gate 25067c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 25077c478bd9Sstevel@tonic-gate while (zone->zone_status < status) { 25087c478bd9Sstevel@tonic-gate cv_wait(&zone->zone_cv, &zone_status_lock); 25097c478bd9Sstevel@tonic-gate } 25107c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 25117c478bd9Sstevel@tonic-gate } 25127c478bd9Sstevel@tonic-gate 25137c478bd9Sstevel@tonic-gate /* 25147c478bd9Sstevel@tonic-gate * Private CPR-safe version of zone_status_wait(). 25157c478bd9Sstevel@tonic-gate */ 25167c478bd9Sstevel@tonic-gate static void 25177c478bd9Sstevel@tonic-gate zone_status_wait_cpr(zone_t *zone, zone_status_t status, char *str) 25187c478bd9Sstevel@tonic-gate { 25197c478bd9Sstevel@tonic-gate callb_cpr_t cprinfo; 25207c478bd9Sstevel@tonic-gate 25217c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE); 25227c478bd9Sstevel@tonic-gate 25237c478bd9Sstevel@tonic-gate CALLB_CPR_INIT(&cprinfo, &zone_status_lock, callb_generic_cpr, 25247c478bd9Sstevel@tonic-gate str); 25257c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 25267c478bd9Sstevel@tonic-gate while (zone->zone_status < status) { 25277c478bd9Sstevel@tonic-gate CALLB_CPR_SAFE_BEGIN(&cprinfo); 25287c478bd9Sstevel@tonic-gate cv_wait(&zone->zone_cv, &zone_status_lock); 25297c478bd9Sstevel@tonic-gate CALLB_CPR_SAFE_END(&cprinfo, &zone_status_lock); 25307c478bd9Sstevel@tonic-gate } 25317c478bd9Sstevel@tonic-gate /* 25327c478bd9Sstevel@tonic-gate * zone_status_lock is implicitly released by the following. 25337c478bd9Sstevel@tonic-gate */ 25347c478bd9Sstevel@tonic-gate CALLB_CPR_EXIT(&cprinfo); 25357c478bd9Sstevel@tonic-gate } 25367c478bd9Sstevel@tonic-gate 25377c478bd9Sstevel@tonic-gate /* 25387c478bd9Sstevel@tonic-gate * Block until zone enters requested state or signal is received. Return (0) 25397c478bd9Sstevel@tonic-gate * if signaled, non-zero otherwise. 25407c478bd9Sstevel@tonic-gate */ 25417c478bd9Sstevel@tonic-gate int 25427c478bd9Sstevel@tonic-gate zone_status_wait_sig(zone_t *zone, zone_status_t status) 25437c478bd9Sstevel@tonic-gate { 25447c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE); 25457c478bd9Sstevel@tonic-gate 25467c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 25477c478bd9Sstevel@tonic-gate while (zone->zone_status < status) { 25487c478bd9Sstevel@tonic-gate if (!cv_wait_sig(&zone->zone_cv, &zone_status_lock)) { 25497c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 25507c478bd9Sstevel@tonic-gate return (0); 25517c478bd9Sstevel@tonic-gate } 25527c478bd9Sstevel@tonic-gate } 25537c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 25547c478bd9Sstevel@tonic-gate return (1); 25557c478bd9Sstevel@tonic-gate } 25567c478bd9Sstevel@tonic-gate 25577c478bd9Sstevel@tonic-gate /* 25587c478bd9Sstevel@tonic-gate * Block until the zone enters the requested state or the timeout expires, 25597c478bd9Sstevel@tonic-gate * whichever happens first. Return (-1) if operation timed out, time remaining 25607c478bd9Sstevel@tonic-gate * otherwise. 25617c478bd9Sstevel@tonic-gate */ 25627c478bd9Sstevel@tonic-gate clock_t 25637c478bd9Sstevel@tonic-gate zone_status_timedwait(zone_t *zone, clock_t tim, zone_status_t status) 25647c478bd9Sstevel@tonic-gate { 25657c478bd9Sstevel@tonic-gate clock_t timeleft = 0; 25667c478bd9Sstevel@tonic-gate 25677c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE); 25687c478bd9Sstevel@tonic-gate 25697c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 25707c478bd9Sstevel@tonic-gate while (zone->zone_status < status && timeleft != -1) { 25717c478bd9Sstevel@tonic-gate timeleft = cv_timedwait(&zone->zone_cv, &zone_status_lock, tim); 25727c478bd9Sstevel@tonic-gate } 25737c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 25747c478bd9Sstevel@tonic-gate return (timeleft); 25757c478bd9Sstevel@tonic-gate } 25767c478bd9Sstevel@tonic-gate 25777c478bd9Sstevel@tonic-gate /* 25787c478bd9Sstevel@tonic-gate * Block until the zone enters the requested state, the current process is 25797c478bd9Sstevel@tonic-gate * signaled, or the timeout expires, whichever happens first. Return (-1) if 25807c478bd9Sstevel@tonic-gate * operation timed out, 0 if signaled, time remaining otherwise. 25817c478bd9Sstevel@tonic-gate */ 25827c478bd9Sstevel@tonic-gate clock_t 25837c478bd9Sstevel@tonic-gate zone_status_timedwait_sig(zone_t *zone, clock_t tim, zone_status_t status) 25847c478bd9Sstevel@tonic-gate { 2585d3d50737SRafael Vanoni clock_t timeleft = tim - ddi_get_lbolt(); 25867c478bd9Sstevel@tonic-gate 25877c478bd9Sstevel@tonic-gate ASSERT(status > ZONE_MIN_STATE && status <= ZONE_MAX_STATE); 25887c478bd9Sstevel@tonic-gate 25897c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 25907c478bd9Sstevel@tonic-gate while (zone->zone_status < status) { 25917c478bd9Sstevel@tonic-gate timeleft = cv_timedwait_sig(&zone->zone_cv, &zone_status_lock, 25927c478bd9Sstevel@tonic-gate tim); 25937c478bd9Sstevel@tonic-gate if (timeleft <= 0) 25947c478bd9Sstevel@tonic-gate break; 25957c478bd9Sstevel@tonic-gate } 25967c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 25977c478bd9Sstevel@tonic-gate return (timeleft); 25987c478bd9Sstevel@tonic-gate } 25997c478bd9Sstevel@tonic-gate 26007c478bd9Sstevel@tonic-gate /* 26017c478bd9Sstevel@tonic-gate * Zones have two reference counts: one for references from credential 26027c478bd9Sstevel@tonic-gate * structures (zone_cred_ref), and one (zone_ref) for everything else. 26037c478bd9Sstevel@tonic-gate * This is so we can allow a zone to be rebooted while there are still 26047c478bd9Sstevel@tonic-gate * outstanding cred references, since certain drivers cache dblks (which 26057c478bd9Sstevel@tonic-gate * implicitly results in cached creds). We wait for zone_ref to drop to 26067c478bd9Sstevel@tonic-gate * 0 (actually 1), but not zone_cred_ref. The zone structure itself is 26077c478bd9Sstevel@tonic-gate * later freed when the zone_cred_ref drops to 0, though nothing other 26087c478bd9Sstevel@tonic-gate * than the zone id and privilege set should be accessed once the zone 26097c478bd9Sstevel@tonic-gate * is "dead". 26107c478bd9Sstevel@tonic-gate * 26117c478bd9Sstevel@tonic-gate * A debugging flag, zone_wait_for_cred, can be set to a non-zero value 26127c478bd9Sstevel@tonic-gate * to force halt/reboot to block waiting for the zone_cred_ref to drop 26137c478bd9Sstevel@tonic-gate * to 0. This can be useful to flush out other sources of cached creds 26147c478bd9Sstevel@tonic-gate * that may be less innocuous than the driver case. 2615a19609f8Sjv * 2616a19609f8Sjv * Zones also provide a tracked reference counting mechanism in which zone 2617a19609f8Sjv * references are represented by "crumbs" (zone_ref structures). Crumbs help 2618a19609f8Sjv * debuggers determine the sources of leaked zone references. See 2619a19609f8Sjv * zone_hold_ref() and zone_rele_ref() below for more information. 26207c478bd9Sstevel@tonic-gate */ 26217c478bd9Sstevel@tonic-gate 26227c478bd9Sstevel@tonic-gate int zone_wait_for_cred = 0; 26237c478bd9Sstevel@tonic-gate 26247c478bd9Sstevel@tonic-gate static void 26257c478bd9Sstevel@tonic-gate zone_hold_locked(zone_t *z) 26267c478bd9Sstevel@tonic-gate { 26277c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&z->zone_lock)); 26287c478bd9Sstevel@tonic-gate z->zone_ref++; 26297c478bd9Sstevel@tonic-gate ASSERT(z->zone_ref != 0); 26307c478bd9Sstevel@tonic-gate } 26317c478bd9Sstevel@tonic-gate 2632a19609f8Sjv /* 2633a19609f8Sjv * Increment the specified zone's reference count. The zone's zone_t structure 2634a19609f8Sjv * will not be freed as long as the zone's reference count is nonzero. 2635a19609f8Sjv * Decrement the zone's reference count via zone_rele(). 2636a19609f8Sjv * 2637a19609f8Sjv * NOTE: This function should only be used to hold zones for short periods of 2638a19609f8Sjv * time. Use zone_hold_ref() if the zone must be held for a long time. 2639a19609f8Sjv */ 26407c478bd9Sstevel@tonic-gate void 26417c478bd9Sstevel@tonic-gate zone_hold(zone_t *z) 26427c478bd9Sstevel@tonic-gate { 26437c478bd9Sstevel@tonic-gate mutex_enter(&z->zone_lock); 26447c478bd9Sstevel@tonic-gate zone_hold_locked(z); 26457c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 26467c478bd9Sstevel@tonic-gate } 26477c478bd9Sstevel@tonic-gate 26487c478bd9Sstevel@tonic-gate /* 26497c478bd9Sstevel@tonic-gate * If the non-cred ref count drops to 1 and either the cred ref count 26507c478bd9Sstevel@tonic-gate * is 0 or we aren't waiting for cred references, the zone is ready to 26517c478bd9Sstevel@tonic-gate * be destroyed. 26527c478bd9Sstevel@tonic-gate */ 26537c478bd9Sstevel@tonic-gate #define ZONE_IS_UNREF(zone) ((zone)->zone_ref == 1 && \ 26547c478bd9Sstevel@tonic-gate (!zone_wait_for_cred || (zone)->zone_cred_ref == 0)) 26557c478bd9Sstevel@tonic-gate 2656a19609f8Sjv /* 2657a19609f8Sjv * Common zone reference release function invoked by zone_rele() and 2658a19609f8Sjv * zone_rele_ref(). If subsys is ZONE_REF_NUM_SUBSYS, then the specified 2659a19609f8Sjv * zone's subsystem-specific reference counters are not affected by the 2660a19609f8Sjv * release. If ref is not NULL, then the zone_ref_t to which it refers is 2661a19609f8Sjv * removed from the specified zone's reference list. ref must be non-NULL iff 2662a19609f8Sjv * subsys is not ZONE_REF_NUM_SUBSYS. 2663a19609f8Sjv */ 2664a19609f8Sjv static void 2665a19609f8Sjv zone_rele_common(zone_t *z, zone_ref_t *ref, zone_ref_subsys_t subsys) 26667c478bd9Sstevel@tonic-gate { 26677c478bd9Sstevel@tonic-gate boolean_t wakeup; 26687c478bd9Sstevel@tonic-gate 26697c478bd9Sstevel@tonic-gate mutex_enter(&z->zone_lock); 26707c478bd9Sstevel@tonic-gate ASSERT(z->zone_ref != 0); 26717c478bd9Sstevel@tonic-gate z->zone_ref--; 2672a19609f8Sjv if (subsys != ZONE_REF_NUM_SUBSYS) { 2673a19609f8Sjv ASSERT(z->zone_subsys_ref[subsys] != 0); 2674a19609f8Sjv z->zone_subsys_ref[subsys]--; 2675a19609f8Sjv list_remove(&z->zone_ref_list, ref); 2676a19609f8Sjv } 26777c478bd9Sstevel@tonic-gate if (z->zone_ref == 0 && z->zone_cred_ref == 0) { 26787c478bd9Sstevel@tonic-gate /* no more refs, free the structure */ 26797c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 26807c478bd9Sstevel@tonic-gate zone_free(z); 26817c478bd9Sstevel@tonic-gate return; 26827c478bd9Sstevel@tonic-gate } 26837c478bd9Sstevel@tonic-gate /* signal zone_destroy so the zone can finish halting */ 26847c478bd9Sstevel@tonic-gate wakeup = (ZONE_IS_UNREF(z) && zone_status_get(z) >= ZONE_IS_DEAD); 26857c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 26867c478bd9Sstevel@tonic-gate 26877c478bd9Sstevel@tonic-gate if (wakeup) { 26887c478bd9Sstevel@tonic-gate /* 26897c478bd9Sstevel@tonic-gate * Grabbing zonehash_lock here effectively synchronizes with 26907c478bd9Sstevel@tonic-gate * zone_destroy() to avoid missed signals. 26917c478bd9Sstevel@tonic-gate */ 26927c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 26937c478bd9Sstevel@tonic-gate cv_broadcast(&zone_destroy_cv); 26947c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 26957c478bd9Sstevel@tonic-gate } 26967c478bd9Sstevel@tonic-gate } 26977c478bd9Sstevel@tonic-gate 2698a19609f8Sjv /* 2699a19609f8Sjv * Decrement the specified zone's reference count. The specified zone will 2700a19609f8Sjv * cease to exist after this function returns if the reference count drops to 2701a19609f8Sjv * zero. This function should be paired with zone_hold(). 2702a19609f8Sjv */ 2703a19609f8Sjv void 2704a19609f8Sjv zone_rele(zone_t *z) 2705a19609f8Sjv { 2706a19609f8Sjv zone_rele_common(z, NULL, ZONE_REF_NUM_SUBSYS); 2707a19609f8Sjv } 2708a19609f8Sjv 2709a19609f8Sjv /* 2710a19609f8Sjv * Initialize a zone reference structure. This function must be invoked for 2711a19609f8Sjv * a reference structure before the structure is passed to zone_hold_ref(). 2712a19609f8Sjv */ 2713a19609f8Sjv void 2714a19609f8Sjv zone_init_ref(zone_ref_t *ref) 2715a19609f8Sjv { 2716a19609f8Sjv ref->zref_zone = NULL; 2717a19609f8Sjv list_link_init(&ref->zref_linkage); 2718a19609f8Sjv } 2719a19609f8Sjv 2720a19609f8Sjv /* 2721a19609f8Sjv * Acquire a reference to zone z. The caller must specify the 2722a19609f8Sjv * zone_ref_subsys_t constant associated with its subsystem. The specified 2723a19609f8Sjv * zone_ref_t structure will represent a reference to the specified zone. Use 2724a19609f8Sjv * zone_rele_ref() to release the reference. 2725a19609f8Sjv * 2726a19609f8Sjv * The referenced zone_t structure will not be freed as long as the zone_t's 2727a19609f8Sjv * zone_status field is not ZONE_IS_DEAD and the zone has outstanding 2728a19609f8Sjv * references. 2729a19609f8Sjv * 2730a19609f8Sjv * NOTE: The zone_ref_t structure must be initialized before it is used. 2731a19609f8Sjv * See zone_init_ref() above. 2732a19609f8Sjv */ 2733a19609f8Sjv void 2734a19609f8Sjv zone_hold_ref(zone_t *z, zone_ref_t *ref, zone_ref_subsys_t subsys) 2735a19609f8Sjv { 2736a19609f8Sjv ASSERT(subsys >= 0 && subsys < ZONE_REF_NUM_SUBSYS); 2737a19609f8Sjv 2738a19609f8Sjv /* 2739a19609f8Sjv * Prevent consumers from reusing a reference structure before 2740a19609f8Sjv * releasing it. 2741a19609f8Sjv */ 2742a19609f8Sjv VERIFY(ref->zref_zone == NULL); 2743a19609f8Sjv 2744a19609f8Sjv ref->zref_zone = z; 2745a19609f8Sjv mutex_enter(&z->zone_lock); 2746a19609f8Sjv zone_hold_locked(z); 2747a19609f8Sjv z->zone_subsys_ref[subsys]++; 2748a19609f8Sjv ASSERT(z->zone_subsys_ref[subsys] != 0); 2749a19609f8Sjv list_insert_head(&z->zone_ref_list, ref); 2750a19609f8Sjv mutex_exit(&z->zone_lock); 2751a19609f8Sjv } 2752a19609f8Sjv 2753a19609f8Sjv /* 2754a19609f8Sjv * Release the zone reference represented by the specified zone_ref_t. 2755a19609f8Sjv * The reference is invalid after it's released; however, the zone_ref_t 2756a19609f8Sjv * structure can be reused without having to invoke zone_init_ref(). 2757a19609f8Sjv * subsys should be the same value that was passed to zone_hold_ref() 2758a19609f8Sjv * when the reference was acquired. 2759a19609f8Sjv */ 2760a19609f8Sjv void 2761a19609f8Sjv zone_rele_ref(zone_ref_t *ref, zone_ref_subsys_t subsys) 2762a19609f8Sjv { 2763a19609f8Sjv zone_rele_common(ref->zref_zone, ref, subsys); 2764a19609f8Sjv 2765a19609f8Sjv /* 2766a19609f8Sjv * Set the zone_ref_t's zref_zone field to NULL to generate panics 2767a19609f8Sjv * when consumers dereference the reference. This helps us catch 2768a19609f8Sjv * consumers who use released references. Furthermore, this lets 2769a19609f8Sjv * consumers reuse the zone_ref_t structure without having to 2770a19609f8Sjv * invoke zone_init_ref(). 2771a19609f8Sjv */ 2772a19609f8Sjv ref->zref_zone = NULL; 2773a19609f8Sjv } 2774a19609f8Sjv 27757c478bd9Sstevel@tonic-gate void 27767c478bd9Sstevel@tonic-gate zone_cred_hold(zone_t *z) 27777c478bd9Sstevel@tonic-gate { 27787c478bd9Sstevel@tonic-gate mutex_enter(&z->zone_lock); 27797c478bd9Sstevel@tonic-gate z->zone_cred_ref++; 27807c478bd9Sstevel@tonic-gate ASSERT(z->zone_cred_ref != 0); 27817c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 27827c478bd9Sstevel@tonic-gate } 27837c478bd9Sstevel@tonic-gate 27847c478bd9Sstevel@tonic-gate void 27857c478bd9Sstevel@tonic-gate zone_cred_rele(zone_t *z) 27867c478bd9Sstevel@tonic-gate { 27877c478bd9Sstevel@tonic-gate boolean_t wakeup; 27887c478bd9Sstevel@tonic-gate 27897c478bd9Sstevel@tonic-gate mutex_enter(&z->zone_lock); 27907c478bd9Sstevel@tonic-gate ASSERT(z->zone_cred_ref != 0); 27917c478bd9Sstevel@tonic-gate z->zone_cred_ref--; 27927c478bd9Sstevel@tonic-gate if (z->zone_ref == 0 && z->zone_cred_ref == 0) { 27937c478bd9Sstevel@tonic-gate /* no more refs, free the structure */ 27947c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 27957c478bd9Sstevel@tonic-gate zone_free(z); 27967c478bd9Sstevel@tonic-gate return; 27977c478bd9Sstevel@tonic-gate } 27987c478bd9Sstevel@tonic-gate /* 27997c478bd9Sstevel@tonic-gate * If zone_destroy is waiting for the cred references to drain 28007c478bd9Sstevel@tonic-gate * out, and they have, signal it. 28017c478bd9Sstevel@tonic-gate */ 28027c478bd9Sstevel@tonic-gate wakeup = (zone_wait_for_cred && ZONE_IS_UNREF(z) && 28037c478bd9Sstevel@tonic-gate zone_status_get(z) >= ZONE_IS_DEAD); 28047c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 28057c478bd9Sstevel@tonic-gate 28067c478bd9Sstevel@tonic-gate if (wakeup) { 28077c478bd9Sstevel@tonic-gate /* 28087c478bd9Sstevel@tonic-gate * Grabbing zonehash_lock here effectively synchronizes with 28097c478bd9Sstevel@tonic-gate * zone_destroy() to avoid missed signals. 28107c478bd9Sstevel@tonic-gate */ 28117c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 28127c478bd9Sstevel@tonic-gate cv_broadcast(&zone_destroy_cv); 28137c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 28147c478bd9Sstevel@tonic-gate } 28157c478bd9Sstevel@tonic-gate } 28167c478bd9Sstevel@tonic-gate 28177c478bd9Sstevel@tonic-gate void 28187c478bd9Sstevel@tonic-gate zone_task_hold(zone_t *z) 28197c478bd9Sstevel@tonic-gate { 28207c478bd9Sstevel@tonic-gate mutex_enter(&z->zone_lock); 28217c478bd9Sstevel@tonic-gate z->zone_ntasks++; 28227c478bd9Sstevel@tonic-gate ASSERT(z->zone_ntasks != 0); 28237c478bd9Sstevel@tonic-gate mutex_exit(&z->zone_lock); 28247c478bd9Sstevel@tonic-gate } 28257c478bd9Sstevel@tonic-gate 28267c478bd9Sstevel@tonic-gate void 28277c478bd9Sstevel@tonic-gate zone_task_rele(zone_t *zone) 28287c478bd9Sstevel@tonic-gate { 28297c478bd9Sstevel@tonic-gate uint_t refcnt; 28307c478bd9Sstevel@tonic-gate 28317c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 28327c478bd9Sstevel@tonic-gate ASSERT(zone->zone_ntasks != 0); 28337c478bd9Sstevel@tonic-gate refcnt = --zone->zone_ntasks; 28347c478bd9Sstevel@tonic-gate if (refcnt > 1) { /* Common case */ 28357c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 28367c478bd9Sstevel@tonic-gate return; 28377c478bd9Sstevel@tonic-gate } 28387c478bd9Sstevel@tonic-gate zone_hold_locked(zone); /* so we can use the zone_t later */ 28397c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 28407c478bd9Sstevel@tonic-gate if (refcnt == 1) { 28417c478bd9Sstevel@tonic-gate /* 28427c478bd9Sstevel@tonic-gate * See if the zone is shutting down. 28437c478bd9Sstevel@tonic-gate */ 28447c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 28457c478bd9Sstevel@tonic-gate if (zone_status_get(zone) != ZONE_IS_SHUTTING_DOWN) { 28467c478bd9Sstevel@tonic-gate goto out; 28477c478bd9Sstevel@tonic-gate } 28487c478bd9Sstevel@tonic-gate 28497c478bd9Sstevel@tonic-gate /* 28507c478bd9Sstevel@tonic-gate * Make sure the ntasks didn't change since we 28517c478bd9Sstevel@tonic-gate * dropped zone_lock. 28527c478bd9Sstevel@tonic-gate */ 28537c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 28547c478bd9Sstevel@tonic-gate if (refcnt != zone->zone_ntasks) { 28557c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 28567c478bd9Sstevel@tonic-gate goto out; 28577c478bd9Sstevel@tonic-gate } 28587c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 28597c478bd9Sstevel@tonic-gate 28607c478bd9Sstevel@tonic-gate /* 28617c478bd9Sstevel@tonic-gate * No more user processes in the zone. The zone is empty. 28627c478bd9Sstevel@tonic-gate */ 28637c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_EMPTY); 28647c478bd9Sstevel@tonic-gate goto out; 28657c478bd9Sstevel@tonic-gate } 28667c478bd9Sstevel@tonic-gate 28677c478bd9Sstevel@tonic-gate ASSERT(refcnt == 0); 28687c478bd9Sstevel@tonic-gate /* 28697c478bd9Sstevel@tonic-gate * zsched has exited; the zone is dead. 28707c478bd9Sstevel@tonic-gate */ 28717c478bd9Sstevel@tonic-gate zone->zone_zsched = NULL; /* paranoia */ 28727c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 28737c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_DEAD); 28747c478bd9Sstevel@tonic-gate out: 28757c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 28767c478bd9Sstevel@tonic-gate zone_rele(zone); 28777c478bd9Sstevel@tonic-gate } 28787c478bd9Sstevel@tonic-gate 28797c478bd9Sstevel@tonic-gate zoneid_t 28807c478bd9Sstevel@tonic-gate getzoneid(void) 28817c478bd9Sstevel@tonic-gate { 28827c478bd9Sstevel@tonic-gate return (curproc->p_zone->zone_id); 28837c478bd9Sstevel@tonic-gate } 28847c478bd9Sstevel@tonic-gate 28857c478bd9Sstevel@tonic-gate /* 28867c478bd9Sstevel@tonic-gate * Internal versions of zone_find_by_*(). These don't zone_hold() or 28877c478bd9Sstevel@tonic-gate * check the validity of a zone's state. 28887c478bd9Sstevel@tonic-gate */ 28897c478bd9Sstevel@tonic-gate static zone_t * 28907c478bd9Sstevel@tonic-gate zone_find_all_by_id(zoneid_t zoneid) 28917c478bd9Sstevel@tonic-gate { 28927c478bd9Sstevel@tonic-gate mod_hash_val_t hv; 28937c478bd9Sstevel@tonic-gate zone_t *zone = NULL; 28947c478bd9Sstevel@tonic-gate 28957c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 28967c478bd9Sstevel@tonic-gate 28977c478bd9Sstevel@tonic-gate if (mod_hash_find(zonehashbyid, 28987c478bd9Sstevel@tonic-gate (mod_hash_key_t)(uintptr_t)zoneid, &hv) == 0) 28997c478bd9Sstevel@tonic-gate zone = (zone_t *)hv; 29007c478bd9Sstevel@tonic-gate return (zone); 29017c478bd9Sstevel@tonic-gate } 29027c478bd9Sstevel@tonic-gate 290345916cd2Sjpk static zone_t * 290445916cd2Sjpk zone_find_all_by_label(const ts_label_t *label) 290545916cd2Sjpk { 290645916cd2Sjpk mod_hash_val_t hv; 290745916cd2Sjpk zone_t *zone = NULL; 290845916cd2Sjpk 290945916cd2Sjpk ASSERT(MUTEX_HELD(&zonehash_lock)); 291045916cd2Sjpk 291145916cd2Sjpk /* 291245916cd2Sjpk * zonehashbylabel is not maintained for unlabeled systems 291345916cd2Sjpk */ 291445916cd2Sjpk if (!is_system_labeled()) 291545916cd2Sjpk return (NULL); 291645916cd2Sjpk if (mod_hash_find(zonehashbylabel, (mod_hash_key_t)label, &hv) == 0) 291745916cd2Sjpk zone = (zone_t *)hv; 291845916cd2Sjpk return (zone); 291945916cd2Sjpk } 292045916cd2Sjpk 29217c478bd9Sstevel@tonic-gate static zone_t * 29227c478bd9Sstevel@tonic-gate zone_find_all_by_name(char *name) 29237c478bd9Sstevel@tonic-gate { 29247c478bd9Sstevel@tonic-gate mod_hash_val_t hv; 29257c478bd9Sstevel@tonic-gate zone_t *zone = NULL; 29267c478bd9Sstevel@tonic-gate 29277c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 29287c478bd9Sstevel@tonic-gate 29297c478bd9Sstevel@tonic-gate if (mod_hash_find(zonehashbyname, (mod_hash_key_t)name, &hv) == 0) 29307c478bd9Sstevel@tonic-gate zone = (zone_t *)hv; 29317c478bd9Sstevel@tonic-gate return (zone); 29327c478bd9Sstevel@tonic-gate } 29337c478bd9Sstevel@tonic-gate 29347c478bd9Sstevel@tonic-gate /* 29357c478bd9Sstevel@tonic-gate * Public interface for looking up a zone by zoneid. Only returns the zone if 29367c478bd9Sstevel@tonic-gate * it is fully initialized, and has not yet begun the zone_destroy() sequence. 29377c478bd9Sstevel@tonic-gate * Caller must call zone_rele() once it is done with the zone. 29387c478bd9Sstevel@tonic-gate * 29397c478bd9Sstevel@tonic-gate * The zone may begin the zone_destroy() sequence immediately after this 29407c478bd9Sstevel@tonic-gate * function returns, but may be safely used until zone_rele() is called. 29417c478bd9Sstevel@tonic-gate */ 29427c478bd9Sstevel@tonic-gate zone_t * 29437c478bd9Sstevel@tonic-gate zone_find_by_id(zoneid_t zoneid) 29447c478bd9Sstevel@tonic-gate { 29457c478bd9Sstevel@tonic-gate zone_t *zone; 29467c478bd9Sstevel@tonic-gate zone_status_t status; 29477c478bd9Sstevel@tonic-gate 29487c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 29497c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 29507c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 29517c478bd9Sstevel@tonic-gate return (NULL); 29527c478bd9Sstevel@tonic-gate } 29537c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 29547c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY || status > ZONE_IS_DOWN) { 29557c478bd9Sstevel@tonic-gate /* 29567c478bd9Sstevel@tonic-gate * For all practical purposes the zone doesn't exist. 29577c478bd9Sstevel@tonic-gate */ 29587c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 29597c478bd9Sstevel@tonic-gate return (NULL); 29607c478bd9Sstevel@tonic-gate } 29617c478bd9Sstevel@tonic-gate zone_hold(zone); 29627c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 29637c478bd9Sstevel@tonic-gate return (zone); 29647c478bd9Sstevel@tonic-gate } 29657c478bd9Sstevel@tonic-gate 296645916cd2Sjpk /* 296745916cd2Sjpk * Similar to zone_find_by_id, but using zone label as the key. 296845916cd2Sjpk */ 296945916cd2Sjpk zone_t * 297045916cd2Sjpk zone_find_by_label(const ts_label_t *label) 297145916cd2Sjpk { 297245916cd2Sjpk zone_t *zone; 297342bc57c4Srica zone_status_t status; 297445916cd2Sjpk 297545916cd2Sjpk mutex_enter(&zonehash_lock); 297645916cd2Sjpk if ((zone = zone_find_all_by_label(label)) == NULL) { 297745916cd2Sjpk mutex_exit(&zonehash_lock); 297845916cd2Sjpk return (NULL); 297945916cd2Sjpk } 298042bc57c4Srica 298142bc57c4Srica status = zone_status_get(zone); 298242bc57c4Srica if (status > ZONE_IS_DOWN) { 298345916cd2Sjpk /* 298445916cd2Sjpk * For all practical purposes the zone doesn't exist. 298545916cd2Sjpk */ 298642bc57c4Srica mutex_exit(&zonehash_lock); 298742bc57c4Srica return (NULL); 298845916cd2Sjpk } 298942bc57c4Srica zone_hold(zone); 299045916cd2Sjpk mutex_exit(&zonehash_lock); 299145916cd2Sjpk return (zone); 299245916cd2Sjpk } 299345916cd2Sjpk 29947c478bd9Sstevel@tonic-gate /* 29957c478bd9Sstevel@tonic-gate * Similar to zone_find_by_id, but using zone name as the key. 29967c478bd9Sstevel@tonic-gate */ 29977c478bd9Sstevel@tonic-gate zone_t * 29987c478bd9Sstevel@tonic-gate zone_find_by_name(char *name) 29997c478bd9Sstevel@tonic-gate { 30007c478bd9Sstevel@tonic-gate zone_t *zone; 30017c478bd9Sstevel@tonic-gate zone_status_t status; 30027c478bd9Sstevel@tonic-gate 30037c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 30047c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_name(name)) == NULL) { 30057c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 30067c478bd9Sstevel@tonic-gate return (NULL); 30077c478bd9Sstevel@tonic-gate } 30087c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 30097c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY || status > ZONE_IS_DOWN) { 30107c478bd9Sstevel@tonic-gate /* 30117c478bd9Sstevel@tonic-gate * For all practical purposes the zone doesn't exist. 30127c478bd9Sstevel@tonic-gate */ 30137c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 30147c478bd9Sstevel@tonic-gate return (NULL); 30157c478bd9Sstevel@tonic-gate } 30167c478bd9Sstevel@tonic-gate zone_hold(zone); 30177c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 30187c478bd9Sstevel@tonic-gate return (zone); 30197c478bd9Sstevel@tonic-gate } 30207c478bd9Sstevel@tonic-gate 30217c478bd9Sstevel@tonic-gate /* 30227c478bd9Sstevel@tonic-gate * Similar to zone_find_by_id(), using the path as a key. For instance, 30237c478bd9Sstevel@tonic-gate * if there is a zone "foo" rooted at /foo/root, and the path argument 30247c478bd9Sstevel@tonic-gate * is "/foo/root/proc", it will return the held zone_t corresponding to 30257c478bd9Sstevel@tonic-gate * zone "foo". 30267c478bd9Sstevel@tonic-gate * 30277c478bd9Sstevel@tonic-gate * zone_find_by_path() always returns a non-NULL value, since at the 30287c478bd9Sstevel@tonic-gate * very least every path will be contained in the global zone. 30297c478bd9Sstevel@tonic-gate * 30307c478bd9Sstevel@tonic-gate * As with the other zone_find_by_*() functions, the caller is 30317c478bd9Sstevel@tonic-gate * responsible for zone_rele()ing the return value of this function. 30327c478bd9Sstevel@tonic-gate */ 30337c478bd9Sstevel@tonic-gate zone_t * 30347c478bd9Sstevel@tonic-gate zone_find_by_path(const char *path) 30357c478bd9Sstevel@tonic-gate { 30367c478bd9Sstevel@tonic-gate zone_t *zone; 30377c478bd9Sstevel@tonic-gate zone_t *zret = NULL; 30387c478bd9Sstevel@tonic-gate zone_status_t status; 30397c478bd9Sstevel@tonic-gate 30407c478bd9Sstevel@tonic-gate if (path == NULL) { 30417c478bd9Sstevel@tonic-gate /* 30427c478bd9Sstevel@tonic-gate * Call from rootconf(). 30437c478bd9Sstevel@tonic-gate */ 30447c478bd9Sstevel@tonic-gate zone_hold(global_zone); 30457c478bd9Sstevel@tonic-gate return (global_zone); 30467c478bd9Sstevel@tonic-gate } 30477c478bd9Sstevel@tonic-gate ASSERT(*path == '/'); 30487c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 30497c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 30507c478bd9Sstevel@tonic-gate zone = list_next(&zone_active, zone)) { 30517c478bd9Sstevel@tonic-gate if (ZONE_PATH_VISIBLE(path, zone)) 30527c478bd9Sstevel@tonic-gate zret = zone; 30537c478bd9Sstevel@tonic-gate } 30547c478bd9Sstevel@tonic-gate ASSERT(zret != NULL); 30557c478bd9Sstevel@tonic-gate status = zone_status_get(zret); 30567c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY || status > ZONE_IS_DOWN) { 30577c478bd9Sstevel@tonic-gate /* 30587c478bd9Sstevel@tonic-gate * Zone practically doesn't exist. 30597c478bd9Sstevel@tonic-gate */ 30607c478bd9Sstevel@tonic-gate zret = global_zone; 30617c478bd9Sstevel@tonic-gate } 30627c478bd9Sstevel@tonic-gate zone_hold(zret); 30637c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 30647c478bd9Sstevel@tonic-gate return (zret); 30657c478bd9Sstevel@tonic-gate } 30667c478bd9Sstevel@tonic-gate 306781d43577SJerry Jelinek /* 306881d43577SJerry Jelinek * Public interface for updating per-zone load averages. Called once per 306981d43577SJerry Jelinek * second. 307081d43577SJerry Jelinek * 307181d43577SJerry Jelinek * Based on loadavg_update(), genloadavg() and calcloadavg() from clock.c. 307281d43577SJerry Jelinek */ 307381d43577SJerry Jelinek void 307481d43577SJerry Jelinek zone_loadavg_update() 307581d43577SJerry Jelinek { 307681d43577SJerry Jelinek zone_t *zp; 307781d43577SJerry Jelinek zone_status_t status; 307881d43577SJerry Jelinek struct loadavg_s *lavg; 307981d43577SJerry Jelinek hrtime_t zone_total; 308081d43577SJerry Jelinek int i; 308181d43577SJerry Jelinek hrtime_t hr_avg; 308281d43577SJerry Jelinek int nrun; 308381d43577SJerry Jelinek static int64_t f[3] = { 135, 27, 9 }; 308481d43577SJerry Jelinek int64_t q, r; 308581d43577SJerry Jelinek 308681d43577SJerry Jelinek mutex_enter(&zonehash_lock); 308781d43577SJerry Jelinek for (zp = list_head(&zone_active); zp != NULL; 308881d43577SJerry Jelinek zp = list_next(&zone_active, zp)) { 308981d43577SJerry Jelinek mutex_enter(&zp->zone_lock); 309081d43577SJerry Jelinek 309181d43577SJerry Jelinek /* Skip zones that are on the way down or not yet up */ 309281d43577SJerry Jelinek status = zone_status_get(zp); 309381d43577SJerry Jelinek if (status < ZONE_IS_READY || status >= ZONE_IS_DOWN) { 309481d43577SJerry Jelinek /* For all practical purposes the zone doesn't exist. */ 309581d43577SJerry Jelinek mutex_exit(&zp->zone_lock); 309681d43577SJerry Jelinek continue; 309781d43577SJerry Jelinek } 309881d43577SJerry Jelinek 309981d43577SJerry Jelinek /* 310081d43577SJerry Jelinek * Update the 10 second moving average data in zone_loadavg. 310181d43577SJerry Jelinek */ 310281d43577SJerry Jelinek lavg = &zp->zone_loadavg; 310381d43577SJerry Jelinek 310481d43577SJerry Jelinek zone_total = zp->zone_utime + zp->zone_stime + zp->zone_wtime; 310581d43577SJerry Jelinek scalehrtime(&zone_total); 310681d43577SJerry Jelinek 310781d43577SJerry Jelinek /* The zone_total should always be increasing. */ 310881d43577SJerry Jelinek lavg->lg_loads[lavg->lg_cur] = (zone_total > lavg->lg_total) ? 310981d43577SJerry Jelinek zone_total - lavg->lg_total : 0; 311081d43577SJerry Jelinek lavg->lg_cur = (lavg->lg_cur + 1) % S_LOADAVG_SZ; 311181d43577SJerry Jelinek /* lg_total holds the prev. 1 sec. total */ 311281d43577SJerry Jelinek lavg->lg_total = zone_total; 311381d43577SJerry Jelinek 311481d43577SJerry Jelinek /* 311581d43577SJerry Jelinek * To simplify the calculation, we don't calculate the load avg. 311681d43577SJerry Jelinek * until the zone has been up for at least 10 seconds and our 311781d43577SJerry Jelinek * moving average is thus full. 311881d43577SJerry Jelinek */ 311981d43577SJerry Jelinek if ((lavg->lg_len + 1) < S_LOADAVG_SZ) { 312081d43577SJerry Jelinek lavg->lg_len++; 312181d43577SJerry Jelinek mutex_exit(&zp->zone_lock); 312281d43577SJerry Jelinek continue; 312381d43577SJerry Jelinek } 312481d43577SJerry Jelinek 312581d43577SJerry Jelinek /* Now calculate the 1min, 5min, 15 min load avg. */ 312681d43577SJerry Jelinek hr_avg = 0; 312781d43577SJerry Jelinek for (i = 0; i < S_LOADAVG_SZ; i++) 312881d43577SJerry Jelinek hr_avg += lavg->lg_loads[i]; 312981d43577SJerry Jelinek hr_avg = hr_avg / S_LOADAVG_SZ; 313081d43577SJerry Jelinek nrun = hr_avg / (NANOSEC / LGRP_LOADAVG_IN_THREAD_MAX); 313181d43577SJerry Jelinek 313281d43577SJerry Jelinek /* Compute load avg. See comment in calcloadavg() */ 313381d43577SJerry Jelinek for (i = 0; i < 3; i++) { 313481d43577SJerry Jelinek q = (zp->zone_hp_avenrun[i] >> 16) << 7; 313581d43577SJerry Jelinek r = (zp->zone_hp_avenrun[i] & 0xffff) << 7; 313681d43577SJerry Jelinek zp->zone_hp_avenrun[i] += 313781d43577SJerry Jelinek ((nrun - q) * f[i] - ((r * f[i]) >> 16)) >> 4; 313881d43577SJerry Jelinek 313981d43577SJerry Jelinek /* avenrun[] can only hold 31 bits of load avg. */ 314081d43577SJerry Jelinek if (zp->zone_hp_avenrun[i] < 314181d43577SJerry Jelinek ((uint64_t)1<<(31+16-FSHIFT))) 314281d43577SJerry Jelinek zp->zone_avenrun[i] = (int32_t) 314381d43577SJerry Jelinek (zp->zone_hp_avenrun[i] >> (16 - FSHIFT)); 314481d43577SJerry Jelinek else 314581d43577SJerry Jelinek zp->zone_avenrun[i] = 0x7fffffff; 314681d43577SJerry Jelinek } 314781d43577SJerry Jelinek 314881d43577SJerry Jelinek mutex_exit(&zp->zone_lock); 314981d43577SJerry Jelinek } 315081d43577SJerry Jelinek mutex_exit(&zonehash_lock); 315181d43577SJerry Jelinek } 315281d43577SJerry Jelinek 31537c478bd9Sstevel@tonic-gate /* 31547c478bd9Sstevel@tonic-gate * Get the number of cpus visible to this zone. The system-wide global 31557c478bd9Sstevel@tonic-gate * 'ncpus' is returned if pools are disabled, the caller is in the 31567c478bd9Sstevel@tonic-gate * global zone, or a NULL zone argument is passed in. 31577c478bd9Sstevel@tonic-gate */ 31587c478bd9Sstevel@tonic-gate int 31597c478bd9Sstevel@tonic-gate zone_ncpus_get(zone_t *zone) 31607c478bd9Sstevel@tonic-gate { 31617c478bd9Sstevel@tonic-gate int myncpus = zone == NULL ? 0 : zone->zone_ncpus; 31627c478bd9Sstevel@tonic-gate 31637c478bd9Sstevel@tonic-gate return (myncpus != 0 ? myncpus : ncpus); 31647c478bd9Sstevel@tonic-gate } 31657c478bd9Sstevel@tonic-gate 31667c478bd9Sstevel@tonic-gate /* 31677c478bd9Sstevel@tonic-gate * Get the number of online cpus visible to this zone. The system-wide 31687c478bd9Sstevel@tonic-gate * global 'ncpus_online' is returned if pools are disabled, the caller 31697c478bd9Sstevel@tonic-gate * is in the global zone, or a NULL zone argument is passed in. 31707c478bd9Sstevel@tonic-gate */ 31717c478bd9Sstevel@tonic-gate int 31727c478bd9Sstevel@tonic-gate zone_ncpus_online_get(zone_t *zone) 31737c478bd9Sstevel@tonic-gate { 31747c478bd9Sstevel@tonic-gate int myncpus_online = zone == NULL ? 0 : zone->zone_ncpus_online; 31757c478bd9Sstevel@tonic-gate 31767c478bd9Sstevel@tonic-gate return (myncpus_online != 0 ? myncpus_online : ncpus_online); 31777c478bd9Sstevel@tonic-gate } 31787c478bd9Sstevel@tonic-gate 31797c478bd9Sstevel@tonic-gate /* 31807c478bd9Sstevel@tonic-gate * Return the pool to which the zone is currently bound. 31817c478bd9Sstevel@tonic-gate */ 31827c478bd9Sstevel@tonic-gate pool_t * 31837c478bd9Sstevel@tonic-gate zone_pool_get(zone_t *zone) 31847c478bd9Sstevel@tonic-gate { 31857c478bd9Sstevel@tonic-gate ASSERT(pool_lock_held()); 31867c478bd9Sstevel@tonic-gate 31877c478bd9Sstevel@tonic-gate return (zone->zone_pool); 31887c478bd9Sstevel@tonic-gate } 31897c478bd9Sstevel@tonic-gate 31907c478bd9Sstevel@tonic-gate /* 31917c478bd9Sstevel@tonic-gate * Set the zone's pool pointer and update the zone's visibility to match 31927c478bd9Sstevel@tonic-gate * the resources in the new pool. 31937c478bd9Sstevel@tonic-gate */ 31947c478bd9Sstevel@tonic-gate void 31957c478bd9Sstevel@tonic-gate zone_pool_set(zone_t *zone, pool_t *pool) 31967c478bd9Sstevel@tonic-gate { 31977c478bd9Sstevel@tonic-gate ASSERT(pool_lock_held()); 31987c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&cpu_lock)); 31997c478bd9Sstevel@tonic-gate 32007c478bd9Sstevel@tonic-gate zone->zone_pool = pool; 32017c478bd9Sstevel@tonic-gate zone_pset_set(zone, pool->pool_pset->pset_id); 32027c478bd9Sstevel@tonic-gate } 32037c478bd9Sstevel@tonic-gate 32047c478bd9Sstevel@tonic-gate /* 32057c478bd9Sstevel@tonic-gate * Return the cached value of the id of the processor set to which the 32067c478bd9Sstevel@tonic-gate * zone is currently bound. The value will be ZONE_PS_INVAL if the pools 32077c478bd9Sstevel@tonic-gate * facility is disabled. 32087c478bd9Sstevel@tonic-gate */ 32097c478bd9Sstevel@tonic-gate psetid_t 32107c478bd9Sstevel@tonic-gate zone_pset_get(zone_t *zone) 32117c478bd9Sstevel@tonic-gate { 32127c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&cpu_lock)); 32137c478bd9Sstevel@tonic-gate 32147c478bd9Sstevel@tonic-gate return (zone->zone_psetid); 32157c478bd9Sstevel@tonic-gate } 32167c478bd9Sstevel@tonic-gate 32177c478bd9Sstevel@tonic-gate /* 32187c478bd9Sstevel@tonic-gate * Set the cached value of the id of the processor set to which the zone 32197c478bd9Sstevel@tonic-gate * is currently bound. Also update the zone's visibility to match the 32207c478bd9Sstevel@tonic-gate * resources in the new processor set. 32217c478bd9Sstevel@tonic-gate */ 32227c478bd9Sstevel@tonic-gate void 32237c478bd9Sstevel@tonic-gate zone_pset_set(zone_t *zone, psetid_t newpsetid) 32247c478bd9Sstevel@tonic-gate { 32257c478bd9Sstevel@tonic-gate psetid_t oldpsetid; 32267c478bd9Sstevel@tonic-gate 32277c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&cpu_lock)); 32287c478bd9Sstevel@tonic-gate oldpsetid = zone_pset_get(zone); 32297c478bd9Sstevel@tonic-gate 32307c478bd9Sstevel@tonic-gate if (oldpsetid == newpsetid) 32317c478bd9Sstevel@tonic-gate return; 32327c478bd9Sstevel@tonic-gate /* 32337c478bd9Sstevel@tonic-gate * Global zone sees all. 32347c478bd9Sstevel@tonic-gate */ 32357c478bd9Sstevel@tonic-gate if (zone != global_zone) { 32367c478bd9Sstevel@tonic-gate zone->zone_psetid = newpsetid; 32377c478bd9Sstevel@tonic-gate if (newpsetid != ZONE_PS_INVAL) 32387c478bd9Sstevel@tonic-gate pool_pset_visibility_add(newpsetid, zone); 32397c478bd9Sstevel@tonic-gate if (oldpsetid != ZONE_PS_INVAL) 32407c478bd9Sstevel@tonic-gate pool_pset_visibility_remove(oldpsetid, zone); 32417c478bd9Sstevel@tonic-gate } 32427c478bd9Sstevel@tonic-gate /* 32437c478bd9Sstevel@tonic-gate * Disabling pools, so we should start using the global values 32447c478bd9Sstevel@tonic-gate * for ncpus and ncpus_online. 32457c478bd9Sstevel@tonic-gate */ 32467c478bd9Sstevel@tonic-gate if (newpsetid == ZONE_PS_INVAL) { 32477c478bd9Sstevel@tonic-gate zone->zone_ncpus = 0; 32487c478bd9Sstevel@tonic-gate zone->zone_ncpus_online = 0; 32497c478bd9Sstevel@tonic-gate } 32507c478bd9Sstevel@tonic-gate } 32517c478bd9Sstevel@tonic-gate 32527c478bd9Sstevel@tonic-gate /* 32537c478bd9Sstevel@tonic-gate * Walk the list of active zones and issue the provided callback for 32547c478bd9Sstevel@tonic-gate * each of them. 32557c478bd9Sstevel@tonic-gate * 32567c478bd9Sstevel@tonic-gate * Caller must not be holding any locks that may be acquired under 32577c478bd9Sstevel@tonic-gate * zonehash_lock. See comment at the beginning of the file for a list of 32587c478bd9Sstevel@tonic-gate * common locks and their interactions with zones. 32597c478bd9Sstevel@tonic-gate */ 32607c478bd9Sstevel@tonic-gate int 32617c478bd9Sstevel@tonic-gate zone_walk(int (*cb)(zone_t *, void *), void *data) 32627c478bd9Sstevel@tonic-gate { 32637c478bd9Sstevel@tonic-gate zone_t *zone; 32647c478bd9Sstevel@tonic-gate int ret = 0; 32657c478bd9Sstevel@tonic-gate zone_status_t status; 32667c478bd9Sstevel@tonic-gate 32677c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 32687c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 32697c478bd9Sstevel@tonic-gate zone = list_next(&zone_active, zone)) { 32707c478bd9Sstevel@tonic-gate /* 32717c478bd9Sstevel@tonic-gate * Skip zones that shouldn't be externally visible. 32727c478bd9Sstevel@tonic-gate */ 32737c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 32747c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY || status > ZONE_IS_DOWN) 32757c478bd9Sstevel@tonic-gate continue; 32767c478bd9Sstevel@tonic-gate /* 32777c478bd9Sstevel@tonic-gate * Bail immediately if any callback invocation returns a 32787c478bd9Sstevel@tonic-gate * non-zero value. 32797c478bd9Sstevel@tonic-gate */ 32807c478bd9Sstevel@tonic-gate ret = (*cb)(zone, data); 32817c478bd9Sstevel@tonic-gate if (ret != 0) 32827c478bd9Sstevel@tonic-gate break; 32837c478bd9Sstevel@tonic-gate } 32847c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 32857c478bd9Sstevel@tonic-gate return (ret); 32867c478bd9Sstevel@tonic-gate } 32877c478bd9Sstevel@tonic-gate 32887c478bd9Sstevel@tonic-gate static int 32897c478bd9Sstevel@tonic-gate zone_set_root(zone_t *zone, const char *upath) 32907c478bd9Sstevel@tonic-gate { 32917c478bd9Sstevel@tonic-gate vnode_t *vp; 32927c478bd9Sstevel@tonic-gate int trycount; 32937c478bd9Sstevel@tonic-gate int error = 0; 32947c478bd9Sstevel@tonic-gate char *path; 32957c478bd9Sstevel@tonic-gate struct pathname upn, pn; 32967c478bd9Sstevel@tonic-gate size_t pathlen; 32977c478bd9Sstevel@tonic-gate 32987c478bd9Sstevel@tonic-gate if ((error = pn_get((char *)upath, UIO_USERSPACE, &upn)) != 0) 32997c478bd9Sstevel@tonic-gate return (error); 33007c478bd9Sstevel@tonic-gate 33017c478bd9Sstevel@tonic-gate pn_alloc(&pn); 33027c478bd9Sstevel@tonic-gate 33037c478bd9Sstevel@tonic-gate /* prevent infinite loop */ 33047c478bd9Sstevel@tonic-gate trycount = 10; 33057c478bd9Sstevel@tonic-gate for (;;) { 33067c478bd9Sstevel@tonic-gate if (--trycount <= 0) { 33077c478bd9Sstevel@tonic-gate error = ESTALE; 33087c478bd9Sstevel@tonic-gate goto out; 33097c478bd9Sstevel@tonic-gate } 33107c478bd9Sstevel@tonic-gate 33117c478bd9Sstevel@tonic-gate if ((error = lookuppn(&upn, &pn, FOLLOW, NULLVPP, &vp)) == 0) { 33127c478bd9Sstevel@tonic-gate /* 33137c478bd9Sstevel@tonic-gate * VOP_ACCESS() may cover 'vp' with a new 33147c478bd9Sstevel@tonic-gate * filesystem, if 'vp' is an autoFS vnode. 33157c478bd9Sstevel@tonic-gate * Get the new 'vp' if so. 33167c478bd9Sstevel@tonic-gate */ 3317da6c28aaSamw if ((error = 3318da6c28aaSamw VOP_ACCESS(vp, VEXEC, 0, CRED(), NULL)) == 0 && 331925d2dc23Seh (!vn_ismntpt(vp) || 33207c478bd9Sstevel@tonic-gate (error = traverse(&vp)) == 0)) { 33217c478bd9Sstevel@tonic-gate pathlen = pn.pn_pathlen + 2; 33227c478bd9Sstevel@tonic-gate path = kmem_alloc(pathlen, KM_SLEEP); 33237c478bd9Sstevel@tonic-gate (void) strncpy(path, pn.pn_path, 33247c478bd9Sstevel@tonic-gate pn.pn_pathlen + 1); 33257c478bd9Sstevel@tonic-gate path[pathlen - 2] = '/'; 33267c478bd9Sstevel@tonic-gate path[pathlen - 1] = '\0'; 33277c478bd9Sstevel@tonic-gate pn_free(&pn); 33287c478bd9Sstevel@tonic-gate pn_free(&upn); 33297c478bd9Sstevel@tonic-gate 33307c478bd9Sstevel@tonic-gate /* Success! */ 33317c478bd9Sstevel@tonic-gate break; 33327c478bd9Sstevel@tonic-gate } 33337c478bd9Sstevel@tonic-gate VN_RELE(vp); 33347c478bd9Sstevel@tonic-gate } 33357c478bd9Sstevel@tonic-gate if (error != ESTALE) 33367c478bd9Sstevel@tonic-gate goto out; 33377c478bd9Sstevel@tonic-gate } 33387c478bd9Sstevel@tonic-gate 33397c478bd9Sstevel@tonic-gate ASSERT(error == 0); 33407c478bd9Sstevel@tonic-gate zone->zone_rootvp = vp; /* we hold a reference to vp */ 33417c478bd9Sstevel@tonic-gate zone->zone_rootpath = path; 33427c478bd9Sstevel@tonic-gate zone->zone_rootpathlen = pathlen; 334348451833Scarlsonj if (pathlen > 5 && strcmp(path + pathlen - 5, "/lu/") == 0) 334448451833Scarlsonj zone->zone_flags |= ZF_IS_SCRATCH; 33457c478bd9Sstevel@tonic-gate return (0); 33467c478bd9Sstevel@tonic-gate 33477c478bd9Sstevel@tonic-gate out: 33487c478bd9Sstevel@tonic-gate pn_free(&pn); 33497c478bd9Sstevel@tonic-gate pn_free(&upn); 33507c478bd9Sstevel@tonic-gate return (error); 33517c478bd9Sstevel@tonic-gate } 33527c478bd9Sstevel@tonic-gate 33537c478bd9Sstevel@tonic-gate #define isalnum(c) (((c) >= '0' && (c) <= '9') || \ 33547c478bd9Sstevel@tonic-gate ((c) >= 'a' && (c) <= 'z') || \ 33557c478bd9Sstevel@tonic-gate ((c) >= 'A' && (c) <= 'Z')) 33567c478bd9Sstevel@tonic-gate 33577c478bd9Sstevel@tonic-gate static int 33587c478bd9Sstevel@tonic-gate zone_set_name(zone_t *zone, const char *uname) 33597c478bd9Sstevel@tonic-gate { 33607c478bd9Sstevel@tonic-gate char *kname = kmem_zalloc(ZONENAME_MAX, KM_SLEEP); 33617c478bd9Sstevel@tonic-gate size_t len; 33627c478bd9Sstevel@tonic-gate int i, err; 33637c478bd9Sstevel@tonic-gate 33647c478bd9Sstevel@tonic-gate if ((err = copyinstr(uname, kname, ZONENAME_MAX, &len)) != 0) { 33657c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 33667c478bd9Sstevel@tonic-gate return (err); /* EFAULT or ENAMETOOLONG */ 33677c478bd9Sstevel@tonic-gate } 33687c478bd9Sstevel@tonic-gate 33697c478bd9Sstevel@tonic-gate /* must be less than ZONENAME_MAX */ 33707c478bd9Sstevel@tonic-gate if (len == ZONENAME_MAX && kname[ZONENAME_MAX - 1] != '\0') { 33717c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 33727c478bd9Sstevel@tonic-gate return (EINVAL); 33737c478bd9Sstevel@tonic-gate } 33747c478bd9Sstevel@tonic-gate 33757c478bd9Sstevel@tonic-gate /* 33767c478bd9Sstevel@tonic-gate * Name must start with an alphanumeric and must contain only 33777c478bd9Sstevel@tonic-gate * alphanumerics, '-', '_' and '.'. 33787c478bd9Sstevel@tonic-gate */ 33797c478bd9Sstevel@tonic-gate if (!isalnum(kname[0])) { 33807c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 33817c478bd9Sstevel@tonic-gate return (EINVAL); 33827c478bd9Sstevel@tonic-gate } 33837c478bd9Sstevel@tonic-gate for (i = 1; i < len - 1; i++) { 33847c478bd9Sstevel@tonic-gate if (!isalnum(kname[i]) && kname[i] != '-' && kname[i] != '_' && 33857c478bd9Sstevel@tonic-gate kname[i] != '.') { 33867c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 33877c478bd9Sstevel@tonic-gate return (EINVAL); 33887c478bd9Sstevel@tonic-gate } 33897c478bd9Sstevel@tonic-gate } 33907c478bd9Sstevel@tonic-gate 33917c478bd9Sstevel@tonic-gate zone->zone_name = kname; 33927c478bd9Sstevel@tonic-gate return (0); 33937c478bd9Sstevel@tonic-gate } 33947c478bd9Sstevel@tonic-gate 33955679c89fSjv /* 33965679c89fSjv * Gets the 32-bit hostid of the specified zone as an unsigned int. If 'zonep' 33975679c89fSjv * is NULL or it points to a zone with no hostid emulation, then the machine's 33985679c89fSjv * hostid (i.e., the global zone's hostid) is returned. This function returns 33995679c89fSjv * zero if neither the zone nor the host machine (global zone) have hostids. It 34005679c89fSjv * returns HW_INVALID_HOSTID if the function attempts to return the machine's 34015679c89fSjv * hostid and the machine's hostid is invalid. 34025679c89fSjv */ 34035679c89fSjv uint32_t 34045679c89fSjv zone_get_hostid(zone_t *zonep) 34055679c89fSjv { 34065679c89fSjv unsigned long machine_hostid; 34075679c89fSjv 34085679c89fSjv if (zonep == NULL || zonep->zone_hostid == HW_INVALID_HOSTID) { 34095679c89fSjv if (ddi_strtoul(hw_serial, NULL, 10, &machine_hostid) != 0) 34105679c89fSjv return (HW_INVALID_HOSTID); 34115679c89fSjv return ((uint32_t)machine_hostid); 34125679c89fSjv } 34135679c89fSjv return (zonep->zone_hostid); 34145679c89fSjv } 34155679c89fSjv 34167c478bd9Sstevel@tonic-gate /* 34177c478bd9Sstevel@tonic-gate * Similar to thread_create(), but makes sure the thread is in the appropriate 34187c478bd9Sstevel@tonic-gate * zone's zsched process (curproc->p_zone->zone_zsched) before returning. 34197c478bd9Sstevel@tonic-gate */ 34207c478bd9Sstevel@tonic-gate /*ARGSUSED*/ 34217c478bd9Sstevel@tonic-gate kthread_t * 34227c478bd9Sstevel@tonic-gate zthread_create( 34237c478bd9Sstevel@tonic-gate caddr_t stk, 34247c478bd9Sstevel@tonic-gate size_t stksize, 34257c478bd9Sstevel@tonic-gate void (*proc)(), 34267c478bd9Sstevel@tonic-gate void *arg, 34277c478bd9Sstevel@tonic-gate size_t len, 34287c478bd9Sstevel@tonic-gate pri_t pri) 34297c478bd9Sstevel@tonic-gate { 34307c478bd9Sstevel@tonic-gate kthread_t *t; 34317c478bd9Sstevel@tonic-gate zone_t *zone = curproc->p_zone; 34327c478bd9Sstevel@tonic-gate proc_t *pp = zone->zone_zsched; 34337c478bd9Sstevel@tonic-gate 34347c478bd9Sstevel@tonic-gate zone_hold(zone); /* Reference to be dropped when thread exits */ 34357c478bd9Sstevel@tonic-gate 34367c478bd9Sstevel@tonic-gate /* 34377c478bd9Sstevel@tonic-gate * No-one should be trying to create threads if the zone is shutting 34387c478bd9Sstevel@tonic-gate * down and there aren't any kernel threads around. See comment 34397c478bd9Sstevel@tonic-gate * in zthread_exit(). 34407c478bd9Sstevel@tonic-gate */ 34417c478bd9Sstevel@tonic-gate ASSERT(!(zone->zone_kthreads == NULL && 34427c478bd9Sstevel@tonic-gate zone_status_get(zone) >= ZONE_IS_EMPTY)); 34437c478bd9Sstevel@tonic-gate /* 34447c478bd9Sstevel@tonic-gate * Create a thread, but don't let it run until we've finished setting 34457c478bd9Sstevel@tonic-gate * things up. 34467c478bd9Sstevel@tonic-gate */ 34477c478bd9Sstevel@tonic-gate t = thread_create(stk, stksize, proc, arg, len, pp, TS_STOPPED, pri); 34487c478bd9Sstevel@tonic-gate ASSERT(t->t_forw == NULL); 34497c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 34507c478bd9Sstevel@tonic-gate if (zone->zone_kthreads == NULL) { 34517c478bd9Sstevel@tonic-gate t->t_forw = t->t_back = t; 34527c478bd9Sstevel@tonic-gate } else { 34537c478bd9Sstevel@tonic-gate kthread_t *tx = zone->zone_kthreads; 34547c478bd9Sstevel@tonic-gate 34557c478bd9Sstevel@tonic-gate t->t_forw = tx; 34567c478bd9Sstevel@tonic-gate t->t_back = tx->t_back; 34577c478bd9Sstevel@tonic-gate tx->t_back->t_forw = t; 34587c478bd9Sstevel@tonic-gate tx->t_back = t; 34597c478bd9Sstevel@tonic-gate } 34607c478bd9Sstevel@tonic-gate zone->zone_kthreads = t; 34617c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 34627c478bd9Sstevel@tonic-gate 34637c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 34647c478bd9Sstevel@tonic-gate t->t_proc_flag |= TP_ZTHREAD; 34657c478bd9Sstevel@tonic-gate project_rele(t->t_proj); 34667c478bd9Sstevel@tonic-gate t->t_proj = project_hold(pp->p_task->tk_proj); 34677c478bd9Sstevel@tonic-gate 34687c478bd9Sstevel@tonic-gate /* 34697c478bd9Sstevel@tonic-gate * Setup complete, let it run. 34707c478bd9Sstevel@tonic-gate */ 34717c478bd9Sstevel@tonic-gate thread_lock(t); 34727c478bd9Sstevel@tonic-gate t->t_schedflag |= TS_ALLSTART; 34737c478bd9Sstevel@tonic-gate setrun_locked(t); 34747c478bd9Sstevel@tonic-gate thread_unlock(t); 34757c478bd9Sstevel@tonic-gate 34767c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 34777c478bd9Sstevel@tonic-gate 34787c478bd9Sstevel@tonic-gate return (t); 34797c478bd9Sstevel@tonic-gate } 34807c478bd9Sstevel@tonic-gate 34817c478bd9Sstevel@tonic-gate /* 34827c478bd9Sstevel@tonic-gate * Similar to thread_exit(). Must be called by threads created via 34837c478bd9Sstevel@tonic-gate * zthread_exit(). 34847c478bd9Sstevel@tonic-gate */ 34857c478bd9Sstevel@tonic-gate void 34867c478bd9Sstevel@tonic-gate zthread_exit(void) 34877c478bd9Sstevel@tonic-gate { 34887c478bd9Sstevel@tonic-gate kthread_t *t = curthread; 34897c478bd9Sstevel@tonic-gate proc_t *pp = curproc; 34907c478bd9Sstevel@tonic-gate zone_t *zone = pp->p_zone; 34917c478bd9Sstevel@tonic-gate 34927c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 34937c478bd9Sstevel@tonic-gate 34947c478bd9Sstevel@tonic-gate /* 34957c478bd9Sstevel@tonic-gate * Reparent to p0 34967c478bd9Sstevel@tonic-gate */ 3497b4b07f87Sjosephb kpreempt_disable(); 34987c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 34997c478bd9Sstevel@tonic-gate t->t_proc_flag &= ~TP_ZTHREAD; 35007c478bd9Sstevel@tonic-gate t->t_procp = &p0; 35017c478bd9Sstevel@tonic-gate hat_thread_exit(t); 35027c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 3503b4b07f87Sjosephb kpreempt_enable(); 35047c478bd9Sstevel@tonic-gate 35057c478bd9Sstevel@tonic-gate if (t->t_back == t) { 35067c478bd9Sstevel@tonic-gate ASSERT(t->t_forw == t); 35077c478bd9Sstevel@tonic-gate /* 35087c478bd9Sstevel@tonic-gate * If the zone is empty, once the thread count 35097c478bd9Sstevel@tonic-gate * goes to zero no further kernel threads can be 35107c478bd9Sstevel@tonic-gate * created. This is because if the creator is a process 35117c478bd9Sstevel@tonic-gate * in the zone, then it must have exited before the zone 35127c478bd9Sstevel@tonic-gate * state could be set to ZONE_IS_EMPTY. 35137c478bd9Sstevel@tonic-gate * Otherwise, if the creator is a kernel thread in the 35147c478bd9Sstevel@tonic-gate * zone, the thread count is non-zero. 35157c478bd9Sstevel@tonic-gate * 35167c478bd9Sstevel@tonic-gate * This really means that non-zone kernel threads should 35177c478bd9Sstevel@tonic-gate * not create zone kernel threads. 35187c478bd9Sstevel@tonic-gate */ 35197c478bd9Sstevel@tonic-gate zone->zone_kthreads = NULL; 35207c478bd9Sstevel@tonic-gate if (zone_status_get(zone) == ZONE_IS_EMPTY) { 35217c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_DOWN); 3522c97ad5cdSakolb /* 3523c97ad5cdSakolb * Remove any CPU caps on this zone. 3524c97ad5cdSakolb */ 3525c97ad5cdSakolb cpucaps_zone_remove(zone); 35267c478bd9Sstevel@tonic-gate } 35277c478bd9Sstevel@tonic-gate } else { 35287c478bd9Sstevel@tonic-gate t->t_forw->t_back = t->t_back; 35297c478bd9Sstevel@tonic-gate t->t_back->t_forw = t->t_forw; 35307c478bd9Sstevel@tonic-gate if (zone->zone_kthreads == t) 35317c478bd9Sstevel@tonic-gate zone->zone_kthreads = t->t_forw; 35327c478bd9Sstevel@tonic-gate } 35337c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 35347c478bd9Sstevel@tonic-gate zone_rele(zone); 35357c478bd9Sstevel@tonic-gate thread_exit(); 35367c478bd9Sstevel@tonic-gate /* NOTREACHED */ 35377c478bd9Sstevel@tonic-gate } 35387c478bd9Sstevel@tonic-gate 35397c478bd9Sstevel@tonic-gate static void 35407c478bd9Sstevel@tonic-gate zone_chdir(vnode_t *vp, vnode_t **vpp, proc_t *pp) 35417c478bd9Sstevel@tonic-gate { 35427c478bd9Sstevel@tonic-gate vnode_t *oldvp; 35437c478bd9Sstevel@tonic-gate 35447c478bd9Sstevel@tonic-gate /* we're going to hold a reference here to the directory */ 35457c478bd9Sstevel@tonic-gate VN_HOLD(vp); 35467c478bd9Sstevel@tonic-gate 3547005d3febSMarek Pospisil /* update abs cwd/root path see c2/audit.c */ 3548005d3febSMarek Pospisil if (AU_AUDITING()) 35497c478bd9Sstevel@tonic-gate audit_chdirec(vp, vpp); 35507c478bd9Sstevel@tonic-gate 35517c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 35527c478bd9Sstevel@tonic-gate oldvp = *vpp; 35537c478bd9Sstevel@tonic-gate *vpp = vp; 35547c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 35557c478bd9Sstevel@tonic-gate if (oldvp != NULL) 35567c478bd9Sstevel@tonic-gate VN_RELE(oldvp); 35577c478bd9Sstevel@tonic-gate } 35587c478bd9Sstevel@tonic-gate 35597c478bd9Sstevel@tonic-gate /* 35607c478bd9Sstevel@tonic-gate * Convert an rctl value represented by an nvlist_t into an rctl_val_t. 35617c478bd9Sstevel@tonic-gate */ 35627c478bd9Sstevel@tonic-gate static int 35637c478bd9Sstevel@tonic-gate nvlist2rctlval(nvlist_t *nvl, rctl_val_t *rv) 35647c478bd9Sstevel@tonic-gate { 35657c478bd9Sstevel@tonic-gate nvpair_t *nvp = NULL; 35667c478bd9Sstevel@tonic-gate boolean_t priv_set = B_FALSE; 35677c478bd9Sstevel@tonic-gate boolean_t limit_set = B_FALSE; 35687c478bd9Sstevel@tonic-gate boolean_t action_set = B_FALSE; 35697c478bd9Sstevel@tonic-gate 35707c478bd9Sstevel@tonic-gate while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { 35717c478bd9Sstevel@tonic-gate const char *name; 35727c478bd9Sstevel@tonic-gate uint64_t ui64; 35737c478bd9Sstevel@tonic-gate 35747c478bd9Sstevel@tonic-gate name = nvpair_name(nvp); 35757c478bd9Sstevel@tonic-gate if (nvpair_type(nvp) != DATA_TYPE_UINT64) 35767c478bd9Sstevel@tonic-gate return (EINVAL); 35777c478bd9Sstevel@tonic-gate (void) nvpair_value_uint64(nvp, &ui64); 35787c478bd9Sstevel@tonic-gate if (strcmp(name, "privilege") == 0) { 35797c478bd9Sstevel@tonic-gate /* 35807c478bd9Sstevel@tonic-gate * Currently only privileged values are allowed, but 35817c478bd9Sstevel@tonic-gate * this may change in the future. 35827c478bd9Sstevel@tonic-gate */ 35837c478bd9Sstevel@tonic-gate if (ui64 != RCPRIV_PRIVILEGED) 35847c478bd9Sstevel@tonic-gate return (EINVAL); 35857c478bd9Sstevel@tonic-gate rv->rcv_privilege = ui64; 35867c478bd9Sstevel@tonic-gate priv_set = B_TRUE; 35877c478bd9Sstevel@tonic-gate } else if (strcmp(name, "limit") == 0) { 35887c478bd9Sstevel@tonic-gate rv->rcv_value = ui64; 35897c478bd9Sstevel@tonic-gate limit_set = B_TRUE; 35907c478bd9Sstevel@tonic-gate } else if (strcmp(name, "action") == 0) { 35917c478bd9Sstevel@tonic-gate if (ui64 != RCTL_LOCAL_NOACTION && 35927c478bd9Sstevel@tonic-gate ui64 != RCTL_LOCAL_DENY) 35937c478bd9Sstevel@tonic-gate return (EINVAL); 35947c478bd9Sstevel@tonic-gate rv->rcv_flagaction = ui64; 35957c478bd9Sstevel@tonic-gate action_set = B_TRUE; 35967c478bd9Sstevel@tonic-gate } else { 35977c478bd9Sstevel@tonic-gate return (EINVAL); 35987c478bd9Sstevel@tonic-gate } 35997c478bd9Sstevel@tonic-gate } 36007c478bd9Sstevel@tonic-gate 36017c478bd9Sstevel@tonic-gate if (!(priv_set && limit_set && action_set)) 36027c478bd9Sstevel@tonic-gate return (EINVAL); 36037c478bd9Sstevel@tonic-gate rv->rcv_action_signal = 0; 36047c478bd9Sstevel@tonic-gate rv->rcv_action_recipient = NULL; 36057c478bd9Sstevel@tonic-gate rv->rcv_action_recip_pid = -1; 36067c478bd9Sstevel@tonic-gate rv->rcv_firing_time = 0; 36077c478bd9Sstevel@tonic-gate 36087c478bd9Sstevel@tonic-gate return (0); 36097c478bd9Sstevel@tonic-gate } 36107c478bd9Sstevel@tonic-gate 36113f2f09c1Sdp /* 36123f2f09c1Sdp * Non-global zone version of start_init. 36133f2f09c1Sdp */ 36147c478bd9Sstevel@tonic-gate void 36153f2f09c1Sdp zone_start_init(void) 36167c478bd9Sstevel@tonic-gate { 36177c478bd9Sstevel@tonic-gate proc_t *p = ttoproc(curthread); 36189acbbeafSnn zone_t *z = p->p_zone; 36197c478bd9Sstevel@tonic-gate 36203f2f09c1Sdp ASSERT(!INGLOBALZONE(curproc)); 36217c478bd9Sstevel@tonic-gate 36229acbbeafSnn /* 36239acbbeafSnn * For all purposes (ZONE_ATTR_INITPID and restart_init), 36249acbbeafSnn * storing just the pid of init is sufficient. 36259acbbeafSnn */ 36269acbbeafSnn z->zone_proc_initpid = p->p_pid; 36279acbbeafSnn 36287c478bd9Sstevel@tonic-gate /* 36293f2f09c1Sdp * We maintain zone_boot_err so that we can return the cause of the 36303f2f09c1Sdp * failure back to the caller of the zone_boot syscall. 36317c478bd9Sstevel@tonic-gate */ 36323f2f09c1Sdp p->p_zone->zone_boot_err = start_init_common(); 36337c478bd9Sstevel@tonic-gate 36348f983ab3Sjv /* 36358f983ab3Sjv * We will prevent booting zones from becoming running zones if the 36368f983ab3Sjv * global zone is shutting down. 36378f983ab3Sjv */ 36387c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 36398f983ab3Sjv if (z->zone_boot_err != 0 || zone_status_get(global_zone) >= 36408f983ab3Sjv ZONE_IS_SHUTTING_DOWN) { 36417c478bd9Sstevel@tonic-gate /* 36427c478bd9Sstevel@tonic-gate * Make sure we are still in the booting state-- we could have 36437c478bd9Sstevel@tonic-gate * raced and already be shutting down, or even further along. 36447c478bd9Sstevel@tonic-gate */ 3645c97ad5cdSakolb if (zone_status_get(z) == ZONE_IS_BOOTING) { 36469acbbeafSnn zone_status_set(z, ZONE_IS_SHUTTING_DOWN); 3647c97ad5cdSakolb } 36487c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 36497c478bd9Sstevel@tonic-gate /* It's gone bad, dispose of the process */ 36509acbbeafSnn if (proc_exit(CLD_EXITED, z->zone_boot_err) != 0) { 365197eda132Sraf mutex_enter(&p->p_lock); 365297eda132Sraf ASSERT(p->p_flag & SEXITLWPS); 36537c478bd9Sstevel@tonic-gate lwp_exit(); 36547c478bd9Sstevel@tonic-gate } 36557c478bd9Sstevel@tonic-gate } else { 36569acbbeafSnn if (zone_status_get(z) == ZONE_IS_BOOTING) 36579acbbeafSnn zone_status_set(z, ZONE_IS_RUNNING); 36587c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 36597c478bd9Sstevel@tonic-gate /* cause the process to return to userland. */ 36607c478bd9Sstevel@tonic-gate lwp_rtt(); 36617c478bd9Sstevel@tonic-gate } 36627c478bd9Sstevel@tonic-gate } 36637c478bd9Sstevel@tonic-gate 36647c478bd9Sstevel@tonic-gate struct zsched_arg { 36657c478bd9Sstevel@tonic-gate zone_t *zone; 36667c478bd9Sstevel@tonic-gate nvlist_t *nvlist; 36677c478bd9Sstevel@tonic-gate }; 36687c478bd9Sstevel@tonic-gate 36697c478bd9Sstevel@tonic-gate /* 36707c478bd9Sstevel@tonic-gate * Per-zone "sched" workalike. The similarity to "sched" doesn't have 36717c478bd9Sstevel@tonic-gate * anything to do with scheduling, but rather with the fact that 36727c478bd9Sstevel@tonic-gate * per-zone kernel threads are parented to zsched, just like regular 36737c478bd9Sstevel@tonic-gate * kernel threads are parented to sched (p0). 36747c478bd9Sstevel@tonic-gate * 36757c478bd9Sstevel@tonic-gate * zsched is also responsible for launching init for the zone. 36767c478bd9Sstevel@tonic-gate */ 36777c478bd9Sstevel@tonic-gate static void 36787c478bd9Sstevel@tonic-gate zsched(void *arg) 36797c478bd9Sstevel@tonic-gate { 36807c478bd9Sstevel@tonic-gate struct zsched_arg *za = arg; 36817c478bd9Sstevel@tonic-gate proc_t *pp = curproc; 36827c478bd9Sstevel@tonic-gate proc_t *initp = proc_init; 36837c478bd9Sstevel@tonic-gate zone_t *zone = za->zone; 36847c478bd9Sstevel@tonic-gate cred_t *cr, *oldcred; 36857c478bd9Sstevel@tonic-gate rctl_set_t *set; 36867c478bd9Sstevel@tonic-gate rctl_alloc_gp_t *gp; 36877c478bd9Sstevel@tonic-gate contract_t *ct = NULL; 36887c478bd9Sstevel@tonic-gate task_t *tk, *oldtk; 36897c478bd9Sstevel@tonic-gate rctl_entity_p_t e; 36907c478bd9Sstevel@tonic-gate kproject_t *pj; 36917c478bd9Sstevel@tonic-gate 36927c478bd9Sstevel@tonic-gate nvlist_t *nvl = za->nvlist; 36937c478bd9Sstevel@tonic-gate nvpair_t *nvp = NULL; 36947c478bd9Sstevel@tonic-gate 3695ae115bc7Smrj bcopy("zsched", PTOU(pp)->u_psargs, sizeof ("zsched")); 3696ae115bc7Smrj bcopy("zsched", PTOU(pp)->u_comm, sizeof ("zsched")); 3697ae115bc7Smrj PTOU(pp)->u_argc = 0; 3698ae115bc7Smrj PTOU(pp)->u_argv = NULL; 3699ae115bc7Smrj PTOU(pp)->u_envp = NULL; 37007c478bd9Sstevel@tonic-gate closeall(P_FINFO(pp)); 37017c478bd9Sstevel@tonic-gate 37027c478bd9Sstevel@tonic-gate /* 37037c478bd9Sstevel@tonic-gate * We are this zone's "zsched" process. As the zone isn't generally 37047c478bd9Sstevel@tonic-gate * visible yet we don't need to grab any locks before initializing its 37057c478bd9Sstevel@tonic-gate * zone_proc pointer. 37067c478bd9Sstevel@tonic-gate */ 37077c478bd9Sstevel@tonic-gate zone_hold(zone); /* this hold is released by zone_destroy() */ 37087c478bd9Sstevel@tonic-gate zone->zone_zsched = pp; 37097c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 37107c478bd9Sstevel@tonic-gate pp->p_zone = zone; 37117c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 37127c478bd9Sstevel@tonic-gate 37137c478bd9Sstevel@tonic-gate /* 37147c478bd9Sstevel@tonic-gate * Disassociate process from its 'parent'; parent ourselves to init 37157c478bd9Sstevel@tonic-gate * (pid 1) and change other values as needed. 37167c478bd9Sstevel@tonic-gate */ 37177c478bd9Sstevel@tonic-gate sess_create(); 37187c478bd9Sstevel@tonic-gate 37197c478bd9Sstevel@tonic-gate mutex_enter(&pidlock); 37207c478bd9Sstevel@tonic-gate proc_detach(pp); 37217c478bd9Sstevel@tonic-gate pp->p_ppid = 1; 37227c478bd9Sstevel@tonic-gate pp->p_flag |= SZONETOP; 37237c478bd9Sstevel@tonic-gate pp->p_ancpid = 1; 37247c478bd9Sstevel@tonic-gate pp->p_parent = initp; 37257c478bd9Sstevel@tonic-gate pp->p_psibling = NULL; 37267c478bd9Sstevel@tonic-gate if (initp->p_child) 37277c478bd9Sstevel@tonic-gate initp->p_child->p_psibling = pp; 37287c478bd9Sstevel@tonic-gate pp->p_sibling = initp->p_child; 37297c478bd9Sstevel@tonic-gate initp->p_child = pp; 37307c478bd9Sstevel@tonic-gate 37317c478bd9Sstevel@tonic-gate /* Decrement what newproc() incremented. */ 37327c478bd9Sstevel@tonic-gate upcount_dec(crgetruid(CRED()), GLOBAL_ZONEID); 37337c478bd9Sstevel@tonic-gate /* 37347c478bd9Sstevel@tonic-gate * Our credentials are about to become kcred-like, so we don't care 37357c478bd9Sstevel@tonic-gate * about the caller's ruid. 37367c478bd9Sstevel@tonic-gate */ 37377c478bd9Sstevel@tonic-gate upcount_inc(crgetruid(kcred), zone->zone_id); 37387c478bd9Sstevel@tonic-gate mutex_exit(&pidlock); 37397c478bd9Sstevel@tonic-gate 37407c478bd9Sstevel@tonic-gate /* 3741ff19e029SMenno Lageman * getting out of global zone, so decrement lwp and process counts 37427c478bd9Sstevel@tonic-gate */ 37437c478bd9Sstevel@tonic-gate pj = pp->p_task->tk_proj; 37447c478bd9Sstevel@tonic-gate mutex_enter(&global_zone->zone_nlwps_lock); 37457c478bd9Sstevel@tonic-gate pj->kpj_nlwps -= pp->p_lwpcnt; 37467c478bd9Sstevel@tonic-gate global_zone->zone_nlwps -= pp->p_lwpcnt; 3747ff19e029SMenno Lageman pj->kpj_nprocs--; 3748ff19e029SMenno Lageman global_zone->zone_nprocs--; 37497c478bd9Sstevel@tonic-gate mutex_exit(&global_zone->zone_nlwps_lock); 37507c478bd9Sstevel@tonic-gate 3751c6939658Ssl /* 3752c6939658Ssl * Decrement locked memory counts on old zone and project. 3753c6939658Ssl */ 37540209230bSgjelinek mutex_enter(&global_zone->zone_mem_lock); 3755c6939658Ssl global_zone->zone_locked_mem -= pp->p_locked_mem; 3756c6939658Ssl pj->kpj_data.kpd_locked_mem -= pp->p_locked_mem; 37570209230bSgjelinek mutex_exit(&global_zone->zone_mem_lock); 3758c6939658Ssl 37597c478bd9Sstevel@tonic-gate /* 37607c478bd9Sstevel@tonic-gate * Create and join a new task in project '0' of this zone. 37617c478bd9Sstevel@tonic-gate * 37627c478bd9Sstevel@tonic-gate * We don't need to call holdlwps() since we know we're the only lwp in 37637c478bd9Sstevel@tonic-gate * this process. 37647c478bd9Sstevel@tonic-gate * 37657c478bd9Sstevel@tonic-gate * task_join() returns with p_lock held. 37667c478bd9Sstevel@tonic-gate */ 37677c478bd9Sstevel@tonic-gate tk = task_create(0, zone); 37687c478bd9Sstevel@tonic-gate mutex_enter(&cpu_lock); 37697c478bd9Sstevel@tonic-gate oldtk = task_join(tk, 0); 3770c6939658Ssl 3771c6939658Ssl pj = pp->p_task->tk_proj; 3772c6939658Ssl 37730209230bSgjelinek mutex_enter(&zone->zone_mem_lock); 3774c6939658Ssl zone->zone_locked_mem += pp->p_locked_mem; 3775c6939658Ssl pj->kpj_data.kpd_locked_mem += pp->p_locked_mem; 37760209230bSgjelinek mutex_exit(&zone->zone_mem_lock); 37777c478bd9Sstevel@tonic-gate 37787c478bd9Sstevel@tonic-gate /* 3779ff19e029SMenno Lageman * add lwp and process counts to zsched's zone, and increment 3780ff19e029SMenno Lageman * project's task and process count due to the task created in 3781ff19e029SMenno Lageman * the above task_create. 37827c478bd9Sstevel@tonic-gate */ 37837c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_nlwps_lock); 37847c478bd9Sstevel@tonic-gate pj->kpj_nlwps += pp->p_lwpcnt; 37857c478bd9Sstevel@tonic-gate pj->kpj_ntasks += 1; 37867c478bd9Sstevel@tonic-gate zone->zone_nlwps += pp->p_lwpcnt; 3787ff19e029SMenno Lageman pj->kpj_nprocs++; 3788ff19e029SMenno Lageman zone->zone_nprocs++; 37897c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_nlwps_lock); 37907c478bd9Sstevel@tonic-gate 3791c6939658Ssl mutex_exit(&curproc->p_lock); 3792c6939658Ssl mutex_exit(&cpu_lock); 3793c6939658Ssl task_rele(oldtk); 3794c6939658Ssl 37957c478bd9Sstevel@tonic-gate /* 37967c478bd9Sstevel@tonic-gate * The process was created by a process in the global zone, hence the 37977c478bd9Sstevel@tonic-gate * credentials are wrong. We might as well have kcred-ish credentials. 37987c478bd9Sstevel@tonic-gate */ 37997c478bd9Sstevel@tonic-gate cr = zone->zone_kcred; 38007c478bd9Sstevel@tonic-gate crhold(cr); 38017c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_crlock); 38027c478bd9Sstevel@tonic-gate oldcred = pp->p_cred; 38037c478bd9Sstevel@tonic-gate pp->p_cred = cr; 38047c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_crlock); 38057c478bd9Sstevel@tonic-gate crfree(oldcred); 38067c478bd9Sstevel@tonic-gate 38077c478bd9Sstevel@tonic-gate /* 38087c478bd9Sstevel@tonic-gate * Hold credentials again (for thread) 38097c478bd9Sstevel@tonic-gate */ 38107c478bd9Sstevel@tonic-gate crhold(cr); 38117c478bd9Sstevel@tonic-gate 38127c478bd9Sstevel@tonic-gate /* 38137c478bd9Sstevel@tonic-gate * p_lwpcnt can't change since this is a kernel process. 38147c478bd9Sstevel@tonic-gate */ 38157c478bd9Sstevel@tonic-gate crset(pp, cr); 38167c478bd9Sstevel@tonic-gate 38177c478bd9Sstevel@tonic-gate /* 38187c478bd9Sstevel@tonic-gate * Chroot 38197c478bd9Sstevel@tonic-gate */ 38207c478bd9Sstevel@tonic-gate zone_chdir(zone->zone_rootvp, &PTOU(pp)->u_cdir, pp); 38217c478bd9Sstevel@tonic-gate zone_chdir(zone->zone_rootvp, &PTOU(pp)->u_rdir, pp); 38227c478bd9Sstevel@tonic-gate 38237c478bd9Sstevel@tonic-gate /* 38247c478bd9Sstevel@tonic-gate * Initialize zone's rctl set. 38257c478bd9Sstevel@tonic-gate */ 38267c478bd9Sstevel@tonic-gate set = rctl_set_create(); 38277c478bd9Sstevel@tonic-gate gp = rctl_set_init_prealloc(RCENTITY_ZONE); 38287c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 38297c478bd9Sstevel@tonic-gate e.rcep_p.zone = zone; 38307c478bd9Sstevel@tonic-gate e.rcep_t = RCENTITY_ZONE; 38317c478bd9Sstevel@tonic-gate zone->zone_rctls = rctl_set_init(RCENTITY_ZONE, pp, &e, set, gp); 38327c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 38337c478bd9Sstevel@tonic-gate rctl_prealloc_destroy(gp); 38347c478bd9Sstevel@tonic-gate 38357c478bd9Sstevel@tonic-gate /* 38367c478bd9Sstevel@tonic-gate * Apply the rctls passed in to zone_create(). This is basically a list 38377c478bd9Sstevel@tonic-gate * assignment: all of the old values are removed and the new ones 38387c478bd9Sstevel@tonic-gate * inserted. That is, if an empty list is passed in, all values are 38397c478bd9Sstevel@tonic-gate * removed. 38407c478bd9Sstevel@tonic-gate */ 38417c478bd9Sstevel@tonic-gate while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { 38427c478bd9Sstevel@tonic-gate rctl_dict_entry_t *rde; 38437c478bd9Sstevel@tonic-gate rctl_hndl_t hndl; 38447c478bd9Sstevel@tonic-gate char *name; 38457c478bd9Sstevel@tonic-gate nvlist_t **nvlarray; 38467c478bd9Sstevel@tonic-gate uint_t i, nelem; 38477c478bd9Sstevel@tonic-gate int error; /* For ASSERT()s */ 38487c478bd9Sstevel@tonic-gate 38497c478bd9Sstevel@tonic-gate name = nvpair_name(nvp); 38507c478bd9Sstevel@tonic-gate hndl = rctl_hndl_lookup(name); 38517c478bd9Sstevel@tonic-gate ASSERT(hndl != -1); 38527c478bd9Sstevel@tonic-gate rde = rctl_dict_lookup_hndl(hndl); 38537c478bd9Sstevel@tonic-gate ASSERT(rde != NULL); 38547c478bd9Sstevel@tonic-gate 38557c478bd9Sstevel@tonic-gate for (; /* ever */; ) { 38567c478bd9Sstevel@tonic-gate rctl_val_t oval; 38577c478bd9Sstevel@tonic-gate 38587c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 38597c478bd9Sstevel@tonic-gate error = rctl_local_get(hndl, NULL, &oval, pp); 38607c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 38617c478bd9Sstevel@tonic-gate ASSERT(error == 0); /* Can't fail for RCTL_FIRST */ 38627c478bd9Sstevel@tonic-gate ASSERT(oval.rcv_privilege != RCPRIV_BASIC); 38637c478bd9Sstevel@tonic-gate if (oval.rcv_privilege == RCPRIV_SYSTEM) 38647c478bd9Sstevel@tonic-gate break; 38657c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 38667c478bd9Sstevel@tonic-gate error = rctl_local_delete(hndl, &oval, pp); 38677c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 38687c478bd9Sstevel@tonic-gate ASSERT(error == 0); 38697c478bd9Sstevel@tonic-gate } 38707c478bd9Sstevel@tonic-gate error = nvpair_value_nvlist_array(nvp, &nvlarray, &nelem); 38717c478bd9Sstevel@tonic-gate ASSERT(error == 0); 38727c478bd9Sstevel@tonic-gate for (i = 0; i < nelem; i++) { 38737c478bd9Sstevel@tonic-gate rctl_val_t *nvalp; 38747c478bd9Sstevel@tonic-gate 38757c478bd9Sstevel@tonic-gate nvalp = kmem_cache_alloc(rctl_val_cache, KM_SLEEP); 38767c478bd9Sstevel@tonic-gate error = nvlist2rctlval(nvlarray[i], nvalp); 38777c478bd9Sstevel@tonic-gate ASSERT(error == 0); 38787c478bd9Sstevel@tonic-gate /* 38797c478bd9Sstevel@tonic-gate * rctl_local_insert can fail if the value being 38807c478bd9Sstevel@tonic-gate * inserted is a duplicate; this is OK. 38817c478bd9Sstevel@tonic-gate */ 38827c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 38837c478bd9Sstevel@tonic-gate if (rctl_local_insert(hndl, nvalp, pp) != 0) 38847c478bd9Sstevel@tonic-gate kmem_cache_free(rctl_val_cache, nvalp); 38857c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 38867c478bd9Sstevel@tonic-gate } 38877c478bd9Sstevel@tonic-gate } 38887c478bd9Sstevel@tonic-gate /* 38897c478bd9Sstevel@tonic-gate * Tell the world that we're done setting up. 38907c478bd9Sstevel@tonic-gate * 3891bd41d0a8Snordmark * At this point we want to set the zone status to ZONE_IS_INITIALIZED 38927c478bd9Sstevel@tonic-gate * and atomically set the zone's processor set visibility. Once 38937c478bd9Sstevel@tonic-gate * we drop pool_lock() this zone will automatically get updated 38947c478bd9Sstevel@tonic-gate * to reflect any future changes to the pools configuration. 3895bd41d0a8Snordmark * 3896bd41d0a8Snordmark * Note that after we drop the locks below (zonehash_lock in 3897bd41d0a8Snordmark * particular) other operations such as a zone_getattr call can 3898bd41d0a8Snordmark * now proceed and observe the zone. That is the reason for doing a 3899bd41d0a8Snordmark * state transition to the INITIALIZED state. 39007c478bd9Sstevel@tonic-gate */ 39017c478bd9Sstevel@tonic-gate pool_lock(); 39027c478bd9Sstevel@tonic-gate mutex_enter(&cpu_lock); 39037c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 39047c478bd9Sstevel@tonic-gate zone_uniqid(zone); 39057c478bd9Sstevel@tonic-gate zone_zsd_configure(zone); 39067c478bd9Sstevel@tonic-gate if (pool_state == POOL_ENABLED) 39077c478bd9Sstevel@tonic-gate zone_pset_set(zone, pool_default->pool_pset->pset_id); 39087c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 39097c478bd9Sstevel@tonic-gate ASSERT(zone_status_get(zone) == ZONE_IS_UNINITIALIZED); 3910bd41d0a8Snordmark zone_status_set(zone, ZONE_IS_INITIALIZED); 39117c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 39127c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 39137c478bd9Sstevel@tonic-gate mutex_exit(&cpu_lock); 39147c478bd9Sstevel@tonic-gate pool_unlock(); 39157c478bd9Sstevel@tonic-gate 3916bd41d0a8Snordmark /* Now call the create callback for this key */ 3917bd41d0a8Snordmark zsd_apply_all_keys(zsd_apply_create, zone); 3918bd41d0a8Snordmark 3919bd41d0a8Snordmark /* The callbacks are complete. Mark ZONE_IS_READY */ 3920bd41d0a8Snordmark mutex_enter(&zone_status_lock); 3921bd41d0a8Snordmark ASSERT(zone_status_get(zone) == ZONE_IS_INITIALIZED); 3922bd41d0a8Snordmark zone_status_set(zone, ZONE_IS_READY); 3923bd41d0a8Snordmark mutex_exit(&zone_status_lock); 3924bd41d0a8Snordmark 39257c478bd9Sstevel@tonic-gate /* 39267c478bd9Sstevel@tonic-gate * Once we see the zone transition to the ZONE_IS_BOOTING state, 39277c478bd9Sstevel@tonic-gate * we launch init, and set the state to running. 39287c478bd9Sstevel@tonic-gate */ 39297c478bd9Sstevel@tonic-gate zone_status_wait_cpr(zone, ZONE_IS_BOOTING, "zsched"); 39307c478bd9Sstevel@tonic-gate 39317c478bd9Sstevel@tonic-gate if (zone_status_get(zone) == ZONE_IS_BOOTING) { 39327c478bd9Sstevel@tonic-gate id_t cid; 39337c478bd9Sstevel@tonic-gate 39347c478bd9Sstevel@tonic-gate /* 39357c478bd9Sstevel@tonic-gate * Ok, this is a little complicated. We need to grab the 39367c478bd9Sstevel@tonic-gate * zone's pool's scheduling class ID; note that by now, we 39377c478bd9Sstevel@tonic-gate * are already bound to a pool if we need to be (zoneadmd 39387c478bd9Sstevel@tonic-gate * will have done that to us while we're in the READY 39397c478bd9Sstevel@tonic-gate * state). *But* the scheduling class for the zone's 'init' 39407c478bd9Sstevel@tonic-gate * must be explicitly passed to newproc, which doesn't 39417c478bd9Sstevel@tonic-gate * respect pool bindings. 39427c478bd9Sstevel@tonic-gate * 39437c478bd9Sstevel@tonic-gate * We hold the pool_lock across the call to newproc() to 39447c478bd9Sstevel@tonic-gate * close the obvious race: the pool's scheduling class 39457c478bd9Sstevel@tonic-gate * could change before we manage to create the LWP with 39467c478bd9Sstevel@tonic-gate * classid 'cid'. 39477c478bd9Sstevel@tonic-gate */ 39487c478bd9Sstevel@tonic-gate pool_lock(); 39490209230bSgjelinek if (zone->zone_defaultcid > 0) 39500209230bSgjelinek cid = zone->zone_defaultcid; 39510209230bSgjelinek else 39520209230bSgjelinek cid = pool_get_class(zone->zone_pool); 39537c478bd9Sstevel@tonic-gate if (cid == -1) 39547c478bd9Sstevel@tonic-gate cid = defaultcid; 39557c478bd9Sstevel@tonic-gate 39567c478bd9Sstevel@tonic-gate /* 39577c478bd9Sstevel@tonic-gate * If this fails, zone_boot will ultimately fail. The 39587c478bd9Sstevel@tonic-gate * state of the zone will be set to SHUTTING_DOWN-- userland 39597c478bd9Sstevel@tonic-gate * will have to tear down the zone, and fail, or try again. 39607c478bd9Sstevel@tonic-gate */ 39613f2f09c1Sdp if ((zone->zone_boot_err = newproc(zone_start_init, NULL, cid, 396235a5a358SJonathan Adams minclsyspri - 1, &ct, 0)) != 0) { 39637c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 39647c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_SHUTTING_DOWN); 39657c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 396686ad481cSJerry Jelinek } else { 396786ad481cSJerry Jelinek zone->zone_boot_time = gethrestime_sec(); 39687c478bd9Sstevel@tonic-gate } 396986ad481cSJerry Jelinek 39707c478bd9Sstevel@tonic-gate pool_unlock(); 39717c478bd9Sstevel@tonic-gate } 39727c478bd9Sstevel@tonic-gate 39737c478bd9Sstevel@tonic-gate /* 39747c478bd9Sstevel@tonic-gate * Wait for zone_destroy() to be called. This is what we spend 39757c478bd9Sstevel@tonic-gate * most of our life doing. 39767c478bd9Sstevel@tonic-gate */ 39777c478bd9Sstevel@tonic-gate zone_status_wait_cpr(zone, ZONE_IS_DYING, "zsched"); 39787c478bd9Sstevel@tonic-gate 39797c478bd9Sstevel@tonic-gate if (ct) 39807c478bd9Sstevel@tonic-gate /* 39817c478bd9Sstevel@tonic-gate * At this point the process contract should be empty. 39827c478bd9Sstevel@tonic-gate * (Though if it isn't, it's not the end of the world.) 39837c478bd9Sstevel@tonic-gate */ 39847c478bd9Sstevel@tonic-gate VERIFY(contract_abandon(ct, curproc, B_TRUE) == 0); 39857c478bd9Sstevel@tonic-gate 39867c478bd9Sstevel@tonic-gate /* 39877c478bd9Sstevel@tonic-gate * Allow kcred to be freed when all referring processes 39887c478bd9Sstevel@tonic-gate * (including this one) go away. We can't just do this in 39897c478bd9Sstevel@tonic-gate * zone_free because we need to wait for the zone_cred_ref to 39907c478bd9Sstevel@tonic-gate * drop to 0 before calling zone_free, and the existence of 39917c478bd9Sstevel@tonic-gate * zone_kcred will prevent that. Thus, we call crfree here to 39927c478bd9Sstevel@tonic-gate * balance the crdup in zone_create. The crhold calls earlier 39937c478bd9Sstevel@tonic-gate * in zsched will be dropped when the thread and process exit. 39947c478bd9Sstevel@tonic-gate */ 39957c478bd9Sstevel@tonic-gate crfree(zone->zone_kcred); 39967c478bd9Sstevel@tonic-gate zone->zone_kcred = NULL; 39977c478bd9Sstevel@tonic-gate 39987c478bd9Sstevel@tonic-gate exit(CLD_EXITED, 0); 39997c478bd9Sstevel@tonic-gate } 40007c478bd9Sstevel@tonic-gate 40017c478bd9Sstevel@tonic-gate /* 40027c478bd9Sstevel@tonic-gate * Helper function to determine if there are any submounts of the 40037c478bd9Sstevel@tonic-gate * provided path. Used to make sure the zone doesn't "inherit" any 40047c478bd9Sstevel@tonic-gate * mounts from before it is created. 40057c478bd9Sstevel@tonic-gate */ 40067c478bd9Sstevel@tonic-gate static uint_t 40077c478bd9Sstevel@tonic-gate zone_mount_count(const char *rootpath) 40087c478bd9Sstevel@tonic-gate { 40097c478bd9Sstevel@tonic-gate vfs_t *vfsp; 40107c478bd9Sstevel@tonic-gate uint_t count = 0; 40117c478bd9Sstevel@tonic-gate size_t rootpathlen = strlen(rootpath); 40127c478bd9Sstevel@tonic-gate 40137c478bd9Sstevel@tonic-gate /* 40147c478bd9Sstevel@tonic-gate * Holding zonehash_lock prevents race conditions with 40157c478bd9Sstevel@tonic-gate * vfs_list_add()/vfs_list_remove() since we serialize with 40167c478bd9Sstevel@tonic-gate * zone_find_by_path(). 40177c478bd9Sstevel@tonic-gate */ 40187c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 40197c478bd9Sstevel@tonic-gate /* 40207c478bd9Sstevel@tonic-gate * The rootpath must end with a '/' 40217c478bd9Sstevel@tonic-gate */ 40227c478bd9Sstevel@tonic-gate ASSERT(rootpath[rootpathlen - 1] == '/'); 40237c478bd9Sstevel@tonic-gate 40247c478bd9Sstevel@tonic-gate /* 40257c478bd9Sstevel@tonic-gate * This intentionally does not count the rootpath itself if that 40267c478bd9Sstevel@tonic-gate * happens to be a mount point. 40277c478bd9Sstevel@tonic-gate */ 40287c478bd9Sstevel@tonic-gate vfs_list_read_lock(); 40297c478bd9Sstevel@tonic-gate vfsp = rootvfs; 40307c478bd9Sstevel@tonic-gate do { 40317c478bd9Sstevel@tonic-gate if (strncmp(rootpath, refstr_value(vfsp->vfs_mntpt), 40327c478bd9Sstevel@tonic-gate rootpathlen) == 0) 40337c478bd9Sstevel@tonic-gate count++; 40347c478bd9Sstevel@tonic-gate vfsp = vfsp->vfs_next; 40357c478bd9Sstevel@tonic-gate } while (vfsp != rootvfs); 40367c478bd9Sstevel@tonic-gate vfs_list_unlock(); 40377c478bd9Sstevel@tonic-gate return (count); 40387c478bd9Sstevel@tonic-gate } 40397c478bd9Sstevel@tonic-gate 40407c478bd9Sstevel@tonic-gate /* 40417c478bd9Sstevel@tonic-gate * Helper function to make sure that a zone created on 'rootpath' 40427c478bd9Sstevel@tonic-gate * wouldn't end up containing other zones' rootpaths. 40437c478bd9Sstevel@tonic-gate */ 40447c478bd9Sstevel@tonic-gate static boolean_t 40457c478bd9Sstevel@tonic-gate zone_is_nested(const char *rootpath) 40467c478bd9Sstevel@tonic-gate { 40477c478bd9Sstevel@tonic-gate zone_t *zone; 40487c478bd9Sstevel@tonic-gate size_t rootpathlen = strlen(rootpath); 40497c478bd9Sstevel@tonic-gate size_t len; 40507c478bd9Sstevel@tonic-gate 40517c478bd9Sstevel@tonic-gate ASSERT(MUTEX_HELD(&zonehash_lock)); 40527c478bd9Sstevel@tonic-gate 40532bb88978SDhanaraj M /* 40542bb88978SDhanaraj M * zone_set_root() appended '/' and '\0' at the end of rootpath 40552bb88978SDhanaraj M */ 40562bb88978SDhanaraj M if ((rootpathlen <= 3) && (rootpath[0] == '/') && 40572bb88978SDhanaraj M (rootpath[1] == '/') && (rootpath[2] == '\0')) 40582bb88978SDhanaraj M return (B_TRUE); 40592bb88978SDhanaraj M 40607c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 40617c478bd9Sstevel@tonic-gate zone = list_next(&zone_active, zone)) { 40627c478bd9Sstevel@tonic-gate if (zone == global_zone) 40637c478bd9Sstevel@tonic-gate continue; 40647c478bd9Sstevel@tonic-gate len = strlen(zone->zone_rootpath); 40657c478bd9Sstevel@tonic-gate if (strncmp(rootpath, zone->zone_rootpath, 40667c478bd9Sstevel@tonic-gate MIN(rootpathlen, len)) == 0) 40677c478bd9Sstevel@tonic-gate return (B_TRUE); 40687c478bd9Sstevel@tonic-gate } 40697c478bd9Sstevel@tonic-gate return (B_FALSE); 40707c478bd9Sstevel@tonic-gate } 40717c478bd9Sstevel@tonic-gate 40727c478bd9Sstevel@tonic-gate static int 4073821c4a97Sdp zone_set_privset(zone_t *zone, const priv_set_t *zone_privs, 4074821c4a97Sdp size_t zone_privssz) 40757c478bd9Sstevel@tonic-gate { 40761a88ce5cSDan Price priv_set_t *privs; 40777c478bd9Sstevel@tonic-gate 4078821c4a97Sdp if (zone_privssz < sizeof (priv_set_t)) 40791a88ce5cSDan Price return (ENOMEM); 40801a88ce5cSDan Price 40811a88ce5cSDan Price privs = kmem_alloc(sizeof (priv_set_t), KM_SLEEP); 4082821c4a97Sdp 40837c478bd9Sstevel@tonic-gate if (copyin(zone_privs, privs, sizeof (priv_set_t))) { 40847c478bd9Sstevel@tonic-gate kmem_free(privs, sizeof (priv_set_t)); 40857c478bd9Sstevel@tonic-gate return (EFAULT); 40867c478bd9Sstevel@tonic-gate } 40877c478bd9Sstevel@tonic-gate 40887c478bd9Sstevel@tonic-gate zone->zone_privset = privs; 40897c478bd9Sstevel@tonic-gate return (0); 40907c478bd9Sstevel@tonic-gate } 40917c478bd9Sstevel@tonic-gate 40927c478bd9Sstevel@tonic-gate /* 40937c478bd9Sstevel@tonic-gate * We make creative use of nvlists to pass in rctls from userland. The list is 40947c478bd9Sstevel@tonic-gate * a list of the following structures: 40957c478bd9Sstevel@tonic-gate * 40967c478bd9Sstevel@tonic-gate * (name = rctl_name, value = nvpair_list_array) 40977c478bd9Sstevel@tonic-gate * 40987c478bd9Sstevel@tonic-gate * Where each element of the nvpair_list_array is of the form: 40997c478bd9Sstevel@tonic-gate * 41007c478bd9Sstevel@tonic-gate * [(name = "privilege", value = RCPRIV_PRIVILEGED), 41017c478bd9Sstevel@tonic-gate * (name = "limit", value = uint64_t), 41027c478bd9Sstevel@tonic-gate * (name = "action", value = (RCTL_LOCAL_NOACTION || RCTL_LOCAL_DENY))] 41037c478bd9Sstevel@tonic-gate */ 41047c478bd9Sstevel@tonic-gate static int 41057c478bd9Sstevel@tonic-gate parse_rctls(caddr_t ubuf, size_t buflen, nvlist_t **nvlp) 41067c478bd9Sstevel@tonic-gate { 41077c478bd9Sstevel@tonic-gate nvpair_t *nvp = NULL; 41087c478bd9Sstevel@tonic-gate nvlist_t *nvl = NULL; 41097c478bd9Sstevel@tonic-gate char *kbuf; 41107c478bd9Sstevel@tonic-gate int error; 41117c478bd9Sstevel@tonic-gate rctl_val_t rv; 41127c478bd9Sstevel@tonic-gate 41137c478bd9Sstevel@tonic-gate *nvlp = NULL; 41147c478bd9Sstevel@tonic-gate 41157c478bd9Sstevel@tonic-gate if (buflen == 0) 41167c478bd9Sstevel@tonic-gate return (0); 41177c478bd9Sstevel@tonic-gate 41187c478bd9Sstevel@tonic-gate if ((kbuf = kmem_alloc(buflen, KM_NOSLEEP)) == NULL) 41197c478bd9Sstevel@tonic-gate return (ENOMEM); 41207c478bd9Sstevel@tonic-gate if (copyin(ubuf, kbuf, buflen)) { 41217c478bd9Sstevel@tonic-gate error = EFAULT; 41227c478bd9Sstevel@tonic-gate goto out; 41237c478bd9Sstevel@tonic-gate } 41247c478bd9Sstevel@tonic-gate if (nvlist_unpack(kbuf, buflen, &nvl, KM_SLEEP) != 0) { 41257c478bd9Sstevel@tonic-gate /* 41267c478bd9Sstevel@tonic-gate * nvl may have been allocated/free'd, but the value set to 41277c478bd9Sstevel@tonic-gate * non-NULL, so we reset it here. 41287c478bd9Sstevel@tonic-gate */ 41297c478bd9Sstevel@tonic-gate nvl = NULL; 41307c478bd9Sstevel@tonic-gate error = EINVAL; 41317c478bd9Sstevel@tonic-gate goto out; 41327c478bd9Sstevel@tonic-gate } 41337c478bd9Sstevel@tonic-gate while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { 41347c478bd9Sstevel@tonic-gate rctl_dict_entry_t *rde; 41357c478bd9Sstevel@tonic-gate rctl_hndl_t hndl; 41367c478bd9Sstevel@tonic-gate nvlist_t **nvlarray; 41377c478bd9Sstevel@tonic-gate uint_t i, nelem; 41387c478bd9Sstevel@tonic-gate char *name; 41397c478bd9Sstevel@tonic-gate 41407c478bd9Sstevel@tonic-gate error = EINVAL; 41417c478bd9Sstevel@tonic-gate name = nvpair_name(nvp); 41427c478bd9Sstevel@tonic-gate if (strncmp(nvpair_name(nvp), "zone.", sizeof ("zone.") - 1) 41437c478bd9Sstevel@tonic-gate != 0 || nvpair_type(nvp) != DATA_TYPE_NVLIST_ARRAY) { 41447c478bd9Sstevel@tonic-gate goto out; 41457c478bd9Sstevel@tonic-gate } 41467c478bd9Sstevel@tonic-gate if ((hndl = rctl_hndl_lookup(name)) == -1) { 41477c478bd9Sstevel@tonic-gate goto out; 41487c478bd9Sstevel@tonic-gate } 41497c478bd9Sstevel@tonic-gate rde = rctl_dict_lookup_hndl(hndl); 41507c478bd9Sstevel@tonic-gate error = nvpair_value_nvlist_array(nvp, &nvlarray, &nelem); 41517c478bd9Sstevel@tonic-gate ASSERT(error == 0); 41527c478bd9Sstevel@tonic-gate for (i = 0; i < nelem; i++) { 41537c478bd9Sstevel@tonic-gate if (error = nvlist2rctlval(nvlarray[i], &rv)) 41547c478bd9Sstevel@tonic-gate goto out; 41557c478bd9Sstevel@tonic-gate } 41567c478bd9Sstevel@tonic-gate if (rctl_invalid_value(rde, &rv)) { 41577c478bd9Sstevel@tonic-gate error = EINVAL; 41587c478bd9Sstevel@tonic-gate goto out; 41597c478bd9Sstevel@tonic-gate } 41607c478bd9Sstevel@tonic-gate } 41617c478bd9Sstevel@tonic-gate error = 0; 41627c478bd9Sstevel@tonic-gate *nvlp = nvl; 41637c478bd9Sstevel@tonic-gate out: 41647c478bd9Sstevel@tonic-gate kmem_free(kbuf, buflen); 41657c478bd9Sstevel@tonic-gate if (error && nvl != NULL) 41667c478bd9Sstevel@tonic-gate nvlist_free(nvl); 41677c478bd9Sstevel@tonic-gate return (error); 41687c478bd9Sstevel@tonic-gate } 41697c478bd9Sstevel@tonic-gate 41707c478bd9Sstevel@tonic-gate int 41717c478bd9Sstevel@tonic-gate zone_create_error(int er_error, int er_ext, int *er_out) { 41727c478bd9Sstevel@tonic-gate if (er_out != NULL) { 41737c478bd9Sstevel@tonic-gate if (copyout(&er_ext, er_out, sizeof (int))) { 41747c478bd9Sstevel@tonic-gate return (set_errno(EFAULT)); 41757c478bd9Sstevel@tonic-gate } 41767c478bd9Sstevel@tonic-gate } 41777c478bd9Sstevel@tonic-gate return (set_errno(er_error)); 41787c478bd9Sstevel@tonic-gate } 41797c478bd9Sstevel@tonic-gate 418045916cd2Sjpk static int 418145916cd2Sjpk zone_set_label(zone_t *zone, const bslabel_t *lab, uint32_t doi) 418245916cd2Sjpk { 418345916cd2Sjpk ts_label_t *tsl; 418445916cd2Sjpk bslabel_t blab; 418545916cd2Sjpk 418645916cd2Sjpk /* Get label from user */ 418745916cd2Sjpk if (copyin(lab, &blab, sizeof (blab)) != 0) 418845916cd2Sjpk return (EFAULT); 418945916cd2Sjpk tsl = labelalloc(&blab, doi, KM_NOSLEEP); 419045916cd2Sjpk if (tsl == NULL) 419145916cd2Sjpk return (ENOMEM); 419245916cd2Sjpk 419345916cd2Sjpk zone->zone_slabel = tsl; 419445916cd2Sjpk return (0); 419545916cd2Sjpk } 419645916cd2Sjpk 4197fa9e4066Sahrens /* 4198fa9e4066Sahrens * Parses a comma-separated list of ZFS datasets into a per-zone dictionary. 4199fa9e4066Sahrens */ 4200fa9e4066Sahrens static int 4201fa9e4066Sahrens parse_zfs(zone_t *zone, caddr_t ubuf, size_t buflen) 4202fa9e4066Sahrens { 4203fa9e4066Sahrens char *kbuf; 4204fa9e4066Sahrens char *dataset, *next; 4205fa9e4066Sahrens zone_dataset_t *zd; 4206fa9e4066Sahrens size_t len; 4207fa9e4066Sahrens 4208fa9e4066Sahrens if (ubuf == NULL || buflen == 0) 4209fa9e4066Sahrens return (0); 4210fa9e4066Sahrens 4211fa9e4066Sahrens if ((kbuf = kmem_alloc(buflen, KM_NOSLEEP)) == NULL) 4212fa9e4066Sahrens return (ENOMEM); 4213fa9e4066Sahrens 4214fa9e4066Sahrens if (copyin(ubuf, kbuf, buflen) != 0) { 4215fa9e4066Sahrens kmem_free(kbuf, buflen); 4216fa9e4066Sahrens return (EFAULT); 4217fa9e4066Sahrens } 4218fa9e4066Sahrens 4219fa9e4066Sahrens dataset = next = kbuf; 4220fa9e4066Sahrens for (;;) { 4221fa9e4066Sahrens zd = kmem_alloc(sizeof (zone_dataset_t), KM_SLEEP); 4222fa9e4066Sahrens 4223fa9e4066Sahrens next = strchr(dataset, ','); 4224fa9e4066Sahrens 4225fa9e4066Sahrens if (next == NULL) 4226fa9e4066Sahrens len = strlen(dataset); 4227fa9e4066Sahrens else 4228fa9e4066Sahrens len = next - dataset; 4229fa9e4066Sahrens 4230fa9e4066Sahrens zd->zd_dataset = kmem_alloc(len + 1, KM_SLEEP); 4231fa9e4066Sahrens bcopy(dataset, zd->zd_dataset, len); 4232fa9e4066Sahrens zd->zd_dataset[len] = '\0'; 4233fa9e4066Sahrens 4234fa9e4066Sahrens list_insert_head(&zone->zone_datasets, zd); 4235fa9e4066Sahrens 4236fa9e4066Sahrens if (next == NULL) 4237fa9e4066Sahrens break; 4238fa9e4066Sahrens 4239fa9e4066Sahrens dataset = next + 1; 4240fa9e4066Sahrens } 4241fa9e4066Sahrens 4242fa9e4066Sahrens kmem_free(kbuf, buflen); 4243fa9e4066Sahrens return (0); 4244fa9e4066Sahrens } 4245fa9e4066Sahrens 42467c478bd9Sstevel@tonic-gate /* 42477c478bd9Sstevel@tonic-gate * System call to create/initialize a new zone named 'zone_name', rooted 42487c478bd9Sstevel@tonic-gate * at 'zone_root', with a zone-wide privilege limit set of 'zone_privs', 424945916cd2Sjpk * and initialized with the zone-wide rctls described in 'rctlbuf', and 425045916cd2Sjpk * with labeling set by 'match', 'doi', and 'label'. 42517c478bd9Sstevel@tonic-gate * 42527c478bd9Sstevel@tonic-gate * If extended error is non-null, we may use it to return more detailed 42537c478bd9Sstevel@tonic-gate * error information. 42547c478bd9Sstevel@tonic-gate */ 42557c478bd9Sstevel@tonic-gate static zoneid_t 42567c478bd9Sstevel@tonic-gate zone_create(const char *zone_name, const char *zone_root, 4257821c4a97Sdp const priv_set_t *zone_privs, size_t zone_privssz, 4258821c4a97Sdp caddr_t rctlbuf, size_t rctlbufsz, 425945916cd2Sjpk caddr_t zfsbuf, size_t zfsbufsz, int *extended_error, 4260f4b3ec61Sdh int match, uint32_t doi, const bslabel_t *label, 4261f4b3ec61Sdh int flags) 42627c478bd9Sstevel@tonic-gate { 42637c478bd9Sstevel@tonic-gate struct zsched_arg zarg; 42647c478bd9Sstevel@tonic-gate nvlist_t *rctls = NULL; 42657c478bd9Sstevel@tonic-gate proc_t *pp = curproc; 42667c478bd9Sstevel@tonic-gate zone_t *zone, *ztmp; 42677c478bd9Sstevel@tonic-gate zoneid_t zoneid; 42687c478bd9Sstevel@tonic-gate int error; 42697c478bd9Sstevel@tonic-gate int error2 = 0; 42707c478bd9Sstevel@tonic-gate char *str; 42717c478bd9Sstevel@tonic-gate cred_t *zkcr; 427248451833Scarlsonj boolean_t insert_label_hash; 42737c478bd9Sstevel@tonic-gate 42747c478bd9Sstevel@tonic-gate if (secpolicy_zone_config(CRED()) != 0) 42757c478bd9Sstevel@tonic-gate return (set_errno(EPERM)); 42767c478bd9Sstevel@tonic-gate 42777c478bd9Sstevel@tonic-gate /* can't boot zone from within chroot environment */ 42787c478bd9Sstevel@tonic-gate if (PTOU(pp)->u_rdir != NULL && PTOU(pp)->u_rdir != rootdir) 42797c478bd9Sstevel@tonic-gate return (zone_create_error(ENOTSUP, ZE_CHROOTED, 4280821c4a97Sdp extended_error)); 42817c478bd9Sstevel@tonic-gate 42827c478bd9Sstevel@tonic-gate zone = kmem_zalloc(sizeof (zone_t), KM_SLEEP); 42837c478bd9Sstevel@tonic-gate zoneid = zone->zone_id = id_alloc(zoneid_space); 42847c478bd9Sstevel@tonic-gate zone->zone_status = ZONE_IS_UNINITIALIZED; 42857c478bd9Sstevel@tonic-gate zone->zone_pool = pool_default; 42867c478bd9Sstevel@tonic-gate zone->zone_pool_mod = gethrtime(); 42877c478bd9Sstevel@tonic-gate zone->zone_psetid = ZONE_PS_INVAL; 42887c478bd9Sstevel@tonic-gate zone->zone_ncpus = 0; 42897c478bd9Sstevel@tonic-gate zone->zone_ncpus_online = 0; 42909acbbeafSnn zone->zone_restart_init = B_TRUE; 42919acbbeafSnn zone->zone_brand = &native_brand; 42929acbbeafSnn zone->zone_initname = NULL; 42937c478bd9Sstevel@tonic-gate mutex_init(&zone->zone_lock, NULL, MUTEX_DEFAULT, NULL); 42947c478bd9Sstevel@tonic-gate mutex_init(&zone->zone_nlwps_lock, NULL, MUTEX_DEFAULT, NULL); 42950209230bSgjelinek mutex_init(&zone->zone_mem_lock, NULL, MUTEX_DEFAULT, NULL); 42967c478bd9Sstevel@tonic-gate cv_init(&zone->zone_cv, NULL, CV_DEFAULT, NULL); 4297a19609f8Sjv list_create(&zone->zone_ref_list, sizeof (zone_ref_t), 4298a19609f8Sjv offsetof(zone_ref_t, zref_linkage)); 42997c478bd9Sstevel@tonic-gate list_create(&zone->zone_zsd, sizeof (struct zsd_entry), 43007c478bd9Sstevel@tonic-gate offsetof(struct zsd_entry, zsd_linkage)); 4301fa9e4066Sahrens list_create(&zone->zone_datasets, sizeof (zone_dataset_t), 4302fa9e4066Sahrens offsetof(zone_dataset_t, zd_linkage)); 43032b24ab6bSSebastien Roy list_create(&zone->zone_dl_list, sizeof (zone_dl_t), 43042b24ab6bSSebastien Roy offsetof(zone_dl_t, zdl_linkage)); 430545916cd2Sjpk rw_init(&zone->zone_mlps.mlpl_rwlock, NULL, RW_DEFAULT, NULL); 4306835ee219SRobert Harris rw_init(&zone->zone_mntfs_db_lock, NULL, RW_DEFAULT, NULL); 43077c478bd9Sstevel@tonic-gate 4308f4b3ec61Sdh if (flags & ZCF_NET_EXCL) { 4309f4b3ec61Sdh zone->zone_flags |= ZF_NET_EXCL; 4310f4b3ec61Sdh } 4311f4b3ec61Sdh 43127c478bd9Sstevel@tonic-gate if ((error = zone_set_name(zone, zone_name)) != 0) { 43137c478bd9Sstevel@tonic-gate zone_free(zone); 43147c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 43157c478bd9Sstevel@tonic-gate } 43167c478bd9Sstevel@tonic-gate 43177c478bd9Sstevel@tonic-gate if ((error = zone_set_root(zone, zone_root)) != 0) { 43187c478bd9Sstevel@tonic-gate zone_free(zone); 43197c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 43207c478bd9Sstevel@tonic-gate } 4321821c4a97Sdp if ((error = zone_set_privset(zone, zone_privs, zone_privssz)) != 0) { 43227c478bd9Sstevel@tonic-gate zone_free(zone); 43237c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 43247c478bd9Sstevel@tonic-gate } 43257c478bd9Sstevel@tonic-gate 43267c478bd9Sstevel@tonic-gate /* initialize node name to be the same as zone name */ 43277c478bd9Sstevel@tonic-gate zone->zone_nodename = kmem_alloc(_SYS_NMLN, KM_SLEEP); 43287c478bd9Sstevel@tonic-gate (void) strncpy(zone->zone_nodename, zone->zone_name, _SYS_NMLN); 43297c478bd9Sstevel@tonic-gate zone->zone_nodename[_SYS_NMLN - 1] = '\0'; 43307c478bd9Sstevel@tonic-gate 43317c478bd9Sstevel@tonic-gate zone->zone_domain = kmem_alloc(_SYS_NMLN, KM_SLEEP); 43327c478bd9Sstevel@tonic-gate zone->zone_domain[0] = '\0'; 43335679c89fSjv zone->zone_hostid = HW_INVALID_HOSTID; 43347c478bd9Sstevel@tonic-gate zone->zone_shares = 1; 4335824c205fSml zone->zone_shmmax = 0; 4336824c205fSml zone->zone_ipc.ipcq_shmmni = 0; 4337824c205fSml zone->zone_ipc.ipcq_semmni = 0; 4338824c205fSml zone->zone_ipc.ipcq_msgmni = 0; 43397c478bd9Sstevel@tonic-gate zone->zone_bootargs = NULL; 43400fbb751dSJohn Levon zone->zone_fs_allowed = NULL; 43413f2f09c1Sdp zone->zone_initname = 43423f2f09c1Sdp kmem_alloc(strlen(zone_default_initname) + 1, KM_SLEEP); 43433f2f09c1Sdp (void) strcpy(zone->zone_initname, zone_default_initname); 43440209230bSgjelinek zone->zone_nlwps = 0; 43450209230bSgjelinek zone->zone_nlwps_ctl = INT_MAX; 4346ff19e029SMenno Lageman zone->zone_nprocs = 0; 4347ff19e029SMenno Lageman zone->zone_nprocs_ctl = INT_MAX; 4348c6939658Ssl zone->zone_locked_mem = 0; 4349c6939658Ssl zone->zone_locked_mem_ctl = UINT64_MAX; 43500209230bSgjelinek zone->zone_max_swap = 0; 43510209230bSgjelinek zone->zone_max_swap_ctl = UINT64_MAX; 43520fbb751dSJohn Levon zone->zone_max_lofi = 0; 43530fbb751dSJohn Levon zone->zone_max_lofi_ctl = UINT64_MAX; 43540209230bSgjelinek zone0.zone_lockedmem_kstat = NULL; 43550209230bSgjelinek zone0.zone_swapresv_kstat = NULL; 43567c478bd9Sstevel@tonic-gate 43577c478bd9Sstevel@tonic-gate /* 43587c478bd9Sstevel@tonic-gate * Zsched initializes the rctls. 43597c478bd9Sstevel@tonic-gate */ 43607c478bd9Sstevel@tonic-gate zone->zone_rctls = NULL; 43617c478bd9Sstevel@tonic-gate 43627c478bd9Sstevel@tonic-gate if ((error = parse_rctls(rctlbuf, rctlbufsz, &rctls)) != 0) { 43637c478bd9Sstevel@tonic-gate zone_free(zone); 43647c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 43657c478bd9Sstevel@tonic-gate } 43667c478bd9Sstevel@tonic-gate 4367fa9e4066Sahrens if ((error = parse_zfs(zone, zfsbuf, zfsbufsz)) != 0) { 4368fa9e4066Sahrens zone_free(zone); 4369fa9e4066Sahrens return (set_errno(error)); 4370fa9e4066Sahrens } 4371fa9e4066Sahrens 437245916cd2Sjpk /* 437345916cd2Sjpk * Read in the trusted system parameters: 437445916cd2Sjpk * match flag and sensitivity label. 437545916cd2Sjpk */ 437645916cd2Sjpk zone->zone_match = match; 437748451833Scarlsonj if (is_system_labeled() && !(zone->zone_flags & ZF_IS_SCRATCH)) { 43787e6639c2Skp /* Fail if requested to set doi to anything but system's doi */ 43797e6639c2Skp if (doi != 0 && doi != default_doi) { 43807e6639c2Skp zone_free(zone); 43817e6639c2Skp return (set_errno(EINVAL)); 43827e6639c2Skp } 43837e6639c2Skp /* Always apply system's doi to the zone */ 43847e6639c2Skp error = zone_set_label(zone, label, default_doi); 438545916cd2Sjpk if (error != 0) { 438645916cd2Sjpk zone_free(zone); 438745916cd2Sjpk return (set_errno(error)); 438845916cd2Sjpk } 438948451833Scarlsonj insert_label_hash = B_TRUE; 439045916cd2Sjpk } else { 439145916cd2Sjpk /* all zones get an admin_low label if system is not labeled */ 439245916cd2Sjpk zone->zone_slabel = l_admin_low; 439345916cd2Sjpk label_hold(l_admin_low); 439448451833Scarlsonj insert_label_hash = B_FALSE; 439545916cd2Sjpk } 439645916cd2Sjpk 43977c478bd9Sstevel@tonic-gate /* 43987c478bd9Sstevel@tonic-gate * Stop all lwps since that's what normally happens as part of fork(). 43997c478bd9Sstevel@tonic-gate * This needs to happen before we grab any locks to avoid deadlock 44007c478bd9Sstevel@tonic-gate * (another lwp in the process could be waiting for the held lock). 44017c478bd9Sstevel@tonic-gate */ 44027c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp && !holdlwps(SHOLDFORK)) { 44037c478bd9Sstevel@tonic-gate zone_free(zone); 44047c478bd9Sstevel@tonic-gate if (rctls) 44057c478bd9Sstevel@tonic-gate nvlist_free(rctls); 44067c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 44077c478bd9Sstevel@tonic-gate } 44087c478bd9Sstevel@tonic-gate 4409*5fd5c689SJerry Jelinek if (block_mounts(zone) == 0) { 44107c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 44117c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp) 44127c478bd9Sstevel@tonic-gate continuelwps(pp); 44137c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 44147c478bd9Sstevel@tonic-gate zone_free(zone); 44157c478bd9Sstevel@tonic-gate if (rctls) 44167c478bd9Sstevel@tonic-gate nvlist_free(rctls); 44177c478bd9Sstevel@tonic-gate return (zone_create_error(error, 0, extended_error)); 44187c478bd9Sstevel@tonic-gate } 44197c478bd9Sstevel@tonic-gate 44207c478bd9Sstevel@tonic-gate /* 44217c478bd9Sstevel@tonic-gate * Set up credential for kernel access. After this, any errors 44227c478bd9Sstevel@tonic-gate * should go through the dance in errout rather than calling 44237c478bd9Sstevel@tonic-gate * zone_free directly. 44247c478bd9Sstevel@tonic-gate */ 44257c478bd9Sstevel@tonic-gate zone->zone_kcred = crdup(kcred); 44267c478bd9Sstevel@tonic-gate crsetzone(zone->zone_kcred, zone); 44277c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_PPRIV(zone->zone_kcred)); 44287c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_EPRIV(zone->zone_kcred)); 44297c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_IPRIV(zone->zone_kcred)); 44307c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_LPRIV(zone->zone_kcred)); 44317c478bd9Sstevel@tonic-gate 44327c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 44337c478bd9Sstevel@tonic-gate /* 44347c478bd9Sstevel@tonic-gate * Make sure zone doesn't already exist. 443545916cd2Sjpk * 443645916cd2Sjpk * If the system and zone are labeled, 443745916cd2Sjpk * make sure no other zone exists that has the same label. 44387c478bd9Sstevel@tonic-gate */ 443945916cd2Sjpk if ((ztmp = zone_find_all_by_name(zone->zone_name)) != NULL || 444048451833Scarlsonj (insert_label_hash && 444145916cd2Sjpk (ztmp = zone_find_all_by_label(zone->zone_slabel)) != NULL)) { 44427c478bd9Sstevel@tonic-gate zone_status_t status; 44437c478bd9Sstevel@tonic-gate 44447c478bd9Sstevel@tonic-gate status = zone_status_get(ztmp); 44457c478bd9Sstevel@tonic-gate if (status == ZONE_IS_READY || status == ZONE_IS_RUNNING) 44467c478bd9Sstevel@tonic-gate error = EEXIST; 44477c478bd9Sstevel@tonic-gate else 44487c478bd9Sstevel@tonic-gate error = EBUSY; 44498c55461bSton 44508c55461bSton if (insert_label_hash) 44518c55461bSton error2 = ZE_LABELINUSE; 44528c55461bSton 44537c478bd9Sstevel@tonic-gate goto errout; 44547c478bd9Sstevel@tonic-gate } 44557c478bd9Sstevel@tonic-gate 44567c478bd9Sstevel@tonic-gate /* 44577c478bd9Sstevel@tonic-gate * Don't allow zone creations which would cause one zone's rootpath to 44587c478bd9Sstevel@tonic-gate * be accessible from that of another (non-global) zone. 44597c478bd9Sstevel@tonic-gate */ 44607c478bd9Sstevel@tonic-gate if (zone_is_nested(zone->zone_rootpath)) { 44617c478bd9Sstevel@tonic-gate error = EBUSY; 44627c478bd9Sstevel@tonic-gate goto errout; 44637c478bd9Sstevel@tonic-gate } 44647c478bd9Sstevel@tonic-gate 44657c478bd9Sstevel@tonic-gate ASSERT(zonecount != 0); /* check for leaks */ 44667c478bd9Sstevel@tonic-gate if (zonecount + 1 > maxzones) { 44677c478bd9Sstevel@tonic-gate error = ENOMEM; 44687c478bd9Sstevel@tonic-gate goto errout; 44697c478bd9Sstevel@tonic-gate } 44707c478bd9Sstevel@tonic-gate 44717c478bd9Sstevel@tonic-gate if (zone_mount_count(zone->zone_rootpath) != 0) { 44727c478bd9Sstevel@tonic-gate error = EBUSY; 44737c478bd9Sstevel@tonic-gate error2 = ZE_AREMOUNTS; 44747c478bd9Sstevel@tonic-gate goto errout; 44757c478bd9Sstevel@tonic-gate } 44767c478bd9Sstevel@tonic-gate 44777c478bd9Sstevel@tonic-gate /* 44787c478bd9Sstevel@tonic-gate * Zone is still incomplete, but we need to drop all locks while 44797c478bd9Sstevel@tonic-gate * zsched() initializes this zone's kernel process. We 44807c478bd9Sstevel@tonic-gate * optimistically add the zone to the hashtable and associated 44817c478bd9Sstevel@tonic-gate * lists so a parallel zone_create() doesn't try to create the 44827c478bd9Sstevel@tonic-gate * same zone. 44837c478bd9Sstevel@tonic-gate */ 44847c478bd9Sstevel@tonic-gate zonecount++; 44857c478bd9Sstevel@tonic-gate (void) mod_hash_insert(zonehashbyid, 44867c478bd9Sstevel@tonic-gate (mod_hash_key_t)(uintptr_t)zone->zone_id, 44877c478bd9Sstevel@tonic-gate (mod_hash_val_t)(uintptr_t)zone); 44887c478bd9Sstevel@tonic-gate str = kmem_alloc(strlen(zone->zone_name) + 1, KM_SLEEP); 44897c478bd9Sstevel@tonic-gate (void) strcpy(str, zone->zone_name); 44907c478bd9Sstevel@tonic-gate (void) mod_hash_insert(zonehashbyname, (mod_hash_key_t)str, 44917c478bd9Sstevel@tonic-gate (mod_hash_val_t)(uintptr_t)zone); 449248451833Scarlsonj if (insert_label_hash) { 449345916cd2Sjpk (void) mod_hash_insert(zonehashbylabel, 449445916cd2Sjpk (mod_hash_key_t)zone->zone_slabel, (mod_hash_val_t)zone); 449548451833Scarlsonj zone->zone_flags |= ZF_HASHED_LABEL; 449645916cd2Sjpk } 449745916cd2Sjpk 44987c478bd9Sstevel@tonic-gate /* 44997c478bd9Sstevel@tonic-gate * Insert into active list. At this point there are no 'hold's 45007c478bd9Sstevel@tonic-gate * on the zone, but everyone else knows not to use it, so we can 45017c478bd9Sstevel@tonic-gate * continue to use it. zsched() will do a zone_hold() if the 45027c478bd9Sstevel@tonic-gate * newproc() is successful. 45037c478bd9Sstevel@tonic-gate */ 45047c478bd9Sstevel@tonic-gate list_insert_tail(&zone_active, zone); 45057c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 45067c478bd9Sstevel@tonic-gate 45077c478bd9Sstevel@tonic-gate zarg.zone = zone; 45087c478bd9Sstevel@tonic-gate zarg.nvlist = rctls; 45097c478bd9Sstevel@tonic-gate /* 45107c478bd9Sstevel@tonic-gate * The process, task, and project rctls are probably wrong; 45117c478bd9Sstevel@tonic-gate * we need an interface to get the default values of all rctls, 45127c478bd9Sstevel@tonic-gate * and initialize zsched appropriately. I'm not sure that that 45137c478bd9Sstevel@tonic-gate * makes much of a difference, though. 45147c478bd9Sstevel@tonic-gate */ 451535a5a358SJonathan Adams error = newproc(zsched, (void *)&zarg, syscid, minclsyspri, NULL, 0); 451635a5a358SJonathan Adams if (error != 0) { 45177c478bd9Sstevel@tonic-gate /* 45187c478bd9Sstevel@tonic-gate * We need to undo all globally visible state. 45197c478bd9Sstevel@tonic-gate */ 45207c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 45217c478bd9Sstevel@tonic-gate list_remove(&zone_active, zone); 452248451833Scarlsonj if (zone->zone_flags & ZF_HASHED_LABEL) { 452345916cd2Sjpk ASSERT(zone->zone_slabel != NULL); 452445916cd2Sjpk (void) mod_hash_destroy(zonehashbylabel, 452545916cd2Sjpk (mod_hash_key_t)zone->zone_slabel); 452645916cd2Sjpk } 45277c478bd9Sstevel@tonic-gate (void) mod_hash_destroy(zonehashbyname, 45287c478bd9Sstevel@tonic-gate (mod_hash_key_t)(uintptr_t)zone->zone_name); 45297c478bd9Sstevel@tonic-gate (void) mod_hash_destroy(zonehashbyid, 45307c478bd9Sstevel@tonic-gate (mod_hash_key_t)(uintptr_t)zone->zone_id); 45317c478bd9Sstevel@tonic-gate ASSERT(zonecount > 1); 45327c478bd9Sstevel@tonic-gate zonecount--; 45337c478bd9Sstevel@tonic-gate goto errout; 45347c478bd9Sstevel@tonic-gate } 45357c478bd9Sstevel@tonic-gate 45367c478bd9Sstevel@tonic-gate /* 45377c478bd9Sstevel@tonic-gate * Zone creation can't fail from now on. 45387c478bd9Sstevel@tonic-gate */ 45397c478bd9Sstevel@tonic-gate 45400209230bSgjelinek /* 45410209230bSgjelinek * Create zone kstats 45420209230bSgjelinek */ 45430209230bSgjelinek zone_kstat_create(zone); 45440209230bSgjelinek 45457c478bd9Sstevel@tonic-gate /* 45467c478bd9Sstevel@tonic-gate * Let the other lwps continue. 45477c478bd9Sstevel@tonic-gate */ 45487c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 45497c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp) 45507c478bd9Sstevel@tonic-gate continuelwps(pp); 45517c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 45527c478bd9Sstevel@tonic-gate 45537c478bd9Sstevel@tonic-gate /* 45547c478bd9Sstevel@tonic-gate * Wait for zsched to finish initializing the zone. 45557c478bd9Sstevel@tonic-gate */ 45567c478bd9Sstevel@tonic-gate zone_status_wait(zone, ZONE_IS_READY); 45577c478bd9Sstevel@tonic-gate /* 45587c478bd9Sstevel@tonic-gate * The zone is fully visible, so we can let mounts progress. 45597c478bd9Sstevel@tonic-gate */ 4560*5fd5c689SJerry Jelinek resume_mounts(zone); 45617c478bd9Sstevel@tonic-gate if (rctls) 45627c478bd9Sstevel@tonic-gate nvlist_free(rctls); 45637c478bd9Sstevel@tonic-gate 45647c478bd9Sstevel@tonic-gate return (zoneid); 45657c478bd9Sstevel@tonic-gate 45667c478bd9Sstevel@tonic-gate errout: 45677c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 45687c478bd9Sstevel@tonic-gate /* 45697c478bd9Sstevel@tonic-gate * Let the other lwps continue. 45707c478bd9Sstevel@tonic-gate */ 45717c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 45727c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp) 45737c478bd9Sstevel@tonic-gate continuelwps(pp); 45747c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 45757c478bd9Sstevel@tonic-gate 4576*5fd5c689SJerry Jelinek resume_mounts(zone); 45777c478bd9Sstevel@tonic-gate if (rctls) 45787c478bd9Sstevel@tonic-gate nvlist_free(rctls); 45797c478bd9Sstevel@tonic-gate /* 45807c478bd9Sstevel@tonic-gate * There is currently one reference to the zone, a cred_ref from 45817c478bd9Sstevel@tonic-gate * zone_kcred. To free the zone, we call crfree, which will call 45827c478bd9Sstevel@tonic-gate * zone_cred_rele, which will call zone_free. 45837c478bd9Sstevel@tonic-gate */ 4584a19609f8Sjv ASSERT(zone->zone_cred_ref == 1); 45857c478bd9Sstevel@tonic-gate ASSERT(zone->zone_kcred->cr_ref == 1); 45867c478bd9Sstevel@tonic-gate ASSERT(zone->zone_ref == 0); 45877c478bd9Sstevel@tonic-gate zkcr = zone->zone_kcred; 45887c478bd9Sstevel@tonic-gate zone->zone_kcred = NULL; 45897c478bd9Sstevel@tonic-gate crfree(zkcr); /* triggers call to zone_free */ 45907c478bd9Sstevel@tonic-gate return (zone_create_error(error, error2, extended_error)); 45917c478bd9Sstevel@tonic-gate } 45927c478bd9Sstevel@tonic-gate 45937c478bd9Sstevel@tonic-gate /* 45947c478bd9Sstevel@tonic-gate * Cause the zone to boot. This is pretty simple, since we let zoneadmd do 45953f2f09c1Sdp * the heavy lifting. initname is the path to the program to launch 45963f2f09c1Sdp * at the "top" of the zone; if this is NULL, we use the system default, 45973f2f09c1Sdp * which is stored at zone_default_initname. 45987c478bd9Sstevel@tonic-gate */ 45997c478bd9Sstevel@tonic-gate static int 46003f2f09c1Sdp zone_boot(zoneid_t zoneid) 46017c478bd9Sstevel@tonic-gate { 46027c478bd9Sstevel@tonic-gate int err; 46037c478bd9Sstevel@tonic-gate zone_t *zone; 46047c478bd9Sstevel@tonic-gate 46057c478bd9Sstevel@tonic-gate if (secpolicy_zone_config(CRED()) != 0) 46067c478bd9Sstevel@tonic-gate return (set_errno(EPERM)); 46077c478bd9Sstevel@tonic-gate if (zoneid < MIN_USERZONEID || zoneid > MAX_ZONEID) 46087c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 46097c478bd9Sstevel@tonic-gate 46107c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 46117c478bd9Sstevel@tonic-gate /* 46127c478bd9Sstevel@tonic-gate * Look for zone under hash lock to prevent races with calls to 46137c478bd9Sstevel@tonic-gate * zone_shutdown, zone_destroy, etc. 46147c478bd9Sstevel@tonic-gate */ 46157c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 46167c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 46177c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 46187c478bd9Sstevel@tonic-gate } 46197c478bd9Sstevel@tonic-gate 46207c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 46217c478bd9Sstevel@tonic-gate if (zone_status_get(zone) != ZONE_IS_READY) { 46227c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 46237c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 46247c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 46257c478bd9Sstevel@tonic-gate } 46267c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_BOOTING); 46277c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 46287c478bd9Sstevel@tonic-gate 46297c478bd9Sstevel@tonic-gate zone_hold(zone); /* so we can use the zone_t later */ 46307c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 46317c478bd9Sstevel@tonic-gate 46327c478bd9Sstevel@tonic-gate if (zone_status_wait_sig(zone, ZONE_IS_RUNNING) == 0) { 46337c478bd9Sstevel@tonic-gate zone_rele(zone); 46347c478bd9Sstevel@tonic-gate return (set_errno(EINTR)); 46357c478bd9Sstevel@tonic-gate } 46367c478bd9Sstevel@tonic-gate 46377c478bd9Sstevel@tonic-gate /* 46387c478bd9Sstevel@tonic-gate * Boot (starting init) might have failed, in which case the zone 46397c478bd9Sstevel@tonic-gate * will go to the SHUTTING_DOWN state; an appropriate errno will 46407c478bd9Sstevel@tonic-gate * be placed in zone->zone_boot_err, and so we return that. 46417c478bd9Sstevel@tonic-gate */ 46427c478bd9Sstevel@tonic-gate err = zone->zone_boot_err; 46437c478bd9Sstevel@tonic-gate zone_rele(zone); 46447c478bd9Sstevel@tonic-gate return (err ? set_errno(err) : 0); 46457c478bd9Sstevel@tonic-gate } 46467c478bd9Sstevel@tonic-gate 46477c478bd9Sstevel@tonic-gate /* 46487c478bd9Sstevel@tonic-gate * Kills all user processes in the zone, waiting for them all to exit 46497c478bd9Sstevel@tonic-gate * before returning. 46507c478bd9Sstevel@tonic-gate */ 46517c478bd9Sstevel@tonic-gate static int 46527c478bd9Sstevel@tonic-gate zone_empty(zone_t *zone) 46537c478bd9Sstevel@tonic-gate { 46547c478bd9Sstevel@tonic-gate int waitstatus; 46557c478bd9Sstevel@tonic-gate 46567c478bd9Sstevel@tonic-gate /* 46577c478bd9Sstevel@tonic-gate * We need to drop zonehash_lock before killing all 46587c478bd9Sstevel@tonic-gate * processes, otherwise we'll deadlock with zone_find_* 46597c478bd9Sstevel@tonic-gate * which can be called from the exit path. 46607c478bd9Sstevel@tonic-gate */ 46617c478bd9Sstevel@tonic-gate ASSERT(MUTEX_NOT_HELD(&zonehash_lock)); 4662d3d50737SRafael Vanoni while ((waitstatus = zone_status_timedwait_sig(zone, 4663d3d50737SRafael Vanoni ddi_get_lbolt() + hz, ZONE_IS_EMPTY)) == -1) { 46647c478bd9Sstevel@tonic-gate killall(zone->zone_id); 46657c478bd9Sstevel@tonic-gate } 46667c478bd9Sstevel@tonic-gate /* 46677c478bd9Sstevel@tonic-gate * return EINTR if we were signaled 46687c478bd9Sstevel@tonic-gate */ 46697c478bd9Sstevel@tonic-gate if (waitstatus == 0) 46707c478bd9Sstevel@tonic-gate return (EINTR); 46717c478bd9Sstevel@tonic-gate return (0); 46727c478bd9Sstevel@tonic-gate } 46737c478bd9Sstevel@tonic-gate 467445916cd2Sjpk /* 467545916cd2Sjpk * This function implements the policy for zone visibility. 467645916cd2Sjpk * 467745916cd2Sjpk * In standard Solaris, a non-global zone can only see itself. 467845916cd2Sjpk * 467945916cd2Sjpk * In Trusted Extensions, a labeled zone can lookup any zone whose label 468045916cd2Sjpk * it dominates. For this test, the label of the global zone is treated as 468145916cd2Sjpk * admin_high so it is special-cased instead of being checked for dominance. 468245916cd2Sjpk * 468345916cd2Sjpk * Returns true if zone attributes are viewable, false otherwise. 468445916cd2Sjpk */ 468545916cd2Sjpk static boolean_t 468645916cd2Sjpk zone_list_access(zone_t *zone) 468745916cd2Sjpk { 468845916cd2Sjpk 468945916cd2Sjpk if (curproc->p_zone == global_zone || 469045916cd2Sjpk curproc->p_zone == zone) { 469145916cd2Sjpk return (B_TRUE); 469248451833Scarlsonj } else if (is_system_labeled() && !(zone->zone_flags & ZF_IS_SCRATCH)) { 469345916cd2Sjpk bslabel_t *curproc_label; 469445916cd2Sjpk bslabel_t *zone_label; 469545916cd2Sjpk 469645916cd2Sjpk curproc_label = label2bslabel(curproc->p_zone->zone_slabel); 469745916cd2Sjpk zone_label = label2bslabel(zone->zone_slabel); 469845916cd2Sjpk 469945916cd2Sjpk if (zone->zone_id != GLOBAL_ZONEID && 470045916cd2Sjpk bldominates(curproc_label, zone_label)) { 470145916cd2Sjpk return (B_TRUE); 470245916cd2Sjpk } else { 470345916cd2Sjpk return (B_FALSE); 470445916cd2Sjpk } 470545916cd2Sjpk } else { 470645916cd2Sjpk return (B_FALSE); 470745916cd2Sjpk } 470845916cd2Sjpk } 470945916cd2Sjpk 47107c478bd9Sstevel@tonic-gate /* 47117c478bd9Sstevel@tonic-gate * Systemcall to start the zone's halt sequence. By the time this 47127c478bd9Sstevel@tonic-gate * function successfully returns, all user processes and kernel threads 47137c478bd9Sstevel@tonic-gate * executing in it will have exited, ZSD shutdown callbacks executed, 47147c478bd9Sstevel@tonic-gate * and the zone status set to ZONE_IS_DOWN. 47157c478bd9Sstevel@tonic-gate * 47167c478bd9Sstevel@tonic-gate * It is possible that the call will interrupt itself if the caller is the 47177c478bd9Sstevel@tonic-gate * parent of any process running in the zone, and doesn't have SIGCHLD blocked. 47187c478bd9Sstevel@tonic-gate */ 47197c478bd9Sstevel@tonic-gate static int 47207c478bd9Sstevel@tonic-gate zone_shutdown(zoneid_t zoneid) 47217c478bd9Sstevel@tonic-gate { 47227c478bd9Sstevel@tonic-gate int error; 47237c478bd9Sstevel@tonic-gate zone_t *zone; 47247c478bd9Sstevel@tonic-gate zone_status_t status; 47257c478bd9Sstevel@tonic-gate 47267c478bd9Sstevel@tonic-gate if (secpolicy_zone_config(CRED()) != 0) 47277c478bd9Sstevel@tonic-gate return (set_errno(EPERM)); 47287c478bd9Sstevel@tonic-gate if (zoneid < MIN_USERZONEID || zoneid > MAX_ZONEID) 47297c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 47307c478bd9Sstevel@tonic-gate 47317c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 47327c478bd9Sstevel@tonic-gate /* 47337c478bd9Sstevel@tonic-gate * Look for zone under hash lock to prevent races with other 47347c478bd9Sstevel@tonic-gate * calls to zone_shutdown and zone_destroy. 47357c478bd9Sstevel@tonic-gate */ 47367c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 47377c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 47387c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 47397c478bd9Sstevel@tonic-gate } 4740*5fd5c689SJerry Jelinek 4741*5fd5c689SJerry Jelinek /* 4742*5fd5c689SJerry Jelinek * We have to drop zonehash_lock before calling block_mounts. 4743*5fd5c689SJerry Jelinek * Hold the zone so we can continue to use the zone_t. 4744*5fd5c689SJerry Jelinek */ 4745*5fd5c689SJerry Jelinek zone_hold(zone); 4746*5fd5c689SJerry Jelinek mutex_exit(&zonehash_lock); 4747*5fd5c689SJerry Jelinek 4748*5fd5c689SJerry Jelinek /* 4749*5fd5c689SJerry Jelinek * Block mounts so that VFS_MOUNT() can get an accurate view of 4750*5fd5c689SJerry Jelinek * the zone's status with regards to ZONE_IS_SHUTTING down. 4751*5fd5c689SJerry Jelinek * 4752*5fd5c689SJerry Jelinek * e.g. NFS can fail the mount if it determines that the zone 4753*5fd5c689SJerry Jelinek * has already begun the shutdown sequence. 4754*5fd5c689SJerry Jelinek * 4755*5fd5c689SJerry Jelinek */ 4756*5fd5c689SJerry Jelinek if (block_mounts(zone) == 0) { 4757*5fd5c689SJerry Jelinek zone_rele(zone); 4758*5fd5c689SJerry Jelinek return (set_errno(EINTR)); 4759*5fd5c689SJerry Jelinek } 4760*5fd5c689SJerry Jelinek 4761*5fd5c689SJerry Jelinek mutex_enter(&zonehash_lock); 47627c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 47637c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 47647c478bd9Sstevel@tonic-gate /* 47657c478bd9Sstevel@tonic-gate * Fail if the zone isn't fully initialized yet. 47667c478bd9Sstevel@tonic-gate */ 47677c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY) { 47687c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 47697c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 4770*5fd5c689SJerry Jelinek resume_mounts(zone); 4771*5fd5c689SJerry Jelinek zone_rele(zone); 47727c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 47737c478bd9Sstevel@tonic-gate } 47747c478bd9Sstevel@tonic-gate /* 47757c478bd9Sstevel@tonic-gate * If conditions required for zone_shutdown() to return have been met, 47767c478bd9Sstevel@tonic-gate * return success. 47777c478bd9Sstevel@tonic-gate */ 47787c478bd9Sstevel@tonic-gate if (status >= ZONE_IS_DOWN) { 47797c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 47807c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 4781*5fd5c689SJerry Jelinek resume_mounts(zone); 4782*5fd5c689SJerry Jelinek zone_rele(zone); 47837c478bd9Sstevel@tonic-gate return (0); 47847c478bd9Sstevel@tonic-gate } 47857c478bd9Sstevel@tonic-gate /* 47867c478bd9Sstevel@tonic-gate * If zone_shutdown() hasn't been called before, go through the motions. 47877c478bd9Sstevel@tonic-gate * If it has, there's nothing to do but wait for the kernel threads to 47887c478bd9Sstevel@tonic-gate * drain. 47897c478bd9Sstevel@tonic-gate */ 47907c478bd9Sstevel@tonic-gate if (status < ZONE_IS_EMPTY) { 47917c478bd9Sstevel@tonic-gate uint_t ntasks; 47927c478bd9Sstevel@tonic-gate 47937c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 47947c478bd9Sstevel@tonic-gate if ((ntasks = zone->zone_ntasks) != 1) { 47957c478bd9Sstevel@tonic-gate /* 47967c478bd9Sstevel@tonic-gate * There's still stuff running. 47977c478bd9Sstevel@tonic-gate */ 47987c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_SHUTTING_DOWN); 47997c478bd9Sstevel@tonic-gate } 48007c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 48017c478bd9Sstevel@tonic-gate if (ntasks == 1) { 48027c478bd9Sstevel@tonic-gate /* 48037c478bd9Sstevel@tonic-gate * The only way to create another task is through 48047c478bd9Sstevel@tonic-gate * zone_enter(), which will block until we drop 48057c478bd9Sstevel@tonic-gate * zonehash_lock. The zone is empty. 48067c478bd9Sstevel@tonic-gate */ 48077c478bd9Sstevel@tonic-gate if (zone->zone_kthreads == NULL) { 48087c478bd9Sstevel@tonic-gate /* 48097c478bd9Sstevel@tonic-gate * Skip ahead to ZONE_IS_DOWN 48107c478bd9Sstevel@tonic-gate */ 48117c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_DOWN); 48127c478bd9Sstevel@tonic-gate } else { 48137c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_EMPTY); 48147c478bd9Sstevel@tonic-gate } 48157c478bd9Sstevel@tonic-gate } 48167c478bd9Sstevel@tonic-gate } 48177c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 48187c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 4819*5fd5c689SJerry Jelinek resume_mounts(zone); 48207c478bd9Sstevel@tonic-gate 48217c478bd9Sstevel@tonic-gate if (error = zone_empty(zone)) { 48227c478bd9Sstevel@tonic-gate zone_rele(zone); 48237c478bd9Sstevel@tonic-gate return (set_errno(error)); 48247c478bd9Sstevel@tonic-gate } 48257c478bd9Sstevel@tonic-gate /* 48267c478bd9Sstevel@tonic-gate * After the zone status goes to ZONE_IS_DOWN this zone will no 48277c478bd9Sstevel@tonic-gate * longer be notified of changes to the pools configuration, so 48287c478bd9Sstevel@tonic-gate * in order to not end up with a stale pool pointer, we point 48297c478bd9Sstevel@tonic-gate * ourselves at the default pool and remove all resource 48307c478bd9Sstevel@tonic-gate * visibility. This is especially important as the zone_t may 48317c478bd9Sstevel@tonic-gate * languish on the deathrow for a very long time waiting for 48327c478bd9Sstevel@tonic-gate * cred's to drain out. 48337c478bd9Sstevel@tonic-gate * 48347c478bd9Sstevel@tonic-gate * This rebinding of the zone can happen multiple times 48357c478bd9Sstevel@tonic-gate * (presumably due to interrupted or parallel systemcalls) 48367c478bd9Sstevel@tonic-gate * without any adverse effects. 48377c478bd9Sstevel@tonic-gate */ 48387c478bd9Sstevel@tonic-gate if (pool_lock_intr() != 0) { 48397c478bd9Sstevel@tonic-gate zone_rele(zone); 48407c478bd9Sstevel@tonic-gate return (set_errno(EINTR)); 48417c478bd9Sstevel@tonic-gate } 48427c478bd9Sstevel@tonic-gate if (pool_state == POOL_ENABLED) { 48437c478bd9Sstevel@tonic-gate mutex_enter(&cpu_lock); 48447c478bd9Sstevel@tonic-gate zone_pool_set(zone, pool_default); 48457c478bd9Sstevel@tonic-gate /* 48467c478bd9Sstevel@tonic-gate * The zone no longer needs to be able to see any cpus. 48477c478bd9Sstevel@tonic-gate */ 48487c478bd9Sstevel@tonic-gate zone_pset_set(zone, ZONE_PS_INVAL); 48497c478bd9Sstevel@tonic-gate mutex_exit(&cpu_lock); 48507c478bd9Sstevel@tonic-gate } 48517c478bd9Sstevel@tonic-gate pool_unlock(); 48527c478bd9Sstevel@tonic-gate 48537c478bd9Sstevel@tonic-gate /* 48547c478bd9Sstevel@tonic-gate * ZSD shutdown callbacks can be executed multiple times, hence 48557c478bd9Sstevel@tonic-gate * it is safe to not be holding any locks across this call. 48567c478bd9Sstevel@tonic-gate */ 48577c478bd9Sstevel@tonic-gate zone_zsd_callbacks(zone, ZSD_SHUTDOWN); 48587c478bd9Sstevel@tonic-gate 48597c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 48607c478bd9Sstevel@tonic-gate if (zone->zone_kthreads == NULL && zone_status_get(zone) < ZONE_IS_DOWN) 48617c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_DOWN); 48627c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 48637c478bd9Sstevel@tonic-gate 48647c478bd9Sstevel@tonic-gate /* 48657c478bd9Sstevel@tonic-gate * Wait for kernel threads to drain. 48667c478bd9Sstevel@tonic-gate */ 48677c478bd9Sstevel@tonic-gate if (!zone_status_wait_sig(zone, ZONE_IS_DOWN)) { 48687c478bd9Sstevel@tonic-gate zone_rele(zone); 48697c478bd9Sstevel@tonic-gate return (set_errno(EINTR)); 48707c478bd9Sstevel@tonic-gate } 48719acbbeafSnn 487252978630Ssl /* 487352978630Ssl * Zone can be become down/destroyable even if the above wait 487452978630Ssl * returns EINTR, so any code added here may never execute. 487552978630Ssl * (i.e. don't add code here) 487652978630Ssl */ 48779acbbeafSnn 48787c478bd9Sstevel@tonic-gate zone_rele(zone); 48797c478bd9Sstevel@tonic-gate return (0); 48807c478bd9Sstevel@tonic-gate } 48817c478bd9Sstevel@tonic-gate 4882a19609f8Sjv /* 4883a19609f8Sjv * Log the specified zone's reference counts. The caller should not be 4884a19609f8Sjv * holding the zone's zone_lock. 4885a19609f8Sjv */ 4886a19609f8Sjv static void 4887a19609f8Sjv zone_log_refcounts(zone_t *zone) 4888a19609f8Sjv { 4889a19609f8Sjv char *buffer; 4890a19609f8Sjv char *buffer_position; 4891a19609f8Sjv uint32_t buffer_size; 4892a19609f8Sjv uint32_t index; 4893a19609f8Sjv uint_t ref; 4894a19609f8Sjv uint_t cred_ref; 4895a19609f8Sjv 4896a19609f8Sjv /* 4897a19609f8Sjv * Construct a string representing the subsystem-specific reference 4898a19609f8Sjv * counts. The counts are printed in ascending order by index into the 4899a19609f8Sjv * zone_t::zone_subsys_ref array. The list will be surrounded by 4900a19609f8Sjv * square brackets [] and will only contain nonzero reference counts. 4901a19609f8Sjv * 4902a19609f8Sjv * The buffer will hold two square bracket characters plus ten digits, 4903a19609f8Sjv * one colon, one space, one comma, and some characters for a 4904a19609f8Sjv * subsystem name per subsystem-specific reference count. (Unsigned 32- 4905a19609f8Sjv * bit integers have at most ten decimal digits.) The last 4906a19609f8Sjv * reference count's comma is replaced by the closing square 4907a19609f8Sjv * bracket and a NULL character to terminate the string. 4908a19609f8Sjv * 4909a19609f8Sjv * NOTE: We have to grab the zone's zone_lock to create a consistent 4910a19609f8Sjv * snapshot of the zone's reference counters. 4911a19609f8Sjv * 4912a19609f8Sjv * First, figure out how much space the string buffer will need. 4913a19609f8Sjv * The buffer's size is stored in buffer_size. 4914a19609f8Sjv */ 4915a19609f8Sjv buffer_size = 2; /* for the square brackets */ 4916a19609f8Sjv mutex_enter(&zone->zone_lock); 4917a19609f8Sjv zone->zone_flags |= ZF_REFCOUNTS_LOGGED; 4918a19609f8Sjv ref = zone->zone_ref; 4919a19609f8Sjv cred_ref = zone->zone_cred_ref; 4920a19609f8Sjv for (index = 0; index < ZONE_REF_NUM_SUBSYS; ++index) 4921a19609f8Sjv if (zone->zone_subsys_ref[index] != 0) 4922a19609f8Sjv buffer_size += strlen(zone_ref_subsys_names[index]) + 4923a19609f8Sjv 13; 4924a19609f8Sjv if (buffer_size == 2) { 4925a19609f8Sjv /* 4926a19609f8Sjv * No subsystems had nonzero reference counts. Don't bother 4927a19609f8Sjv * with allocating a buffer; just log the general-purpose and 4928a19609f8Sjv * credential reference counts. 4929a19609f8Sjv */ 4930a19609f8Sjv mutex_exit(&zone->zone_lock); 4931a19609f8Sjv (void) strlog(0, 0, 1, SL_CONSOLE | SL_NOTE, 4932a19609f8Sjv "Zone '%s' (ID: %d) is shutting down, but %u zone " 4933a19609f8Sjv "references and %u credential references are still extant", 4934a19609f8Sjv zone->zone_name, zone->zone_id, ref, cred_ref); 4935a19609f8Sjv return; 4936a19609f8Sjv } 4937a19609f8Sjv 4938a19609f8Sjv /* 4939a19609f8Sjv * buffer_size contains the exact number of characters that the 4940a19609f8Sjv * buffer will need. Allocate the buffer and fill it with nonzero 4941a19609f8Sjv * subsystem-specific reference counts. Surround the results with 4942a19609f8Sjv * square brackets afterwards. 4943a19609f8Sjv */ 4944a19609f8Sjv buffer = kmem_alloc(buffer_size, KM_SLEEP); 4945a19609f8Sjv buffer_position = &buffer[1]; 4946a19609f8Sjv for (index = 0; index < ZONE_REF_NUM_SUBSYS; ++index) { 4947a19609f8Sjv /* 4948a19609f8Sjv * NOTE: The DDI's version of sprintf() returns a pointer to 4949a19609f8Sjv * the modified buffer rather than the number of bytes written 4950a19609f8Sjv * (as in snprintf(3C)). This is unfortunate and annoying. 4951a19609f8Sjv * Therefore, we'll use snprintf() with INT_MAX to get the 4952a19609f8Sjv * number of bytes written. Using INT_MAX is safe because 4953a19609f8Sjv * the buffer is perfectly sized for the data: we'll never 4954a19609f8Sjv * overrun the buffer. 4955a19609f8Sjv */ 4956a19609f8Sjv if (zone->zone_subsys_ref[index] != 0) 4957a19609f8Sjv buffer_position += snprintf(buffer_position, INT_MAX, 4958a19609f8Sjv "%s: %u,", zone_ref_subsys_names[index], 4959a19609f8Sjv zone->zone_subsys_ref[index]); 4960a19609f8Sjv } 4961a19609f8Sjv mutex_exit(&zone->zone_lock); 4962a19609f8Sjv buffer[0] = '['; 4963a19609f8Sjv ASSERT((uintptr_t)(buffer_position - buffer) < buffer_size); 4964a19609f8Sjv ASSERT(buffer_position[0] == '\0' && buffer_position[-1] == ','); 4965a19609f8Sjv buffer_position[-1] = ']'; 4966a19609f8Sjv 4967a19609f8Sjv /* 4968a19609f8Sjv * Log the reference counts and free the message buffer. 4969a19609f8Sjv */ 4970a19609f8Sjv (void) strlog(0, 0, 1, SL_CONSOLE | SL_NOTE, 4971a19609f8Sjv "Zone '%s' (ID: %d) is shutting down, but %u zone references and " 4972a19609f8Sjv "%u credential references are still extant %s", zone->zone_name, 4973a19609f8Sjv zone->zone_id, ref, cred_ref, buffer); 4974a19609f8Sjv kmem_free(buffer, buffer_size); 4975a19609f8Sjv } 4976a19609f8Sjv 49777c478bd9Sstevel@tonic-gate /* 49787c478bd9Sstevel@tonic-gate * Systemcall entry point to finalize the zone halt process. The caller 4979824c205fSml * must have already successfully called zone_shutdown(). 49807c478bd9Sstevel@tonic-gate * 49817c478bd9Sstevel@tonic-gate * Upon successful completion, the zone will have been fully destroyed: 49827c478bd9Sstevel@tonic-gate * zsched will have exited, destructor callbacks executed, and the zone 49837c478bd9Sstevel@tonic-gate * removed from the list of active zones. 49847c478bd9Sstevel@tonic-gate */ 49857c478bd9Sstevel@tonic-gate static int 49867c478bd9Sstevel@tonic-gate zone_destroy(zoneid_t zoneid) 49877c478bd9Sstevel@tonic-gate { 49887c478bd9Sstevel@tonic-gate uint64_t uniqid; 49897c478bd9Sstevel@tonic-gate zone_t *zone; 49907c478bd9Sstevel@tonic-gate zone_status_t status; 4991a19609f8Sjv clock_t wait_time; 4992a19609f8Sjv boolean_t log_refcounts; 49937c478bd9Sstevel@tonic-gate 49947c478bd9Sstevel@tonic-gate if (secpolicy_zone_config(CRED()) != 0) 49957c478bd9Sstevel@tonic-gate return (set_errno(EPERM)); 49967c478bd9Sstevel@tonic-gate if (zoneid < MIN_USERZONEID || zoneid > MAX_ZONEID) 49977c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 49987c478bd9Sstevel@tonic-gate 49997c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 50007c478bd9Sstevel@tonic-gate /* 50017c478bd9Sstevel@tonic-gate * Look for zone under hash lock to prevent races with other 50027c478bd9Sstevel@tonic-gate * calls to zone_destroy. 50037c478bd9Sstevel@tonic-gate */ 50047c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 50057c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 50067c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 50077c478bd9Sstevel@tonic-gate } 50087c478bd9Sstevel@tonic-gate 50097c478bd9Sstevel@tonic-gate if (zone_mount_count(zone->zone_rootpath) != 0) { 50107c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 50117c478bd9Sstevel@tonic-gate return (set_errno(EBUSY)); 50127c478bd9Sstevel@tonic-gate } 50137c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 50147c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 50157c478bd9Sstevel@tonic-gate if (status < ZONE_IS_DOWN) { 50167c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 50177c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 50187c478bd9Sstevel@tonic-gate return (set_errno(EBUSY)); 50197c478bd9Sstevel@tonic-gate } else if (status == ZONE_IS_DOWN) { 50207c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_DYING); /* Tell zsched to exit */ 50217c478bd9Sstevel@tonic-gate } 50227c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 50237c478bd9Sstevel@tonic-gate zone_hold(zone); 50247c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 50257c478bd9Sstevel@tonic-gate 50267c478bd9Sstevel@tonic-gate /* 50277c478bd9Sstevel@tonic-gate * wait for zsched to exit 50287c478bd9Sstevel@tonic-gate */ 50297c478bd9Sstevel@tonic-gate zone_status_wait(zone, ZONE_IS_DEAD); 50307c478bd9Sstevel@tonic-gate zone_zsd_callbacks(zone, ZSD_DESTROY); 5031f4b3ec61Sdh zone->zone_netstack = NULL; 50327c478bd9Sstevel@tonic-gate uniqid = zone->zone_uniqid; 50337c478bd9Sstevel@tonic-gate zone_rele(zone); 50347c478bd9Sstevel@tonic-gate zone = NULL; /* potentially free'd */ 50357c478bd9Sstevel@tonic-gate 5036a19609f8Sjv log_refcounts = B_FALSE; 5037a19609f8Sjv wait_time = SEC_TO_TICK(ZONE_DESTROY_TIMEOUT_SECS); 50387c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 50397c478bd9Sstevel@tonic-gate for (; /* ever */; ) { 50407c478bd9Sstevel@tonic-gate boolean_t unref; 5041a19609f8Sjv boolean_t refs_have_been_logged; 50427c478bd9Sstevel@tonic-gate 50437c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL || 50447c478bd9Sstevel@tonic-gate zone->zone_uniqid != uniqid) { 50457c478bd9Sstevel@tonic-gate /* 50467c478bd9Sstevel@tonic-gate * The zone has gone away. Necessary conditions 50477c478bd9Sstevel@tonic-gate * are met, so we return success. 50487c478bd9Sstevel@tonic-gate */ 50497c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 50507c478bd9Sstevel@tonic-gate return (0); 50517c478bd9Sstevel@tonic-gate } 50527c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_lock); 50537c478bd9Sstevel@tonic-gate unref = ZONE_IS_UNREF(zone); 5054a19609f8Sjv refs_have_been_logged = (zone->zone_flags & 5055a19609f8Sjv ZF_REFCOUNTS_LOGGED); 50567c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_lock); 50577c478bd9Sstevel@tonic-gate if (unref) { 50587c478bd9Sstevel@tonic-gate /* 50597c478bd9Sstevel@tonic-gate * There is only one reference to the zone -- that 50607c478bd9Sstevel@tonic-gate * added when the zone was added to the hashtables -- 50617c478bd9Sstevel@tonic-gate * and things will remain this way until we drop 50627c478bd9Sstevel@tonic-gate * zonehash_lock... we can go ahead and cleanup the 50637c478bd9Sstevel@tonic-gate * zone. 50647c478bd9Sstevel@tonic-gate */ 50657c478bd9Sstevel@tonic-gate break; 50667c478bd9Sstevel@tonic-gate } 50677c478bd9Sstevel@tonic-gate 5068a19609f8Sjv /* 5069a19609f8Sjv * Wait for zone_rele_common() or zone_cred_rele() to signal 5070a19609f8Sjv * zone_destroy_cv. zone_destroy_cv is signaled only when 5071a19609f8Sjv * some zone's general-purpose reference count reaches one. 5072a19609f8Sjv * If ZONE_DESTROY_TIMEOUT_SECS seconds elapse while waiting 5073a19609f8Sjv * on zone_destroy_cv, then log the zone's reference counts and 5074a19609f8Sjv * continue to wait for zone_rele() and zone_cred_rele(). 5075a19609f8Sjv */ 5076a19609f8Sjv if (!refs_have_been_logged) { 5077a19609f8Sjv if (!log_refcounts) { 5078a19609f8Sjv /* 5079a19609f8Sjv * This thread hasn't timed out waiting on 5080a19609f8Sjv * zone_destroy_cv yet. Wait wait_time clock 5081a19609f8Sjv * ticks (initially ZONE_DESTROY_TIMEOUT_SECS 5082a19609f8Sjv * seconds) for the zone's references to clear. 5083a19609f8Sjv */ 5084a19609f8Sjv ASSERT(wait_time > 0); 5085a19609f8Sjv wait_time = cv_reltimedwait_sig( 5086a19609f8Sjv &zone_destroy_cv, &zonehash_lock, wait_time, 5087a19609f8Sjv TR_SEC); 5088a19609f8Sjv if (wait_time > 0) { 5089a19609f8Sjv /* 5090a19609f8Sjv * A thread in zone_rele() or 5091a19609f8Sjv * zone_cred_rele() signaled 5092a19609f8Sjv * zone_destroy_cv before this thread's 5093a19609f8Sjv * wait timed out. The zone might have 5094a19609f8Sjv * only one reference left; find out! 5095a19609f8Sjv */ 5096a19609f8Sjv continue; 5097a19609f8Sjv } else if (wait_time == 0) { 5098a19609f8Sjv /* The thread's process was signaled. */ 5099a19609f8Sjv mutex_exit(&zonehash_lock); 5100a19609f8Sjv return (set_errno(EINTR)); 5101a19609f8Sjv } 5102a19609f8Sjv 5103a19609f8Sjv /* 5104a19609f8Sjv * The thread timed out while waiting on 5105a19609f8Sjv * zone_destroy_cv. Even though the thread 5106a19609f8Sjv * timed out, it has to check whether another 5107a19609f8Sjv * thread woke up from zone_destroy_cv and 5108a19609f8Sjv * destroyed the zone. 5109a19609f8Sjv * 5110a19609f8Sjv * If the zone still exists and has more than 5111a19609f8Sjv * one unreleased general-purpose reference, 5112a19609f8Sjv * then log the zone's reference counts. 5113a19609f8Sjv */ 5114a19609f8Sjv log_refcounts = B_TRUE; 5115a19609f8Sjv continue; 5116a19609f8Sjv } 5117a19609f8Sjv 5118a19609f8Sjv /* 5119a19609f8Sjv * The thread already timed out on zone_destroy_cv while 5120a19609f8Sjv * waiting for subsystems to release the zone's last 5121a19609f8Sjv * general-purpose references. Log the zone's reference 5122a19609f8Sjv * counts and wait indefinitely on zone_destroy_cv. 5123a19609f8Sjv */ 5124a19609f8Sjv zone_log_refcounts(zone); 5125a19609f8Sjv } 51267c478bd9Sstevel@tonic-gate if (cv_wait_sig(&zone_destroy_cv, &zonehash_lock) == 0) { 5127a19609f8Sjv /* The thread's process was signaled. */ 51287c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 51297c478bd9Sstevel@tonic-gate return (set_errno(EINTR)); 51307c478bd9Sstevel@tonic-gate } 51317c478bd9Sstevel@tonic-gate } 51327c478bd9Sstevel@tonic-gate 5133c97ad5cdSakolb /* 5134c97ad5cdSakolb * Remove CPU cap for this zone now since we're not going to 5135c97ad5cdSakolb * fail below this point. 5136c97ad5cdSakolb */ 5137c97ad5cdSakolb cpucaps_zone_remove(zone); 5138c97ad5cdSakolb 5139c97ad5cdSakolb /* Get rid of the zone's kstats */ 51400209230bSgjelinek zone_kstat_delete(zone); 51410209230bSgjelinek 5142134a1f4eSCasper H.S. Dik /* remove the pfexecd doors */ 5143134a1f4eSCasper H.S. Dik if (zone->zone_pfexecd != NULL) { 5144134a1f4eSCasper H.S. Dik klpd_freelist(&zone->zone_pfexecd); 5145134a1f4eSCasper H.S. Dik zone->zone_pfexecd = NULL; 5146134a1f4eSCasper H.S. Dik } 5147134a1f4eSCasper H.S. Dik 5148319378d9Seh /* free brand specific data */ 5149319378d9Seh if (ZONE_IS_BRANDED(zone)) 5150319378d9Seh ZBROP(zone)->b_free_brand_data(zone); 5151319378d9Seh 515252978630Ssl /* Say goodbye to brand framework. */ 515352978630Ssl brand_unregister_zone(zone->zone_brand); 515452978630Ssl 51557c478bd9Sstevel@tonic-gate /* 51567c478bd9Sstevel@tonic-gate * It is now safe to let the zone be recreated; remove it from the 51577c478bd9Sstevel@tonic-gate * lists. The memory will not be freed until the last cred 51587c478bd9Sstevel@tonic-gate * reference goes away. 51597c478bd9Sstevel@tonic-gate */ 51607c478bd9Sstevel@tonic-gate ASSERT(zonecount > 1); /* must be > 1; can't destroy global zone */ 51617c478bd9Sstevel@tonic-gate zonecount--; 51627c478bd9Sstevel@tonic-gate /* remove from active list and hash tables */ 51637c478bd9Sstevel@tonic-gate list_remove(&zone_active, zone); 51647c478bd9Sstevel@tonic-gate (void) mod_hash_destroy(zonehashbyname, 51657c478bd9Sstevel@tonic-gate (mod_hash_key_t)zone->zone_name); 51667c478bd9Sstevel@tonic-gate (void) mod_hash_destroy(zonehashbyid, 51677c478bd9Sstevel@tonic-gate (mod_hash_key_t)(uintptr_t)zone->zone_id); 516848451833Scarlsonj if (zone->zone_flags & ZF_HASHED_LABEL) 516945916cd2Sjpk (void) mod_hash_destroy(zonehashbylabel, 517045916cd2Sjpk (mod_hash_key_t)zone->zone_slabel); 51717c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 51727c478bd9Sstevel@tonic-gate 5173108322fbScarlsonj /* 5174108322fbScarlsonj * Release the root vnode; we're not using it anymore. Nor should any 5175108322fbScarlsonj * other thread that might access it exist. 5176108322fbScarlsonj */ 5177108322fbScarlsonj if (zone->zone_rootvp != NULL) { 5178108322fbScarlsonj VN_RELE(zone->zone_rootvp); 5179108322fbScarlsonj zone->zone_rootvp = NULL; 5180108322fbScarlsonj } 5181108322fbScarlsonj 51827c478bd9Sstevel@tonic-gate /* add to deathrow list */ 51837c478bd9Sstevel@tonic-gate mutex_enter(&zone_deathrow_lock); 51847c478bd9Sstevel@tonic-gate list_insert_tail(&zone_deathrow, zone); 51857c478bd9Sstevel@tonic-gate mutex_exit(&zone_deathrow_lock); 51867c478bd9Sstevel@tonic-gate 51877c478bd9Sstevel@tonic-gate /* 51887c478bd9Sstevel@tonic-gate * Drop last reference (which was added by zsched()), this will 51897c478bd9Sstevel@tonic-gate * free the zone unless there are outstanding cred references. 51907c478bd9Sstevel@tonic-gate */ 51917c478bd9Sstevel@tonic-gate zone_rele(zone); 51927c478bd9Sstevel@tonic-gate return (0); 51937c478bd9Sstevel@tonic-gate } 51947c478bd9Sstevel@tonic-gate 51957c478bd9Sstevel@tonic-gate /* 51967c478bd9Sstevel@tonic-gate * Systemcall entry point for zone_getattr(2). 51977c478bd9Sstevel@tonic-gate */ 51987c478bd9Sstevel@tonic-gate static ssize_t 51997c478bd9Sstevel@tonic-gate zone_getattr(zoneid_t zoneid, int attr, void *buf, size_t bufsize) 52007c478bd9Sstevel@tonic-gate { 52017c478bd9Sstevel@tonic-gate size_t size; 52027c478bd9Sstevel@tonic-gate int error = 0, err; 52037c478bd9Sstevel@tonic-gate zone_t *zone; 52047c478bd9Sstevel@tonic-gate char *zonepath; 52053f2f09c1Sdp char *outstr; 52067c478bd9Sstevel@tonic-gate zone_status_t zone_status; 52077c478bd9Sstevel@tonic-gate pid_t initpid; 5208c97ad5cdSakolb boolean_t global = (curzone == global_zone); 5209c97ad5cdSakolb boolean_t inzone = (curzone->zone_id == zoneid); 5210f4b3ec61Sdh ushort_t flags; 5211550b6e40SSowmini Varadhan zone_net_data_t *zbuf; 52127c478bd9Sstevel@tonic-gate 52137c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 52147c478bd9Sstevel@tonic-gate if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 52157c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 52167c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 52177c478bd9Sstevel@tonic-gate } 52187c478bd9Sstevel@tonic-gate zone_status = zone_status_get(zone); 5219bd41d0a8Snordmark if (zone_status < ZONE_IS_INITIALIZED) { 52207c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 52217c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 52227c478bd9Sstevel@tonic-gate } 52237c478bd9Sstevel@tonic-gate zone_hold(zone); 52247c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 52257c478bd9Sstevel@tonic-gate 52267c478bd9Sstevel@tonic-gate /* 522745916cd2Sjpk * If not in the global zone, don't show information about other zones, 522845916cd2Sjpk * unless the system is labeled and the local zone's label dominates 522945916cd2Sjpk * the other zone. 52307c478bd9Sstevel@tonic-gate */ 523145916cd2Sjpk if (!zone_list_access(zone)) { 52327c478bd9Sstevel@tonic-gate zone_rele(zone); 52337c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 52347c478bd9Sstevel@tonic-gate } 52357c478bd9Sstevel@tonic-gate 52367c478bd9Sstevel@tonic-gate switch (attr) { 52377c478bd9Sstevel@tonic-gate case ZONE_ATTR_ROOT: 52387c478bd9Sstevel@tonic-gate if (global) { 52397c478bd9Sstevel@tonic-gate /* 52407c478bd9Sstevel@tonic-gate * Copy the path to trim the trailing "/" (except for 52417c478bd9Sstevel@tonic-gate * the global zone). 52427c478bd9Sstevel@tonic-gate */ 52437c478bd9Sstevel@tonic-gate if (zone != global_zone) 52447c478bd9Sstevel@tonic-gate size = zone->zone_rootpathlen - 1; 52457c478bd9Sstevel@tonic-gate else 52467c478bd9Sstevel@tonic-gate size = zone->zone_rootpathlen; 52477c478bd9Sstevel@tonic-gate zonepath = kmem_alloc(size, KM_SLEEP); 52487c478bd9Sstevel@tonic-gate bcopy(zone->zone_rootpath, zonepath, size); 52497c478bd9Sstevel@tonic-gate zonepath[size - 1] = '\0'; 52507c478bd9Sstevel@tonic-gate } else { 5251c97ad5cdSakolb if (inzone || !is_system_labeled()) { 525245916cd2Sjpk /* 525345916cd2Sjpk * Caller is not in the global zone. 525445916cd2Sjpk * if the query is on the current zone 525545916cd2Sjpk * or the system is not labeled, 525645916cd2Sjpk * just return faked-up path for current zone. 525745916cd2Sjpk */ 525845916cd2Sjpk zonepath = "/"; 525945916cd2Sjpk size = 2; 526045916cd2Sjpk } else { 526145916cd2Sjpk /* 526245916cd2Sjpk * Return related path for current zone. 526345916cd2Sjpk */ 526445916cd2Sjpk int prefix_len = strlen(zone_prefix); 526545916cd2Sjpk int zname_len = strlen(zone->zone_name); 526645916cd2Sjpk 526745916cd2Sjpk size = prefix_len + zname_len + 1; 526845916cd2Sjpk zonepath = kmem_alloc(size, KM_SLEEP); 526945916cd2Sjpk bcopy(zone_prefix, zonepath, prefix_len); 527045916cd2Sjpk bcopy(zone->zone_name, zonepath + 52713f2f09c1Sdp prefix_len, zname_len); 527245916cd2Sjpk zonepath[size - 1] = '\0'; 527345916cd2Sjpk } 52747c478bd9Sstevel@tonic-gate } 52757c478bd9Sstevel@tonic-gate if (bufsize > size) 52767c478bd9Sstevel@tonic-gate bufsize = size; 52777c478bd9Sstevel@tonic-gate if (buf != NULL) { 52787c478bd9Sstevel@tonic-gate err = copyoutstr(zonepath, buf, bufsize, NULL); 52797c478bd9Sstevel@tonic-gate if (err != 0 && err != ENAMETOOLONG) 52807c478bd9Sstevel@tonic-gate error = EFAULT; 52817c478bd9Sstevel@tonic-gate } 5282c97ad5cdSakolb if (global || (is_system_labeled() && !inzone)) 52837c478bd9Sstevel@tonic-gate kmem_free(zonepath, size); 52847c478bd9Sstevel@tonic-gate break; 52857c478bd9Sstevel@tonic-gate 52867c478bd9Sstevel@tonic-gate case ZONE_ATTR_NAME: 52877c478bd9Sstevel@tonic-gate size = strlen(zone->zone_name) + 1; 52887c478bd9Sstevel@tonic-gate if (bufsize > size) 52897c478bd9Sstevel@tonic-gate bufsize = size; 52907c478bd9Sstevel@tonic-gate if (buf != NULL) { 52917c478bd9Sstevel@tonic-gate err = copyoutstr(zone->zone_name, buf, bufsize, NULL); 52927c478bd9Sstevel@tonic-gate if (err != 0 && err != ENAMETOOLONG) 52937c478bd9Sstevel@tonic-gate error = EFAULT; 52947c478bd9Sstevel@tonic-gate } 52957c478bd9Sstevel@tonic-gate break; 52967c478bd9Sstevel@tonic-gate 52977c478bd9Sstevel@tonic-gate case ZONE_ATTR_STATUS: 52987c478bd9Sstevel@tonic-gate /* 52997c478bd9Sstevel@tonic-gate * Since we're not holding zonehash_lock, the zone status 53007c478bd9Sstevel@tonic-gate * may be anything; leave it up to userland to sort it out. 53017c478bd9Sstevel@tonic-gate */ 53027c478bd9Sstevel@tonic-gate size = sizeof (zone_status); 53037c478bd9Sstevel@tonic-gate if (bufsize > size) 53047c478bd9Sstevel@tonic-gate bufsize = size; 53057c478bd9Sstevel@tonic-gate zone_status = zone_status_get(zone); 53067c478bd9Sstevel@tonic-gate if (buf != NULL && 53077c478bd9Sstevel@tonic-gate copyout(&zone_status, buf, bufsize) != 0) 53087c478bd9Sstevel@tonic-gate error = EFAULT; 53097c478bd9Sstevel@tonic-gate break; 5310f4b3ec61Sdh case ZONE_ATTR_FLAGS: 5311f4b3ec61Sdh size = sizeof (zone->zone_flags); 5312f4b3ec61Sdh if (bufsize > size) 5313f4b3ec61Sdh bufsize = size; 5314f4b3ec61Sdh flags = zone->zone_flags; 5315f4b3ec61Sdh if (buf != NULL && 5316f4b3ec61Sdh copyout(&flags, buf, bufsize) != 0) 5317f4b3ec61Sdh error = EFAULT; 5318f4b3ec61Sdh break; 53197c478bd9Sstevel@tonic-gate case ZONE_ATTR_PRIVSET: 53207c478bd9Sstevel@tonic-gate size = sizeof (priv_set_t); 53217c478bd9Sstevel@tonic-gate if (bufsize > size) 53227c478bd9Sstevel@tonic-gate bufsize = size; 53237c478bd9Sstevel@tonic-gate if (buf != NULL && 53247c478bd9Sstevel@tonic-gate copyout(zone->zone_privset, buf, bufsize) != 0) 53257c478bd9Sstevel@tonic-gate error = EFAULT; 53267c478bd9Sstevel@tonic-gate break; 53277c478bd9Sstevel@tonic-gate case ZONE_ATTR_UNIQID: 53287c478bd9Sstevel@tonic-gate size = sizeof (zone->zone_uniqid); 53297c478bd9Sstevel@tonic-gate if (bufsize > size) 53307c478bd9Sstevel@tonic-gate bufsize = size; 53317c478bd9Sstevel@tonic-gate if (buf != NULL && 53327c478bd9Sstevel@tonic-gate copyout(&zone->zone_uniqid, buf, bufsize) != 0) 53337c478bd9Sstevel@tonic-gate error = EFAULT; 53347c478bd9Sstevel@tonic-gate break; 53357c478bd9Sstevel@tonic-gate case ZONE_ATTR_POOLID: 53367c478bd9Sstevel@tonic-gate { 53377c478bd9Sstevel@tonic-gate pool_t *pool; 53387c478bd9Sstevel@tonic-gate poolid_t poolid; 53397c478bd9Sstevel@tonic-gate 53407c478bd9Sstevel@tonic-gate if (pool_lock_intr() != 0) { 53417c478bd9Sstevel@tonic-gate error = EINTR; 53427c478bd9Sstevel@tonic-gate break; 53437c478bd9Sstevel@tonic-gate } 53447c478bd9Sstevel@tonic-gate pool = zone_pool_get(zone); 53457c478bd9Sstevel@tonic-gate poolid = pool->pool_id; 53467c478bd9Sstevel@tonic-gate pool_unlock(); 53477c478bd9Sstevel@tonic-gate size = sizeof (poolid); 53487c478bd9Sstevel@tonic-gate if (bufsize > size) 53497c478bd9Sstevel@tonic-gate bufsize = size; 53507c478bd9Sstevel@tonic-gate if (buf != NULL && copyout(&poolid, buf, size) != 0) 53517c478bd9Sstevel@tonic-gate error = EFAULT; 53527c478bd9Sstevel@tonic-gate } 53537c478bd9Sstevel@tonic-gate break; 535445916cd2Sjpk case ZONE_ATTR_SLBL: 535545916cd2Sjpk size = sizeof (bslabel_t); 535645916cd2Sjpk if (bufsize > size) 535745916cd2Sjpk bufsize = size; 535845916cd2Sjpk if (zone->zone_slabel == NULL) 535945916cd2Sjpk error = EINVAL; 536045916cd2Sjpk else if (buf != NULL && 536145916cd2Sjpk copyout(label2bslabel(zone->zone_slabel), buf, 536245916cd2Sjpk bufsize) != 0) 536345916cd2Sjpk error = EFAULT; 536445916cd2Sjpk break; 53657c478bd9Sstevel@tonic-gate case ZONE_ATTR_INITPID: 53667c478bd9Sstevel@tonic-gate size = sizeof (initpid); 53677c478bd9Sstevel@tonic-gate if (bufsize > size) 53687c478bd9Sstevel@tonic-gate bufsize = size; 53697c478bd9Sstevel@tonic-gate initpid = zone->zone_proc_initpid; 53707c478bd9Sstevel@tonic-gate if (initpid == -1) { 53717c478bd9Sstevel@tonic-gate error = ESRCH; 53727c478bd9Sstevel@tonic-gate break; 53737c478bd9Sstevel@tonic-gate } 53747c478bd9Sstevel@tonic-gate if (buf != NULL && 53757c478bd9Sstevel@tonic-gate copyout(&initpid, buf, bufsize) != 0) 53767c478bd9Sstevel@tonic-gate error = EFAULT; 53777c478bd9Sstevel@tonic-gate break; 53789acbbeafSnn case ZONE_ATTR_BRAND: 53799acbbeafSnn size = strlen(zone->zone_brand->b_name) + 1; 53809acbbeafSnn 53819acbbeafSnn if (bufsize > size) 53829acbbeafSnn bufsize = size; 53839acbbeafSnn if (buf != NULL) { 53849acbbeafSnn err = copyoutstr(zone->zone_brand->b_name, buf, 53859acbbeafSnn bufsize, NULL); 53869acbbeafSnn if (err != 0 && err != ENAMETOOLONG) 53879acbbeafSnn error = EFAULT; 53889acbbeafSnn } 53899acbbeafSnn break; 53903f2f09c1Sdp case ZONE_ATTR_INITNAME: 53913f2f09c1Sdp size = strlen(zone->zone_initname) + 1; 53923f2f09c1Sdp if (bufsize > size) 53933f2f09c1Sdp bufsize = size; 53943f2f09c1Sdp if (buf != NULL) { 53953f2f09c1Sdp err = copyoutstr(zone->zone_initname, buf, bufsize, 53963f2f09c1Sdp NULL); 53973f2f09c1Sdp if (err != 0 && err != ENAMETOOLONG) 53983f2f09c1Sdp error = EFAULT; 53993f2f09c1Sdp } 54003f2f09c1Sdp break; 54013f2f09c1Sdp case ZONE_ATTR_BOOTARGS: 54023f2f09c1Sdp if (zone->zone_bootargs == NULL) 54033f2f09c1Sdp outstr = ""; 54043f2f09c1Sdp else 54053f2f09c1Sdp outstr = zone->zone_bootargs; 54063f2f09c1Sdp size = strlen(outstr) + 1; 54073f2f09c1Sdp if (bufsize > size) 54083f2f09c1Sdp bufsize = size; 54093f2f09c1Sdp if (buf != NULL) { 54103f2f09c1Sdp err = copyoutstr(outstr, buf, bufsize, NULL); 54113f2f09c1Sdp if (err != 0 && err != ENAMETOOLONG) 54123f2f09c1Sdp error = EFAULT; 54133f2f09c1Sdp } 54143f2f09c1Sdp break; 54150209230bSgjelinek case ZONE_ATTR_PHYS_MCAP: 54160209230bSgjelinek size = sizeof (zone->zone_phys_mcap); 54170209230bSgjelinek if (bufsize > size) 54180209230bSgjelinek bufsize = size; 54190209230bSgjelinek if (buf != NULL && 54200209230bSgjelinek copyout(&zone->zone_phys_mcap, buf, bufsize) != 0) 54210209230bSgjelinek error = EFAULT; 54220209230bSgjelinek break; 54230209230bSgjelinek case ZONE_ATTR_SCHED_CLASS: 54240209230bSgjelinek mutex_enter(&class_lock); 54250209230bSgjelinek 54260209230bSgjelinek if (zone->zone_defaultcid >= loaded_classes) 54270209230bSgjelinek outstr = ""; 54280209230bSgjelinek else 54290209230bSgjelinek outstr = sclass[zone->zone_defaultcid].cl_name; 54300209230bSgjelinek size = strlen(outstr) + 1; 54310209230bSgjelinek if (bufsize > size) 54320209230bSgjelinek bufsize = size; 54330209230bSgjelinek if (buf != NULL) { 54340209230bSgjelinek err = copyoutstr(outstr, buf, bufsize, NULL); 54350209230bSgjelinek if (err != 0 && err != ENAMETOOLONG) 54360209230bSgjelinek error = EFAULT; 54370209230bSgjelinek } 54380209230bSgjelinek 54390209230bSgjelinek mutex_exit(&class_lock); 54400209230bSgjelinek break; 54415679c89fSjv case ZONE_ATTR_HOSTID: 54425679c89fSjv if (zone->zone_hostid != HW_INVALID_HOSTID && 54435679c89fSjv bufsize == sizeof (zone->zone_hostid)) { 54445679c89fSjv size = sizeof (zone->zone_hostid); 54455679c89fSjv if (buf != NULL && copyout(&zone->zone_hostid, buf, 54465679c89fSjv bufsize) != 0) 54475679c89fSjv error = EFAULT; 54485679c89fSjv } else { 54495679c89fSjv error = EINVAL; 54505679c89fSjv } 54515679c89fSjv break; 54520fbb751dSJohn Levon case ZONE_ATTR_FS_ALLOWED: 54530fbb751dSJohn Levon if (zone->zone_fs_allowed == NULL) 54540fbb751dSJohn Levon outstr = ""; 54550fbb751dSJohn Levon else 54560fbb751dSJohn Levon outstr = zone->zone_fs_allowed; 54570fbb751dSJohn Levon size = strlen(outstr) + 1; 54580fbb751dSJohn Levon if (bufsize > size) 54590fbb751dSJohn Levon bufsize = size; 54600fbb751dSJohn Levon if (buf != NULL) { 54610fbb751dSJohn Levon err = copyoutstr(outstr, buf, bufsize, NULL); 54620fbb751dSJohn Levon if (err != 0 && err != ENAMETOOLONG) 54630fbb751dSJohn Levon error = EFAULT; 54640fbb751dSJohn Levon } 54650fbb751dSJohn Levon break; 5466550b6e40SSowmini Varadhan case ZONE_ATTR_NETWORK: 5467550b6e40SSowmini Varadhan zbuf = kmem_alloc(bufsize, KM_SLEEP); 5468550b6e40SSowmini Varadhan if (copyin(buf, zbuf, bufsize) != 0) { 5469550b6e40SSowmini Varadhan error = EFAULT; 5470550b6e40SSowmini Varadhan } else { 5471550b6e40SSowmini Varadhan error = zone_get_network(zoneid, zbuf); 5472550b6e40SSowmini Varadhan if (error == 0 && copyout(zbuf, buf, bufsize) != 0) 5473550b6e40SSowmini Varadhan error = EFAULT; 5474550b6e40SSowmini Varadhan } 5475550b6e40SSowmini Varadhan kmem_free(zbuf, bufsize); 5476550b6e40SSowmini Varadhan break; 54777c478bd9Sstevel@tonic-gate default: 54789acbbeafSnn if ((attr >= ZONE_ATTR_BRAND_ATTRS) && ZONE_IS_BRANDED(zone)) { 54799acbbeafSnn size = bufsize; 54809acbbeafSnn error = ZBROP(zone)->b_getattr(zone, attr, buf, &size); 54819acbbeafSnn } else { 54829acbbeafSnn error = EINVAL; 54839acbbeafSnn } 54847c478bd9Sstevel@tonic-gate } 54857c478bd9Sstevel@tonic-gate zone_rele(zone); 54867c478bd9Sstevel@tonic-gate 54877c478bd9Sstevel@tonic-gate if (error) 54887c478bd9Sstevel@tonic-gate return (set_errno(error)); 54897c478bd9Sstevel@tonic-gate return ((ssize_t)size); 54907c478bd9Sstevel@tonic-gate } 54917c478bd9Sstevel@tonic-gate 54923f2f09c1Sdp /* 54933f2f09c1Sdp * Systemcall entry point for zone_setattr(2). 54943f2f09c1Sdp */ 54953f2f09c1Sdp /*ARGSUSED*/ 54963f2f09c1Sdp static int 54973f2f09c1Sdp zone_setattr(zoneid_t zoneid, int attr, void *buf, size_t bufsize) 54983f2f09c1Sdp { 54993f2f09c1Sdp zone_t *zone; 55003f2f09c1Sdp zone_status_t zone_status; 55011a88ce5cSDan Price int err = -1; 5502550b6e40SSowmini Varadhan zone_net_data_t *zbuf; 55033f2f09c1Sdp 55043f2f09c1Sdp if (secpolicy_zone_config(CRED()) != 0) 55053f2f09c1Sdp return (set_errno(EPERM)); 55063f2f09c1Sdp 55073f2f09c1Sdp /* 55080209230bSgjelinek * Only the ZONE_ATTR_PHYS_MCAP attribute can be set on the 55090209230bSgjelinek * global zone. 55103f2f09c1Sdp */ 55110209230bSgjelinek if (zoneid == GLOBAL_ZONEID && attr != ZONE_ATTR_PHYS_MCAP) { 55123f2f09c1Sdp return (set_errno(EINVAL)); 55133f2f09c1Sdp } 55143f2f09c1Sdp 55153f2f09c1Sdp mutex_enter(&zonehash_lock); 55163f2f09c1Sdp if ((zone = zone_find_all_by_id(zoneid)) == NULL) { 55173f2f09c1Sdp mutex_exit(&zonehash_lock); 55183f2f09c1Sdp return (set_errno(EINVAL)); 55193f2f09c1Sdp } 55203f2f09c1Sdp zone_hold(zone); 55213f2f09c1Sdp mutex_exit(&zonehash_lock); 55223f2f09c1Sdp 55230209230bSgjelinek /* 55240209230bSgjelinek * At present most attributes can only be set on non-running, 55250209230bSgjelinek * non-global zones. 55260209230bSgjelinek */ 55273f2f09c1Sdp zone_status = zone_status_get(zone); 55281a88ce5cSDan Price if (attr != ZONE_ATTR_PHYS_MCAP && zone_status > ZONE_IS_READY) { 55291a88ce5cSDan Price err = EINVAL; 55303f2f09c1Sdp goto done; 55311a88ce5cSDan Price } 55323f2f09c1Sdp 55333f2f09c1Sdp switch (attr) { 55343f2f09c1Sdp case ZONE_ATTR_INITNAME: 55353f2f09c1Sdp err = zone_set_initname(zone, (const char *)buf); 55363f2f09c1Sdp break; 5537bafd1f14SJerry Jelinek case ZONE_ATTR_INITNORESTART: 5538bafd1f14SJerry Jelinek zone->zone_restart_init = B_FALSE; 5539bafd1f14SJerry Jelinek err = 0; 5540bafd1f14SJerry Jelinek break; 55413f2f09c1Sdp case ZONE_ATTR_BOOTARGS: 55423f2f09c1Sdp err = zone_set_bootargs(zone, (const char *)buf); 55433f2f09c1Sdp break; 55449acbbeafSnn case ZONE_ATTR_BRAND: 554559f2ff5cSedp err = zone_set_brand(zone, (const char *)buf); 55469acbbeafSnn break; 55470fbb751dSJohn Levon case ZONE_ATTR_FS_ALLOWED: 55480fbb751dSJohn Levon err = zone_set_fs_allowed(zone, (const char *)buf); 55490fbb751dSJohn Levon break; 55500209230bSgjelinek case ZONE_ATTR_PHYS_MCAP: 55510209230bSgjelinek err = zone_set_phys_mcap(zone, (const uint64_t *)buf); 55520209230bSgjelinek break; 55530209230bSgjelinek case ZONE_ATTR_SCHED_CLASS: 55540209230bSgjelinek err = zone_set_sched_class(zone, (const char *)buf); 55550209230bSgjelinek break; 55565679c89fSjv case ZONE_ATTR_HOSTID: 55575679c89fSjv if (bufsize == sizeof (zone->zone_hostid)) { 55585679c89fSjv if (copyin(buf, &zone->zone_hostid, bufsize) == 0) 55595679c89fSjv err = 0; 55605679c89fSjv else 55615679c89fSjv err = EFAULT; 55625679c89fSjv } else { 55635679c89fSjv err = EINVAL; 55645679c89fSjv } 55655679c89fSjv break; 5566550b6e40SSowmini Varadhan case ZONE_ATTR_NETWORK: 5567550b6e40SSowmini Varadhan if (bufsize > (PIPE_BUF + sizeof (zone_net_data_t))) { 5568550b6e40SSowmini Varadhan err = EINVAL; 55691a88ce5cSDan Price break; 5570550b6e40SSowmini Varadhan } 5571550b6e40SSowmini Varadhan zbuf = kmem_alloc(bufsize, KM_SLEEP); 5572550b6e40SSowmini Varadhan if (copyin(buf, zbuf, bufsize) != 0) { 55731a88ce5cSDan Price kmem_free(zbuf, bufsize); 5574550b6e40SSowmini Varadhan err = EFAULT; 55751a88ce5cSDan Price break; 5576550b6e40SSowmini Varadhan } 5577550b6e40SSowmini Varadhan err = zone_set_network(zoneid, zbuf); 5578550b6e40SSowmini Varadhan kmem_free(zbuf, bufsize); 5579550b6e40SSowmini Varadhan break; 55803f2f09c1Sdp default: 55819acbbeafSnn if ((attr >= ZONE_ATTR_BRAND_ATTRS) && ZONE_IS_BRANDED(zone)) 55829acbbeafSnn err = ZBROP(zone)->b_setattr(zone, attr, buf, bufsize); 55839acbbeafSnn else 55849acbbeafSnn err = EINVAL; 55853f2f09c1Sdp } 55863f2f09c1Sdp 55873f2f09c1Sdp done: 55883f2f09c1Sdp zone_rele(zone); 55891a88ce5cSDan Price ASSERT(err != -1); 55903f2f09c1Sdp return (err != 0 ? set_errno(err) : 0); 55913f2f09c1Sdp } 55923f2f09c1Sdp 55937c478bd9Sstevel@tonic-gate /* 55947c478bd9Sstevel@tonic-gate * Return zero if the process has at least one vnode mapped in to its 55957c478bd9Sstevel@tonic-gate * address space which shouldn't be allowed to change zones. 55960209230bSgjelinek * 55970209230bSgjelinek * Also return zero if the process has any shared mappings which reserve 55980209230bSgjelinek * swap. This is because the counting for zone.max-swap does not allow swap 5599da6c28aaSamw * reservation to be shared between zones. zone swap reservation is counted 56000209230bSgjelinek * on zone->zone_max_swap. 56017c478bd9Sstevel@tonic-gate */ 56027c478bd9Sstevel@tonic-gate static int 56037c478bd9Sstevel@tonic-gate as_can_change_zones(void) 56047c478bd9Sstevel@tonic-gate { 56057c478bd9Sstevel@tonic-gate proc_t *pp = curproc; 56067c478bd9Sstevel@tonic-gate struct seg *seg; 56077c478bd9Sstevel@tonic-gate struct as *as = pp->p_as; 56087c478bd9Sstevel@tonic-gate vnode_t *vp; 56097c478bd9Sstevel@tonic-gate int allow = 1; 56107c478bd9Sstevel@tonic-gate 56117c478bd9Sstevel@tonic-gate ASSERT(pp->p_as != &kas); 56120209230bSgjelinek AS_LOCK_ENTER(as, &as->a_lock, RW_READER); 56137c478bd9Sstevel@tonic-gate for (seg = AS_SEGFIRST(as); seg != NULL; seg = AS_SEGNEXT(as, seg)) { 56140209230bSgjelinek 56150209230bSgjelinek /* 56160209230bSgjelinek * Cannot enter zone with shared anon memory which 56170209230bSgjelinek * reserves swap. See comment above. 56180209230bSgjelinek */ 56190209230bSgjelinek if (seg_can_change_zones(seg) == B_FALSE) { 56200209230bSgjelinek allow = 0; 56210209230bSgjelinek break; 56220209230bSgjelinek } 56237c478bd9Sstevel@tonic-gate /* 56247c478bd9Sstevel@tonic-gate * if we can't get a backing vnode for this segment then skip 56257c478bd9Sstevel@tonic-gate * it. 56267c478bd9Sstevel@tonic-gate */ 56277c478bd9Sstevel@tonic-gate vp = NULL; 56287c478bd9Sstevel@tonic-gate if (SEGOP_GETVP(seg, seg->s_base, &vp) != 0 || vp == NULL) 56297c478bd9Sstevel@tonic-gate continue; 56307c478bd9Sstevel@tonic-gate if (!vn_can_change_zones(vp)) { /* bail on first match */ 56317c478bd9Sstevel@tonic-gate allow = 0; 56327c478bd9Sstevel@tonic-gate break; 56337c478bd9Sstevel@tonic-gate } 56347c478bd9Sstevel@tonic-gate } 56350209230bSgjelinek AS_LOCK_EXIT(as, &as->a_lock); 56367c478bd9Sstevel@tonic-gate return (allow); 56377c478bd9Sstevel@tonic-gate } 56387c478bd9Sstevel@tonic-gate 56390209230bSgjelinek /* 56400209230bSgjelinek * Count swap reserved by curproc's address space 56410209230bSgjelinek */ 56420209230bSgjelinek static size_t 56430209230bSgjelinek as_swresv(void) 56440209230bSgjelinek { 56450209230bSgjelinek proc_t *pp = curproc; 56460209230bSgjelinek struct seg *seg; 56470209230bSgjelinek struct as *as = pp->p_as; 56480209230bSgjelinek size_t swap = 0; 56490209230bSgjelinek 56500209230bSgjelinek ASSERT(pp->p_as != &kas); 56510209230bSgjelinek ASSERT(AS_WRITE_HELD(as, &as->a_lock)); 56520209230bSgjelinek for (seg = AS_SEGFIRST(as); seg != NULL; seg = AS_SEGNEXT(as, seg)) 56530209230bSgjelinek swap += seg_swresv(seg); 56540209230bSgjelinek 56550209230bSgjelinek return (swap); 56560209230bSgjelinek } 56570209230bSgjelinek 56587c478bd9Sstevel@tonic-gate /* 56597c478bd9Sstevel@tonic-gate * Systemcall entry point for zone_enter(). 56607c478bd9Sstevel@tonic-gate * 56617c478bd9Sstevel@tonic-gate * The current process is injected into said zone. In the process 56627c478bd9Sstevel@tonic-gate * it will change its project membership, privileges, rootdir/cwd, 56637c478bd9Sstevel@tonic-gate * zone-wide rctls, and pool association to match those of the zone. 56647c478bd9Sstevel@tonic-gate * 56657c478bd9Sstevel@tonic-gate * The first zone_enter() called while the zone is in the ZONE_IS_READY 56667c478bd9Sstevel@tonic-gate * state will transition it to ZONE_IS_RUNNING. Processes may only 56677c478bd9Sstevel@tonic-gate * enter a zone that is "ready" or "running". 56687c478bd9Sstevel@tonic-gate */ 56697c478bd9Sstevel@tonic-gate static int 56707c478bd9Sstevel@tonic-gate zone_enter(zoneid_t zoneid) 56717c478bd9Sstevel@tonic-gate { 56727c478bd9Sstevel@tonic-gate zone_t *zone; 56737c478bd9Sstevel@tonic-gate vnode_t *vp; 56747c478bd9Sstevel@tonic-gate proc_t *pp = curproc; 56757c478bd9Sstevel@tonic-gate contract_t *ct; 56767c478bd9Sstevel@tonic-gate cont_process_t *ctp; 56777c478bd9Sstevel@tonic-gate task_t *tk, *oldtk; 56787c478bd9Sstevel@tonic-gate kproject_t *zone_proj0; 56797c478bd9Sstevel@tonic-gate cred_t *cr, *newcr; 56807c478bd9Sstevel@tonic-gate pool_t *oldpool, *newpool; 56817c478bd9Sstevel@tonic-gate sess_t *sp; 56827c478bd9Sstevel@tonic-gate uid_t uid; 56837c478bd9Sstevel@tonic-gate zone_status_t status; 56847c478bd9Sstevel@tonic-gate int err = 0; 56857c478bd9Sstevel@tonic-gate rctl_entity_p_t e; 56860209230bSgjelinek size_t swap; 5687c97ad5cdSakolb kthread_id_t t; 56887c478bd9Sstevel@tonic-gate 56897c478bd9Sstevel@tonic-gate if (secpolicy_zone_config(CRED()) != 0) 56907c478bd9Sstevel@tonic-gate return (set_errno(EPERM)); 56917c478bd9Sstevel@tonic-gate if (zoneid < MIN_USERZONEID || zoneid > MAX_ZONEID) 56927c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 56937c478bd9Sstevel@tonic-gate 56947c478bd9Sstevel@tonic-gate /* 56957c478bd9Sstevel@tonic-gate * Stop all lwps so we don't need to hold a lock to look at 56967c478bd9Sstevel@tonic-gate * curproc->p_zone. This needs to happen before we grab any 56977c478bd9Sstevel@tonic-gate * locks to avoid deadlock (another lwp in the process could 56987c478bd9Sstevel@tonic-gate * be waiting for the held lock). 56997c478bd9Sstevel@tonic-gate */ 57007c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp && !holdlwps(SHOLDFORK)) 57017c478bd9Sstevel@tonic-gate return (set_errno(EINTR)); 57027c478bd9Sstevel@tonic-gate 57037c478bd9Sstevel@tonic-gate /* 57047c478bd9Sstevel@tonic-gate * Make sure we're not changing zones with files open or mapped in 57057c478bd9Sstevel@tonic-gate * to our address space which shouldn't be changing zones. 57067c478bd9Sstevel@tonic-gate */ 57077c478bd9Sstevel@tonic-gate if (!files_can_change_zones()) { 57087c478bd9Sstevel@tonic-gate err = EBADF; 57097c478bd9Sstevel@tonic-gate goto out; 57107c478bd9Sstevel@tonic-gate } 57117c478bd9Sstevel@tonic-gate if (!as_can_change_zones()) { 57127c478bd9Sstevel@tonic-gate err = EFAULT; 57137c478bd9Sstevel@tonic-gate goto out; 57147c478bd9Sstevel@tonic-gate } 57157c478bd9Sstevel@tonic-gate 57167c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 57177c478bd9Sstevel@tonic-gate if (pp->p_zone != global_zone) { 57187c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57197c478bd9Sstevel@tonic-gate err = EINVAL; 57207c478bd9Sstevel@tonic-gate goto out; 57217c478bd9Sstevel@tonic-gate } 57227c478bd9Sstevel@tonic-gate 57237c478bd9Sstevel@tonic-gate zone = zone_find_all_by_id(zoneid); 57247c478bd9Sstevel@tonic-gate if (zone == NULL) { 57257c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57267c478bd9Sstevel@tonic-gate err = EINVAL; 57277c478bd9Sstevel@tonic-gate goto out; 57287c478bd9Sstevel@tonic-gate } 57297c478bd9Sstevel@tonic-gate 57307c478bd9Sstevel@tonic-gate /* 57317c478bd9Sstevel@tonic-gate * To prevent processes in a zone from holding contracts on 57327c478bd9Sstevel@tonic-gate * extrazonal resources, and to avoid process contract 57337c478bd9Sstevel@tonic-gate * memberships which span zones, contract holders and processes 57347c478bd9Sstevel@tonic-gate * which aren't the sole members of their encapsulating process 57357c478bd9Sstevel@tonic-gate * contracts are not allowed to zone_enter. 57367c478bd9Sstevel@tonic-gate */ 57377c478bd9Sstevel@tonic-gate ctp = pp->p_ct_process; 57387c478bd9Sstevel@tonic-gate ct = &ctp->conp_contract; 57397c478bd9Sstevel@tonic-gate mutex_enter(&ct->ct_lock); 57407c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 57417c478bd9Sstevel@tonic-gate if ((avl_numnodes(&pp->p_ct_held) != 0) || (ctp->conp_nmembers != 1)) { 57427c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 57437c478bd9Sstevel@tonic-gate mutex_exit(&ct->ct_lock); 57447c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57457c478bd9Sstevel@tonic-gate err = EINVAL; 57467c478bd9Sstevel@tonic-gate goto out; 57477c478bd9Sstevel@tonic-gate } 57487c478bd9Sstevel@tonic-gate 57497c478bd9Sstevel@tonic-gate /* 57507c478bd9Sstevel@tonic-gate * Moreover, we don't allow processes whose encapsulating 57517c478bd9Sstevel@tonic-gate * process contracts have inherited extrazonal contracts. 57527c478bd9Sstevel@tonic-gate * While it would be easier to eliminate all process contracts 57537c478bd9Sstevel@tonic-gate * with inherited contracts, we need to be able to give a 57547c478bd9Sstevel@tonic-gate * restarted init (or other zone-penetrating process) its 57557c478bd9Sstevel@tonic-gate * predecessor's contracts. 57567c478bd9Sstevel@tonic-gate */ 57577c478bd9Sstevel@tonic-gate if (ctp->conp_ninherited != 0) { 57587c478bd9Sstevel@tonic-gate contract_t *next; 57597c478bd9Sstevel@tonic-gate for (next = list_head(&ctp->conp_inherited); next; 57607c478bd9Sstevel@tonic-gate next = list_next(&ctp->conp_inherited, next)) { 57617c478bd9Sstevel@tonic-gate if (contract_getzuniqid(next) != zone->zone_uniqid) { 57627c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 57637c478bd9Sstevel@tonic-gate mutex_exit(&ct->ct_lock); 57647c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57657c478bd9Sstevel@tonic-gate err = EINVAL; 57667c478bd9Sstevel@tonic-gate goto out; 57677c478bd9Sstevel@tonic-gate } 57687c478bd9Sstevel@tonic-gate } 57697c478bd9Sstevel@tonic-gate } 57707b209c2cSacruz 57717c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 57727c478bd9Sstevel@tonic-gate mutex_exit(&ct->ct_lock); 57737c478bd9Sstevel@tonic-gate 57747c478bd9Sstevel@tonic-gate status = zone_status_get(zone); 57757c478bd9Sstevel@tonic-gate if (status < ZONE_IS_READY || status >= ZONE_IS_SHUTTING_DOWN) { 57767c478bd9Sstevel@tonic-gate /* 57777c478bd9Sstevel@tonic-gate * Can't join 57787c478bd9Sstevel@tonic-gate */ 57797c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57807c478bd9Sstevel@tonic-gate err = EINVAL; 57817c478bd9Sstevel@tonic-gate goto out; 57827c478bd9Sstevel@tonic-gate } 57837c478bd9Sstevel@tonic-gate 57847c478bd9Sstevel@tonic-gate /* 57857c478bd9Sstevel@tonic-gate * Make sure new priv set is within the permitted set for caller 57867c478bd9Sstevel@tonic-gate */ 57877c478bd9Sstevel@tonic-gate if (!priv_issubset(zone->zone_privset, &CR_OPPRIV(CRED()))) { 57887c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57897c478bd9Sstevel@tonic-gate err = EPERM; 57907c478bd9Sstevel@tonic-gate goto out; 57917c478bd9Sstevel@tonic-gate } 57927c478bd9Sstevel@tonic-gate /* 57937c478bd9Sstevel@tonic-gate * We want to momentarily drop zonehash_lock while we optimistically 57947c478bd9Sstevel@tonic-gate * bind curproc to the pool it should be running in. This is safe 57957c478bd9Sstevel@tonic-gate * since the zone can't disappear (we have a hold on it). 57967c478bd9Sstevel@tonic-gate */ 57977c478bd9Sstevel@tonic-gate zone_hold(zone); 57987c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 57997c478bd9Sstevel@tonic-gate 58007c478bd9Sstevel@tonic-gate /* 58017c478bd9Sstevel@tonic-gate * Grab pool_lock to keep the pools configuration from changing 58027c478bd9Sstevel@tonic-gate * and to stop ourselves from getting rebound to another pool 58037c478bd9Sstevel@tonic-gate * until we join the zone. 58047c478bd9Sstevel@tonic-gate */ 58057c478bd9Sstevel@tonic-gate if (pool_lock_intr() != 0) { 58067c478bd9Sstevel@tonic-gate zone_rele(zone); 58077c478bd9Sstevel@tonic-gate err = EINTR; 58087c478bd9Sstevel@tonic-gate goto out; 58097c478bd9Sstevel@tonic-gate } 58107c478bd9Sstevel@tonic-gate ASSERT(secpolicy_pool(CRED()) == 0); 58117c478bd9Sstevel@tonic-gate /* 58127c478bd9Sstevel@tonic-gate * Bind ourselves to the pool currently associated with the zone. 58137c478bd9Sstevel@tonic-gate */ 58147c478bd9Sstevel@tonic-gate oldpool = curproc->p_pool; 58157c478bd9Sstevel@tonic-gate newpool = zone_pool_get(zone); 58167c478bd9Sstevel@tonic-gate if (pool_state == POOL_ENABLED && newpool != oldpool && 58177c478bd9Sstevel@tonic-gate (err = pool_do_bind(newpool, P_PID, P_MYID, 58187c478bd9Sstevel@tonic-gate POOL_BIND_ALL)) != 0) { 58197c478bd9Sstevel@tonic-gate pool_unlock(); 58207c478bd9Sstevel@tonic-gate zone_rele(zone); 58217c478bd9Sstevel@tonic-gate goto out; 58227c478bd9Sstevel@tonic-gate } 58237c478bd9Sstevel@tonic-gate 58247c478bd9Sstevel@tonic-gate /* 58257c478bd9Sstevel@tonic-gate * Grab cpu_lock now; we'll need it later when we call 58267c478bd9Sstevel@tonic-gate * task_join(). 58277c478bd9Sstevel@tonic-gate */ 58287c478bd9Sstevel@tonic-gate mutex_enter(&cpu_lock); 58297c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 58307c478bd9Sstevel@tonic-gate /* 58317c478bd9Sstevel@tonic-gate * Make sure the zone hasn't moved on since we dropped zonehash_lock. 58327c478bd9Sstevel@tonic-gate */ 58337c478bd9Sstevel@tonic-gate if (zone_status_get(zone) >= ZONE_IS_SHUTTING_DOWN) { 58347c478bd9Sstevel@tonic-gate /* 58357c478bd9Sstevel@tonic-gate * Can't join anymore. 58367c478bd9Sstevel@tonic-gate */ 58377c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 58387c478bd9Sstevel@tonic-gate mutex_exit(&cpu_lock); 58397c478bd9Sstevel@tonic-gate if (pool_state == POOL_ENABLED && 58407c478bd9Sstevel@tonic-gate newpool != oldpool) 58417c478bd9Sstevel@tonic-gate (void) pool_do_bind(oldpool, P_PID, P_MYID, 58427c478bd9Sstevel@tonic-gate POOL_BIND_ALL); 58437c478bd9Sstevel@tonic-gate pool_unlock(); 58447c478bd9Sstevel@tonic-gate zone_rele(zone); 58457c478bd9Sstevel@tonic-gate err = EINVAL; 58467c478bd9Sstevel@tonic-gate goto out; 58477c478bd9Sstevel@tonic-gate } 58487c478bd9Sstevel@tonic-gate 58490209230bSgjelinek /* 58500209230bSgjelinek * a_lock must be held while transfering locked memory and swap 58510209230bSgjelinek * reservation from the global zone to the non global zone because 58520209230bSgjelinek * asynchronous faults on the processes' address space can lock 58530209230bSgjelinek * memory and reserve swap via MCL_FUTURE and MAP_NORESERVE 58540209230bSgjelinek * segments respectively. 58550209230bSgjelinek */ 58560209230bSgjelinek AS_LOCK_ENTER(pp->as, &pp->p_as->a_lock, RW_WRITER); 58570209230bSgjelinek swap = as_swresv(); 58587c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 58597c478bd9Sstevel@tonic-gate zone_proj0 = zone->zone_zsched->p_task->tk_proj; 58607c478bd9Sstevel@tonic-gate /* verify that we do not exceed and task or lwp limits */ 58617c478bd9Sstevel@tonic-gate mutex_enter(&zone->zone_nlwps_lock); 58627c478bd9Sstevel@tonic-gate /* add new lwps to zone and zone's proj0 */ 58637c478bd9Sstevel@tonic-gate zone_proj0->kpj_nlwps += pp->p_lwpcnt; 58647c478bd9Sstevel@tonic-gate zone->zone_nlwps += pp->p_lwpcnt; 58657c478bd9Sstevel@tonic-gate /* add 1 task to zone's proj0 */ 58667c478bd9Sstevel@tonic-gate zone_proj0->kpj_ntasks += 1; 5867ff19e029SMenno Lageman 5868ff19e029SMenno Lageman zone_proj0->kpj_nprocs++; 5869ff19e029SMenno Lageman zone->zone_nprocs++; 58707c478bd9Sstevel@tonic-gate mutex_exit(&zone->zone_nlwps_lock); 58717c478bd9Sstevel@tonic-gate 58720209230bSgjelinek mutex_enter(&zone->zone_mem_lock); 5873c6939658Ssl zone->zone_locked_mem += pp->p_locked_mem; 5874c6939658Ssl zone_proj0->kpj_data.kpd_locked_mem += pp->p_locked_mem; 58750209230bSgjelinek zone->zone_max_swap += swap; 58760209230bSgjelinek mutex_exit(&zone->zone_mem_lock); 5877c6939658Ssl 5878c1a9a9c3Skrishna mutex_enter(&(zone_proj0->kpj_data.kpd_crypto_lock)); 5879c1a9a9c3Skrishna zone_proj0->kpj_data.kpd_crypto_mem += pp->p_crypto_mem; 5880c1a9a9c3Skrishna mutex_exit(&(zone_proj0->kpj_data.kpd_crypto_lock)); 5881c1a9a9c3Skrishna 5882ff19e029SMenno Lageman /* remove lwps and process from proc's old zone and old project */ 58837c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_zone->zone_nlwps_lock); 58847c478bd9Sstevel@tonic-gate pp->p_zone->zone_nlwps -= pp->p_lwpcnt; 58857c478bd9Sstevel@tonic-gate pp->p_task->tk_proj->kpj_nlwps -= pp->p_lwpcnt; 5886ff19e029SMenno Lageman pp->p_task->tk_proj->kpj_nprocs--; 5887ff19e029SMenno Lageman pp->p_zone->zone_nprocs--; 58887c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_zone->zone_nlwps_lock); 58897c478bd9Sstevel@tonic-gate 58900209230bSgjelinek mutex_enter(&pp->p_zone->zone_mem_lock); 5891c6939658Ssl pp->p_zone->zone_locked_mem -= pp->p_locked_mem; 5892c6939658Ssl pp->p_task->tk_proj->kpj_data.kpd_locked_mem -= pp->p_locked_mem; 58930209230bSgjelinek pp->p_zone->zone_max_swap -= swap; 58940209230bSgjelinek mutex_exit(&pp->p_zone->zone_mem_lock); 5895c6939658Ssl 5896c1a9a9c3Skrishna mutex_enter(&(pp->p_task->tk_proj->kpj_data.kpd_crypto_lock)); 5897c1a9a9c3Skrishna pp->p_task->tk_proj->kpj_data.kpd_crypto_mem -= pp->p_crypto_mem; 5898c1a9a9c3Skrishna mutex_exit(&(pp->p_task->tk_proj->kpj_data.kpd_crypto_lock)); 5899c1a9a9c3Skrishna 5900bb5ca623SVamsi Nagineni pp->p_flag |= SZONETOP; 5901bb5ca623SVamsi Nagineni pp->p_zone = zone; 5902c6939658Ssl mutex_exit(&pp->p_lock); 59030209230bSgjelinek AS_LOCK_EXIT(pp->p_as, &pp->p_as->a_lock); 5904c6939658Ssl 59057c478bd9Sstevel@tonic-gate /* 59067c478bd9Sstevel@tonic-gate * Joining the zone cannot fail from now on. 59077c478bd9Sstevel@tonic-gate * 59087c478bd9Sstevel@tonic-gate * This means that a lot of the following code can be commonized and 59097c478bd9Sstevel@tonic-gate * shared with zsched(). 59107c478bd9Sstevel@tonic-gate */ 59117c478bd9Sstevel@tonic-gate 59127b209c2cSacruz /* 59137b209c2cSacruz * If the process contract fmri was inherited, we need to 59147b209c2cSacruz * flag this so that any contract status will not leak 59157b209c2cSacruz * extra zone information, svc_fmri in this case 59167b209c2cSacruz */ 59177b209c2cSacruz if (ctp->conp_svc_ctid != ct->ct_id) { 59187b209c2cSacruz mutex_enter(&ct->ct_lock); 59197b209c2cSacruz ctp->conp_svc_zone_enter = ct->ct_id; 59207b209c2cSacruz mutex_exit(&ct->ct_lock); 59217b209c2cSacruz } 59227b209c2cSacruz 59237c478bd9Sstevel@tonic-gate /* 59247c478bd9Sstevel@tonic-gate * Reset the encapsulating process contract's zone. 59257c478bd9Sstevel@tonic-gate */ 59267c478bd9Sstevel@tonic-gate ASSERT(ct->ct_mzuniqid == GLOBAL_ZONEUNIQID); 59277c478bd9Sstevel@tonic-gate contract_setzuniqid(ct, zone->zone_uniqid); 59287c478bd9Sstevel@tonic-gate 59297c478bd9Sstevel@tonic-gate /* 59307c478bd9Sstevel@tonic-gate * Create a new task and associate the process with the project keyed 59317c478bd9Sstevel@tonic-gate * by (projid,zoneid). 59327c478bd9Sstevel@tonic-gate * 59337c478bd9Sstevel@tonic-gate * We might as well be in project 0; the global zone's projid doesn't 59347c478bd9Sstevel@tonic-gate * make much sense in a zone anyhow. 59357c478bd9Sstevel@tonic-gate * 59367c478bd9Sstevel@tonic-gate * This also increments zone_ntasks, and returns with p_lock held. 59377c478bd9Sstevel@tonic-gate */ 59387c478bd9Sstevel@tonic-gate tk = task_create(0, zone); 59397c478bd9Sstevel@tonic-gate oldtk = task_join(tk, 0); 59407c478bd9Sstevel@tonic-gate mutex_exit(&cpu_lock); 59417c478bd9Sstevel@tonic-gate 59427c478bd9Sstevel@tonic-gate /* 59437c478bd9Sstevel@tonic-gate * call RCTLOP_SET functions on this proc 59447c478bd9Sstevel@tonic-gate */ 59457c478bd9Sstevel@tonic-gate e.rcep_p.zone = zone; 59467c478bd9Sstevel@tonic-gate e.rcep_t = RCENTITY_ZONE; 59477c478bd9Sstevel@tonic-gate (void) rctl_set_dup(NULL, NULL, pp, &e, zone->zone_rctls, NULL, 59487c478bd9Sstevel@tonic-gate RCD_CALLBACK); 59497c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 59507c478bd9Sstevel@tonic-gate 59517c478bd9Sstevel@tonic-gate /* 59527c478bd9Sstevel@tonic-gate * We don't need to hold any of zsched's locks here; not only do we know 59537c478bd9Sstevel@tonic-gate * the process and zone aren't going away, we know its session isn't 59547c478bd9Sstevel@tonic-gate * changing either. 59557c478bd9Sstevel@tonic-gate * 59567c478bd9Sstevel@tonic-gate * By joining zsched's session here, we mimic the behavior in the 59577c478bd9Sstevel@tonic-gate * global zone of init's sid being the pid of sched. We extend this 59587c478bd9Sstevel@tonic-gate * to all zlogin-like zone_enter()'ing processes as well. 59597c478bd9Sstevel@tonic-gate */ 59607c478bd9Sstevel@tonic-gate mutex_enter(&pidlock); 59617c478bd9Sstevel@tonic-gate sp = zone->zone_zsched->p_sessp; 59629acbbeafSnn sess_hold(zone->zone_zsched); 59637c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 59647c478bd9Sstevel@tonic-gate pgexit(pp); 59659acbbeafSnn sess_rele(pp->p_sessp, B_TRUE); 59667c478bd9Sstevel@tonic-gate pp->p_sessp = sp; 59677c478bd9Sstevel@tonic-gate pgjoin(pp, zone->zone_zsched->p_pidp); 59680209230bSgjelinek 5969c97ad5cdSakolb /* 5970c97ad5cdSakolb * If any threads are scheduled to be placed on zone wait queue they 5971c97ad5cdSakolb * should abandon the idea since the wait queue is changing. 5972c97ad5cdSakolb * We need to be holding pidlock & p_lock to do this. 5973c97ad5cdSakolb */ 5974c97ad5cdSakolb if ((t = pp->p_tlist) != NULL) { 5975c97ad5cdSakolb do { 5976c97ad5cdSakolb thread_lock(t); 5977c97ad5cdSakolb /* 5978c97ad5cdSakolb * Kick this thread so that he doesn't sit 5979c97ad5cdSakolb * on a wrong wait queue. 5980c97ad5cdSakolb */ 5981c97ad5cdSakolb if (ISWAITING(t)) 5982c97ad5cdSakolb setrun_locked(t); 5983c97ad5cdSakolb 5984c97ad5cdSakolb if (t->t_schedflag & TS_ANYWAITQ) 5985c97ad5cdSakolb t->t_schedflag &= ~ TS_ANYWAITQ; 5986c97ad5cdSakolb 5987c97ad5cdSakolb thread_unlock(t); 5988c97ad5cdSakolb } while ((t = t->t_forw) != pp->p_tlist); 5989c97ad5cdSakolb } 5990c97ad5cdSakolb 59910209230bSgjelinek /* 59920209230bSgjelinek * If there is a default scheduling class for the zone and it is not 59930209230bSgjelinek * the class we are currently in, change all of the threads in the 59940209230bSgjelinek * process to the new class. We need to be holding pidlock & p_lock 59950209230bSgjelinek * when we call parmsset so this is a good place to do it. 59960209230bSgjelinek */ 59970209230bSgjelinek if (zone->zone_defaultcid > 0 && 59980209230bSgjelinek zone->zone_defaultcid != curthread->t_cid) { 59990209230bSgjelinek pcparms_t pcparms; 60000209230bSgjelinek 60010209230bSgjelinek pcparms.pc_cid = zone->zone_defaultcid; 60020209230bSgjelinek pcparms.pc_clparms[0] = 0; 60030209230bSgjelinek 60040209230bSgjelinek /* 60050209230bSgjelinek * If setting the class fails, we still want to enter the zone. 60060209230bSgjelinek */ 60070209230bSgjelinek if ((t = pp->p_tlist) != NULL) { 60080209230bSgjelinek do { 60090209230bSgjelinek (void) parmsset(&pcparms, t); 60100209230bSgjelinek } while ((t = t->t_forw) != pp->p_tlist); 60110209230bSgjelinek } 60120209230bSgjelinek } 60130209230bSgjelinek 60147c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 60157c478bd9Sstevel@tonic-gate mutex_exit(&pidlock); 60167c478bd9Sstevel@tonic-gate 60177c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 60187c478bd9Sstevel@tonic-gate /* 60197c478bd9Sstevel@tonic-gate * We're firmly in the zone; let pools progress. 60207c478bd9Sstevel@tonic-gate */ 60217c478bd9Sstevel@tonic-gate pool_unlock(); 60227c478bd9Sstevel@tonic-gate task_rele(oldtk); 60237c478bd9Sstevel@tonic-gate /* 60247c478bd9Sstevel@tonic-gate * We don't need to retain a hold on the zone since we already 60257c478bd9Sstevel@tonic-gate * incremented zone_ntasks, so the zone isn't going anywhere. 60267c478bd9Sstevel@tonic-gate */ 60277c478bd9Sstevel@tonic-gate zone_rele(zone); 60287c478bd9Sstevel@tonic-gate 60297c478bd9Sstevel@tonic-gate /* 60307c478bd9Sstevel@tonic-gate * Chroot 60317c478bd9Sstevel@tonic-gate */ 60327c478bd9Sstevel@tonic-gate vp = zone->zone_rootvp; 60337c478bd9Sstevel@tonic-gate zone_chdir(vp, &PTOU(pp)->u_cdir, pp); 60347c478bd9Sstevel@tonic-gate zone_chdir(vp, &PTOU(pp)->u_rdir, pp); 60357c478bd9Sstevel@tonic-gate 60367c478bd9Sstevel@tonic-gate /* 60377c478bd9Sstevel@tonic-gate * Change process credentials 60387c478bd9Sstevel@tonic-gate */ 60397c478bd9Sstevel@tonic-gate newcr = cralloc(); 60407c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_crlock); 60417c478bd9Sstevel@tonic-gate cr = pp->p_cred; 60427c478bd9Sstevel@tonic-gate crcopy_to(cr, newcr); 60437c478bd9Sstevel@tonic-gate crsetzone(newcr, zone); 60447c478bd9Sstevel@tonic-gate pp->p_cred = newcr; 60457c478bd9Sstevel@tonic-gate 60467c478bd9Sstevel@tonic-gate /* 60477c478bd9Sstevel@tonic-gate * Restrict all process privilege sets to zone limit 60487c478bd9Sstevel@tonic-gate */ 60497c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_PPRIV(newcr)); 60507c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_EPRIV(newcr)); 60517c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_IPRIV(newcr)); 60527c478bd9Sstevel@tonic-gate priv_intersect(zone->zone_privset, &CR_LPRIV(newcr)); 60537c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_crlock); 60547c478bd9Sstevel@tonic-gate crset(pp, newcr); 60557c478bd9Sstevel@tonic-gate 60567c478bd9Sstevel@tonic-gate /* 60577c478bd9Sstevel@tonic-gate * Adjust upcount to reflect zone entry. 60587c478bd9Sstevel@tonic-gate */ 60597c478bd9Sstevel@tonic-gate uid = crgetruid(newcr); 60607c478bd9Sstevel@tonic-gate mutex_enter(&pidlock); 60617c478bd9Sstevel@tonic-gate upcount_dec(uid, GLOBAL_ZONEID); 60627c478bd9Sstevel@tonic-gate upcount_inc(uid, zoneid); 60637c478bd9Sstevel@tonic-gate mutex_exit(&pidlock); 60647c478bd9Sstevel@tonic-gate 60657c478bd9Sstevel@tonic-gate /* 60667c478bd9Sstevel@tonic-gate * Set up core file path and content. 60677c478bd9Sstevel@tonic-gate */ 60687c478bd9Sstevel@tonic-gate set_core_defaults(); 60697c478bd9Sstevel@tonic-gate 60707c478bd9Sstevel@tonic-gate out: 60717c478bd9Sstevel@tonic-gate /* 60727c478bd9Sstevel@tonic-gate * Let the other lwps continue. 60737c478bd9Sstevel@tonic-gate */ 60747c478bd9Sstevel@tonic-gate mutex_enter(&pp->p_lock); 60757c478bd9Sstevel@tonic-gate if (curthread != pp->p_agenttp) 60767c478bd9Sstevel@tonic-gate continuelwps(pp); 60777c478bd9Sstevel@tonic-gate mutex_exit(&pp->p_lock); 60787c478bd9Sstevel@tonic-gate 60797c478bd9Sstevel@tonic-gate return (err != 0 ? set_errno(err) : 0); 60807c478bd9Sstevel@tonic-gate } 60817c478bd9Sstevel@tonic-gate 60827c478bd9Sstevel@tonic-gate /* 60837c478bd9Sstevel@tonic-gate * Systemcall entry point for zone_list(2). 60847c478bd9Sstevel@tonic-gate * 60857c478bd9Sstevel@tonic-gate * Processes running in a (non-global) zone only see themselves. 608645916cd2Sjpk * On labeled systems, they see all zones whose label they dominate. 60877c478bd9Sstevel@tonic-gate */ 60887c478bd9Sstevel@tonic-gate static int 60897c478bd9Sstevel@tonic-gate zone_list(zoneid_t *zoneidlist, uint_t *numzones) 60907c478bd9Sstevel@tonic-gate { 60917c478bd9Sstevel@tonic-gate zoneid_t *zoneids; 609248451833Scarlsonj zone_t *zone, *myzone; 60937c478bd9Sstevel@tonic-gate uint_t user_nzones, real_nzones; 609445916cd2Sjpk uint_t domi_nzones; 609545916cd2Sjpk int error; 60967c478bd9Sstevel@tonic-gate 60977c478bd9Sstevel@tonic-gate if (copyin(numzones, &user_nzones, sizeof (uint_t)) != 0) 60987c478bd9Sstevel@tonic-gate return (set_errno(EFAULT)); 60997c478bd9Sstevel@tonic-gate 610048451833Scarlsonj myzone = curproc->p_zone; 610148451833Scarlsonj if (myzone != global_zone) { 610245916cd2Sjpk bslabel_t *mybslab; 610345916cd2Sjpk 610445916cd2Sjpk if (!is_system_labeled()) { 610545916cd2Sjpk /* just return current zone */ 610645916cd2Sjpk real_nzones = domi_nzones = 1; 610745916cd2Sjpk zoneids = kmem_alloc(sizeof (zoneid_t), KM_SLEEP); 610848451833Scarlsonj zoneids[0] = myzone->zone_id; 610945916cd2Sjpk } else { 611045916cd2Sjpk /* return all zones that are dominated */ 611145916cd2Sjpk mutex_enter(&zonehash_lock); 611245916cd2Sjpk real_nzones = zonecount; 611345916cd2Sjpk domi_nzones = 0; 611445916cd2Sjpk if (real_nzones > 0) { 611545916cd2Sjpk zoneids = kmem_alloc(real_nzones * 611645916cd2Sjpk sizeof (zoneid_t), KM_SLEEP); 611748451833Scarlsonj mybslab = label2bslabel(myzone->zone_slabel); 611845916cd2Sjpk for (zone = list_head(&zone_active); 611945916cd2Sjpk zone != NULL; 612045916cd2Sjpk zone = list_next(&zone_active, zone)) { 612145916cd2Sjpk if (zone->zone_id == GLOBAL_ZONEID) 612245916cd2Sjpk continue; 612348451833Scarlsonj if (zone != myzone && 612448451833Scarlsonj (zone->zone_flags & ZF_IS_SCRATCH)) 612548451833Scarlsonj continue; 612648451833Scarlsonj /* 612748451833Scarlsonj * Note that a label always dominates 612848451833Scarlsonj * itself, so myzone is always included 612948451833Scarlsonj * in the list. 613048451833Scarlsonj */ 613145916cd2Sjpk if (bldominates(mybslab, 613245916cd2Sjpk label2bslabel(zone->zone_slabel))) { 613345916cd2Sjpk zoneids[domi_nzones++] = 613445916cd2Sjpk zone->zone_id; 613545916cd2Sjpk } 613645916cd2Sjpk } 613745916cd2Sjpk } 613845916cd2Sjpk mutex_exit(&zonehash_lock); 613945916cd2Sjpk } 61407c478bd9Sstevel@tonic-gate } else { 61417c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 61427c478bd9Sstevel@tonic-gate real_nzones = zonecount; 614345916cd2Sjpk domi_nzones = 0; 614445916cd2Sjpk if (real_nzones > 0) { 61457c478bd9Sstevel@tonic-gate zoneids = kmem_alloc(real_nzones * sizeof (zoneid_t), 61467c478bd9Sstevel@tonic-gate KM_SLEEP); 61477c478bd9Sstevel@tonic-gate for (zone = list_head(&zone_active); zone != NULL; 61487c478bd9Sstevel@tonic-gate zone = list_next(&zone_active, zone)) 614945916cd2Sjpk zoneids[domi_nzones++] = zone->zone_id; 615045916cd2Sjpk ASSERT(domi_nzones == real_nzones); 61517c478bd9Sstevel@tonic-gate } 61527c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 61537c478bd9Sstevel@tonic-gate } 61547c478bd9Sstevel@tonic-gate 615545916cd2Sjpk /* 615645916cd2Sjpk * If user has allocated space for fewer entries than we found, then 615745916cd2Sjpk * return only up to his limit. Either way, tell him exactly how many 615845916cd2Sjpk * we found. 615945916cd2Sjpk */ 616045916cd2Sjpk if (domi_nzones < user_nzones) 616145916cd2Sjpk user_nzones = domi_nzones; 616245916cd2Sjpk error = 0; 616345916cd2Sjpk if (copyout(&domi_nzones, numzones, sizeof (uint_t)) != 0) { 61647c478bd9Sstevel@tonic-gate error = EFAULT; 616545916cd2Sjpk } else if (zoneidlist != NULL && user_nzones != 0) { 61667c478bd9Sstevel@tonic-gate if (copyout(zoneids, zoneidlist, 61677c478bd9Sstevel@tonic-gate user_nzones * sizeof (zoneid_t)) != 0) 61687c478bd9Sstevel@tonic-gate error = EFAULT; 61697c478bd9Sstevel@tonic-gate } 61707c478bd9Sstevel@tonic-gate 617145916cd2Sjpk if (real_nzones > 0) 61727c478bd9Sstevel@tonic-gate kmem_free(zoneids, real_nzones * sizeof (zoneid_t)); 61737c478bd9Sstevel@tonic-gate 617445916cd2Sjpk if (error != 0) 61757c478bd9Sstevel@tonic-gate return (set_errno(error)); 61767c478bd9Sstevel@tonic-gate else 61777c478bd9Sstevel@tonic-gate return (0); 61787c478bd9Sstevel@tonic-gate } 61797c478bd9Sstevel@tonic-gate 61807c478bd9Sstevel@tonic-gate /* 61817c478bd9Sstevel@tonic-gate * Systemcall entry point for zone_lookup(2). 61827c478bd9Sstevel@tonic-gate * 618345916cd2Sjpk * Non-global zones are only able to see themselves and (on labeled systems) 618445916cd2Sjpk * the zones they dominate. 61857c478bd9Sstevel@tonic-gate */ 61867c478bd9Sstevel@tonic-gate static zoneid_t 61877c478bd9Sstevel@tonic-gate zone_lookup(const char *zone_name) 61887c478bd9Sstevel@tonic-gate { 61897c478bd9Sstevel@tonic-gate char *kname; 61907c478bd9Sstevel@tonic-gate zone_t *zone; 61917c478bd9Sstevel@tonic-gate zoneid_t zoneid; 61927c478bd9Sstevel@tonic-gate int err; 61937c478bd9Sstevel@tonic-gate 61947c478bd9Sstevel@tonic-gate if (zone_name == NULL) { 61957c478bd9Sstevel@tonic-gate /* return caller's zone id */ 61967c478bd9Sstevel@tonic-gate return (getzoneid()); 61977c478bd9Sstevel@tonic-gate } 61987c478bd9Sstevel@tonic-gate 61997c478bd9Sstevel@tonic-gate kname = kmem_zalloc(ZONENAME_MAX, KM_SLEEP); 62007c478bd9Sstevel@tonic-gate if ((err = copyinstr(zone_name, kname, ZONENAME_MAX, NULL)) != 0) { 62017c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 62027c478bd9Sstevel@tonic-gate return (set_errno(err)); 62037c478bd9Sstevel@tonic-gate } 62047c478bd9Sstevel@tonic-gate 62057c478bd9Sstevel@tonic-gate mutex_enter(&zonehash_lock); 62067c478bd9Sstevel@tonic-gate zone = zone_find_all_by_name(kname); 62077c478bd9Sstevel@tonic-gate kmem_free(kname, ZONENAME_MAX); 620845916cd2Sjpk /* 620945916cd2Sjpk * In a non-global zone, can only lookup global and own name. 621045916cd2Sjpk * In Trusted Extensions zone label dominance rules apply. 621145916cd2Sjpk */ 621245916cd2Sjpk if (zone == NULL || 621345916cd2Sjpk zone_status_get(zone) < ZONE_IS_READY || 621445916cd2Sjpk !zone_list_access(zone)) { 62157c478bd9Sstevel@tonic-gate mutex_exit(&zonehash_lock); 62167c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 621745916cd2Sjpk } else { 621845916cd2Sjpk zoneid = zone->zone_id; 621945916cd2Sjpk mutex_exit(&zonehash_lock); 622045916cd2Sjpk return (zoneid); 62217c478bd9Sstevel@tonic-gate } 62227c478bd9Sstevel@tonic-gate } 62237c478bd9Sstevel@tonic-gate 6224821c4a97Sdp static int 6225821c4a97Sdp zone_version(int *version_arg) 6226821c4a97Sdp { 6227821c4a97Sdp int version = ZONE_SYSCALL_API_VERSION; 6228821c4a97Sdp 6229821c4a97Sdp if (copyout(&version, version_arg, sizeof (int)) != 0) 6230821c4a97Sdp return (set_errno(EFAULT)); 6231821c4a97Sdp return (0); 6232821c4a97Sdp } 6233821c4a97Sdp 62347c478bd9Sstevel@tonic-gate /* ARGSUSED */ 62357c478bd9Sstevel@tonic-gate long 6236fa9e4066Sahrens zone(int cmd, void *arg1, void *arg2, void *arg3, void *arg4) 62377c478bd9Sstevel@tonic-gate { 62387c478bd9Sstevel@tonic-gate zone_def zs; 62392b24ab6bSSebastien Roy int err; 62407c478bd9Sstevel@tonic-gate 62417c478bd9Sstevel@tonic-gate switch (cmd) { 62427c478bd9Sstevel@tonic-gate case ZONE_CREATE: 62437c478bd9Sstevel@tonic-gate if (get_udatamodel() == DATAMODEL_NATIVE) { 62447c478bd9Sstevel@tonic-gate if (copyin(arg1, &zs, sizeof (zone_def))) { 62457c478bd9Sstevel@tonic-gate return (set_errno(EFAULT)); 62467c478bd9Sstevel@tonic-gate } 62477c478bd9Sstevel@tonic-gate } else { 62487c478bd9Sstevel@tonic-gate #ifdef _SYSCALL32_IMPL 62497c478bd9Sstevel@tonic-gate zone_def32 zs32; 62507c478bd9Sstevel@tonic-gate 62517c478bd9Sstevel@tonic-gate if (copyin(arg1, &zs32, sizeof (zone_def32))) { 62527c478bd9Sstevel@tonic-gate return (set_errno(EFAULT)); 62537c478bd9Sstevel@tonic-gate } 62547c478bd9Sstevel@tonic-gate zs.zone_name = 62557c478bd9Sstevel@tonic-gate (const char *)(unsigned long)zs32.zone_name; 62567c478bd9Sstevel@tonic-gate zs.zone_root = 62577c478bd9Sstevel@tonic-gate (const char *)(unsigned long)zs32.zone_root; 62587c478bd9Sstevel@tonic-gate zs.zone_privs = 62597c478bd9Sstevel@tonic-gate (const struct priv_set *) 62607c478bd9Sstevel@tonic-gate (unsigned long)zs32.zone_privs; 62616f70df68Sdp zs.zone_privssz = zs32.zone_privssz; 62627c478bd9Sstevel@tonic-gate zs.rctlbuf = (caddr_t)(unsigned long)zs32.rctlbuf; 62637c478bd9Sstevel@tonic-gate zs.rctlbufsz = zs32.rctlbufsz; 6264fa9e4066Sahrens zs.zfsbuf = (caddr_t)(unsigned long)zs32.zfsbuf; 6265fa9e4066Sahrens zs.zfsbufsz = zs32.zfsbufsz; 62667c478bd9Sstevel@tonic-gate zs.extended_error = 62677c478bd9Sstevel@tonic-gate (int *)(unsigned long)zs32.extended_error; 626845916cd2Sjpk zs.match = zs32.match; 626945916cd2Sjpk zs.doi = zs32.doi; 627045916cd2Sjpk zs.label = (const bslabel_t *)(uintptr_t)zs32.label; 6271f4b3ec61Sdh zs.flags = zs32.flags; 62727c478bd9Sstevel@tonic-gate #else 62737c478bd9Sstevel@tonic-gate panic("get_udatamodel() returned bogus result\n"); 62747c478bd9Sstevel@tonic-gate #endif 62757c478bd9Sstevel@tonic-gate } 62767c478bd9Sstevel@tonic-gate 62777c478bd9Sstevel@tonic-gate return (zone_create(zs.zone_name, zs.zone_root, 6278821c4a97Sdp zs.zone_privs, zs.zone_privssz, 6279821c4a97Sdp (caddr_t)zs.rctlbuf, zs.rctlbufsz, 6280821c4a97Sdp (caddr_t)zs.zfsbuf, zs.zfsbufsz, 628145916cd2Sjpk zs.extended_error, zs.match, zs.doi, 6282f4b3ec61Sdh zs.label, zs.flags)); 62837c478bd9Sstevel@tonic-gate case ZONE_BOOT: 62843f2f09c1Sdp return (zone_boot((zoneid_t)(uintptr_t)arg1)); 62857c478bd9Sstevel@tonic-gate case ZONE_DESTROY: 62867c478bd9Sstevel@tonic-gate return (zone_destroy((zoneid_t)(uintptr_t)arg1)); 62877c478bd9Sstevel@tonic-gate case ZONE_GETATTR: 62887c478bd9Sstevel@tonic-gate return (zone_getattr((zoneid_t)(uintptr_t)arg1, 62897c478bd9Sstevel@tonic-gate (int)(uintptr_t)arg2, arg3, (size_t)arg4)); 62903f2f09c1Sdp case ZONE_SETATTR: 62913f2f09c1Sdp return (zone_setattr((zoneid_t)(uintptr_t)arg1, 62923f2f09c1Sdp (int)(uintptr_t)arg2, arg3, (size_t)arg4)); 62937c478bd9Sstevel@tonic-gate case ZONE_ENTER: 62947c478bd9Sstevel@tonic-gate return (zone_enter((zoneid_t)(uintptr_t)arg1)); 62957c478bd9Sstevel@tonic-gate case ZONE_LIST: 62967c478bd9Sstevel@tonic-gate return (zone_list((zoneid_t *)arg1, (uint_t *)arg2)); 62977c478bd9Sstevel@tonic-gate case ZONE_SHUTDOWN: 62987c478bd9Sstevel@tonic-gate return (zone_shutdown((zoneid_t)(uintptr_t)arg1)); 62997c478bd9Sstevel@tonic-gate case ZONE_LOOKUP: 63007c478bd9Sstevel@tonic-gate return (zone_lookup((const char *)arg1)); 6301821c4a97Sdp case ZONE_VERSION: 6302821c4a97Sdp return (zone_version((int *)arg1)); 6303f4b3ec61Sdh case ZONE_ADD_DATALINK: 6304f4b3ec61Sdh return (zone_add_datalink((zoneid_t)(uintptr_t)arg1, 63052b24ab6bSSebastien Roy (datalink_id_t)(uintptr_t)arg2)); 6306f4b3ec61Sdh case ZONE_DEL_DATALINK: 6307f4b3ec61Sdh return (zone_remove_datalink((zoneid_t)(uintptr_t)arg1, 63082b24ab6bSSebastien Roy (datalink_id_t)(uintptr_t)arg2)); 63092b24ab6bSSebastien Roy case ZONE_CHECK_DATALINK: { 63102b24ab6bSSebastien Roy zoneid_t zoneid; 63112b24ab6bSSebastien Roy boolean_t need_copyout; 63122b24ab6bSSebastien Roy 63132b24ab6bSSebastien Roy if (copyin(arg1, &zoneid, sizeof (zoneid)) != 0) 63142b24ab6bSSebastien Roy return (EFAULT); 63152b24ab6bSSebastien Roy need_copyout = (zoneid == ALL_ZONES); 63162b24ab6bSSebastien Roy err = zone_check_datalink(&zoneid, 63172b24ab6bSSebastien Roy (datalink_id_t)(uintptr_t)arg2); 63182b24ab6bSSebastien Roy if (err == 0 && need_copyout) { 63192b24ab6bSSebastien Roy if (copyout(&zoneid, arg1, sizeof (zoneid)) != 0) 63202b24ab6bSSebastien Roy err = EFAULT; 63212b24ab6bSSebastien Roy } 63222b24ab6bSSebastien Roy return (err == 0 ? 0 : set_errno(err)); 63232b24ab6bSSebastien Roy } 6324f4b3ec61Sdh case ZONE_LIST_DATALINK: 6325f4b3ec61Sdh return (zone_list_datalink((zoneid_t)(uintptr_t)arg1, 63262b24ab6bSSebastien Roy (int *)arg2, (datalink_id_t *)(uintptr_t)arg3)); 63277c478bd9Sstevel@tonic-gate default: 63287c478bd9Sstevel@tonic-gate return (set_errno(EINVAL)); 63297c478bd9Sstevel@tonic-gate } 63307c478bd9Sstevel@tonic-gate } 63317c478bd9Sstevel@tonic-gate 63327c478bd9Sstevel@tonic-gate struct zarg { 63337c478bd9Sstevel@tonic-gate zone_t *zone; 63347c478bd9Sstevel@tonic-gate zone_cmd_arg_t arg; 63357c478bd9Sstevel@tonic-gate }; 63367c478bd9Sstevel@tonic-gate 63377c478bd9Sstevel@tonic-gate static int 63387c478bd9Sstevel@tonic-gate zone_lookup_door(const char *zone_name, door_handle_t *doorp) 63397c478bd9Sstevel@tonic-gate { 63407c478bd9Sstevel@tonic-gate char *buf; 63417c478bd9Sstevel@tonic-gate size_t buflen; 63427c478bd9Sstevel@tonic-gate int error; 63437c478bd9Sstevel@tonic-gate 63447c478bd9Sstevel@tonic-gate buflen = sizeof (ZONE_DOOR_PATH) + strlen(zone_name); 63457c478bd9Sstevel@tonic-gate buf = kmem_alloc(buflen, KM_SLEEP); 63467c478bd9Sstevel@tonic-gate (void) snprintf(buf, buflen, ZONE_DOOR_PATH, zone_name); 63477c478bd9Sstevel@tonic-gate error = door_ki_open(buf, doorp); 63487c478bd9Sstevel@tonic-gate kmem_free(buf, buflen); 63497c478bd9Sstevel@tonic-gate return (error); 63507c478bd9Sstevel@tonic-gate } 63517c478bd9Sstevel@tonic-gate 63527c478bd9Sstevel@tonic-gate static void 63537c478bd9Sstevel@tonic-gate zone_release_door(door_handle_t *doorp) 63547c478bd9Sstevel@tonic-gate { 63557c478bd9Sstevel@tonic-gate door_ki_rele(*doorp); 63567c478bd9Sstevel@tonic-gate *doorp = NULL; 63577c478bd9Sstevel@tonic-gate } 63587c478bd9Sstevel@tonic-gate 63597c478bd9Sstevel@tonic-gate static void 63607c478bd9Sstevel@tonic-gate zone_ki_call_zoneadmd(struct zarg *zargp) 63617c478bd9Sstevel@tonic-gate { 63627c478bd9Sstevel@tonic-gate door_handle_t door = NULL; 63637c478bd9Sstevel@tonic-gate door_arg_t darg, save_arg; 63647c478bd9Sstevel@tonic-gate char *zone_name; 63657c478bd9Sstevel@tonic-gate size_t zone_namelen; 63667c478bd9Sstevel@tonic-gate zoneid_t zoneid; 63677c478bd9Sstevel@tonic-gate zone_t *zone; 63687c478bd9Sstevel@tonic-gate zone_cmd_arg_t arg; 63697c478bd9Sstevel@tonic-gate uint64_t uniqid; 63707c478bd9Sstevel@tonic-gate size_t size; 63717c478bd9Sstevel@tonic-gate int error; 63727c478bd9Sstevel@tonic-gate int retry; 63737c478bd9Sstevel@tonic-gate 63747c478bd9Sstevel@tonic-gate zone = zargp->zone; 63757c478bd9Sstevel@tonic-gate arg = zargp->arg; 63767c478bd9Sstevel@tonic-gate kmem_free(zargp, sizeof (*zargp)); 63777c478bd9Sstevel@tonic-gate 63787c478bd9Sstevel@tonic-gate zone_namelen = strlen(zone->zone_name) + 1; 63797c478bd9Sstevel@tonic-gate zone_name = kmem_alloc(zone_namelen, KM_SLEEP); 63807c478bd9Sstevel@tonic-gate bcopy(zone->zone_name, zone_name, zone_namelen); 63817c478bd9Sstevel@tonic-gate zoneid = zone->zone_id; 63827c478bd9Sstevel@tonic-gate uniqid = zone->zone_uniqid; 63837c478bd9Sstevel@tonic-gate /* 63847c478bd9Sstevel@tonic-gate * zoneadmd may be down, but at least we can empty out the zone. 63857c478bd9Sstevel@tonic-gate * We can ignore the return value of zone_empty() since we're called 63867c478bd9Sstevel@tonic-gate * from a kernel thread and know we won't be delivered any signals. 63877c478bd9Sstevel@tonic-gate */ 63887c478bd9Sstevel@tonic-gate ASSERT(curproc == &p0); 63897c478bd9Sstevel@tonic-gate (void) zone_empty(zone); 63907c478bd9Sstevel@tonic-gate ASSERT(zone_status_get(zone) >= ZONE_IS_EMPTY); 63917c478bd9Sstevel@tonic-gate zone_rele(zone); 63927c478bd9Sstevel@tonic-gate 63937c478bd9Sstevel@tonic-gate size = sizeof (arg); 63947c478bd9Sstevel@tonic-gate darg.rbuf = (char *)&arg; 63957c478bd9Sstevel@tonic-gate darg.data_ptr = (char *)&arg; 63967c478bd9Sstevel@tonic-gate darg.rsize = size; 63977c478bd9Sstevel@tonic-gate darg.data_size = size; 63987c478bd9Sstevel@tonic-gate darg.desc_ptr = NULL; 63997c478bd9Sstevel@tonic-gate darg.desc_num = 0; 64007c478bd9Sstevel@tonic-gate 64017c478bd9Sstevel@tonic-gate save_arg = darg; 64027c478bd9Sstevel@tonic-gate /* 64037c478bd9Sstevel@tonic-gate * Since we're not holding a reference to the zone, any number of 64047c478bd9Sstevel@tonic-gate * things can go wrong, including the zone disappearing before we get a 64057c478bd9Sstevel@tonic-gate * chance to talk to zoneadmd. 64067c478bd9Sstevel@tonic-gate */ 64077c478bd9Sstevel@tonic-gate for (retry = 0; /* forever */; retry++) { 64087c478bd9Sstevel@tonic-gate if (door == NULL && 64097c478bd9Sstevel@tonic-gate (error = zone_lookup_door(zone_name, &door)) != 0) { 64107c478bd9Sstevel@tonic-gate goto next; 64117c478bd9Sstevel@tonic-gate } 64127c478bd9Sstevel@tonic-gate ASSERT(door != NULL); 64137c478bd9Sstevel@tonic-gate 6414323a81d9Sjwadams if ((error = door_ki_upcall_limited(door, &darg, NULL, 6415323a81d9Sjwadams SIZE_MAX, 0)) == 0) { 64167c478bd9Sstevel@tonic-gate break; 64177c478bd9Sstevel@tonic-gate } 64187c478bd9Sstevel@tonic-gate switch (error) { 64197c478bd9Sstevel@tonic-gate case EINTR: 64207c478bd9Sstevel@tonic-gate /* FALLTHROUGH */ 64217c478bd9Sstevel@tonic-gate case EAGAIN: /* process may be forking */ 64227c478bd9Sstevel@tonic-gate /* 64237c478bd9Sstevel@tonic-gate * Back off for a bit 64247c478bd9Sstevel@tonic-gate */ 64257c478bd9Sstevel@tonic-gate break; 64267c478bd9Sstevel@tonic-gate case EBADF: 64277c478bd9Sstevel@tonic-gate zone_release_door(&door); 64287c478bd9Sstevel@tonic-gate if (zone_lookup_door(zone_name, &door) != 0) { 64297c478bd9Sstevel@tonic-gate /* 64307c478bd9Sstevel@tonic-gate * zoneadmd may be dead, but it may come back to 64317c478bd9Sstevel@tonic-gate * life later. 64327c478bd9Sstevel@tonic-gate */ 64337c478bd9Sstevel@tonic-gate break; 64347c478bd9Sstevel@tonic-gate } 64357c478bd9Sstevel@tonic-gate break; 64367c478bd9Sstevel@tonic-gate default: 64377c478bd9Sstevel@tonic-gate cmn_err(CE_WARN, 64387c478bd9Sstevel@tonic-gate "zone_ki_call_zoneadmd: door_ki_upcall error %d\n", 64397c478bd9Sstevel@tonic-gate error); 64407c478bd9Sstevel@tonic-gate goto out; 64417c478bd9Sstevel@tonic-gate } 64427c478bd9Sstevel@tonic-gate next: 64437c478bd9Sstevel@tonic-gate /* 64447c478bd9Sstevel@tonic-gate * If this isn't the same zone_t that we originally had in mind, 64457c478bd9Sstevel@tonic-gate * then this is the same as if two kadmin requests come in at 64467c478bd9Sstevel@tonic-gate * the same time: the first one wins. This means we lose, so we 64477c478bd9Sstevel@tonic-gate * bail. 64487c478bd9Sstevel@tonic-gate */ 64497c478bd9Sstevel@tonic-gate if ((zone = zone_find_by_id(zoneid)) == NULL) { 64507c478bd9Sstevel@tonic-gate /* 64517c478bd9Sstevel@tonic-gate * Problem is solved. 64527c478bd9Sstevel@tonic-gate */ 64537c478bd9Sstevel@tonic-gate break; 64547c478bd9Sstevel@tonic-gate } 64557c478bd9Sstevel@tonic-gate if (zone->zone_uniqid != uniqid) { 64567c478bd9Sstevel@tonic-gate /* 64577c478bd9Sstevel@tonic-gate * zoneid recycled 64587c478bd9Sstevel@tonic-gate */ 64597c478bd9Sstevel@tonic-gate zone_rele(zone); 64607c478bd9Sstevel@tonic-gate break; 64617c478bd9Sstevel@tonic-gate } 64627c478bd9Sstevel@tonic-gate /* 64637c478bd9Sstevel@tonic-gate * We could zone_status_timedwait(), but there doesn't seem to 64647c478bd9Sstevel@tonic-gate * be much point in doing that (plus, it would mean that 64657c478bd9Sstevel@tonic-gate * zone_free() isn't called until this thread exits). 64667c478bd9Sstevel@tonic-gate */ 64677c478bd9Sstevel@tonic-gate zone_rele(zone); 64687c478bd9Sstevel@tonic-gate delay(hz); 64697c478bd9Sstevel@tonic-gate darg = save_arg; 64707c478bd9Sstevel@tonic-gate } 64717c478bd9Sstevel@tonic-gate out: 64727c478bd9Sstevel@tonic-gate if (door != NULL) { 64737c478bd9Sstevel@tonic-gate zone_release_door(&door); 64747c478bd9Sstevel@tonic-gate } 64757c478bd9Sstevel@tonic-gate kmem_free(zone_name, zone_namelen); 64767c478bd9Sstevel@tonic-gate thread_exit(); 64777c478bd9Sstevel@tonic-gate } 64787c478bd9Sstevel@tonic-gate 64797c478bd9Sstevel@tonic-gate /* 64803f2f09c1Sdp * Entry point for uadmin() to tell the zone to go away or reboot. Analog to 64813f2f09c1Sdp * kadmin(). The caller is a process in the zone. 64827c478bd9Sstevel@tonic-gate * 64837c478bd9Sstevel@tonic-gate * In order to shutdown the zone, we will hand off control to zoneadmd 64847c478bd9Sstevel@tonic-gate * (running in the global zone) via a door. We do a half-hearted job at 64857c478bd9Sstevel@tonic-gate * killing all processes in the zone, create a kernel thread to contact 64867c478bd9Sstevel@tonic-gate * zoneadmd, and make note of the "uniqid" of the zone. The uniqid is 64877c478bd9Sstevel@tonic-gate * a form of generation number used to let zoneadmd (as well as 64887c478bd9Sstevel@tonic-gate * zone_destroy()) know exactly which zone they're re talking about. 64897c478bd9Sstevel@tonic-gate */ 64907c478bd9Sstevel@tonic-gate int 64913f2f09c1Sdp zone_kadmin(int cmd, int fcn, const char *mdep, cred_t *credp) 64927c478bd9Sstevel@tonic-gate { 64937c478bd9Sstevel@tonic-gate struct zarg *zargp; 64947c478bd9Sstevel@tonic-gate zone_cmd_t zcmd; 64957c478bd9Sstevel@tonic-gate zone_t *zone; 64967c478bd9Sstevel@tonic-gate 64977c478bd9Sstevel@tonic-gate zone = curproc->p_zone; 64987c478bd9Sstevel@tonic-gate ASSERT(getzoneid() != GLOBAL_ZONEID); 64997c478bd9Sstevel@tonic-gate 65007c478bd9Sstevel@tonic-gate switch (cmd) { 65017c478bd9Sstevel@tonic-gate case A_SHUTDOWN: 65027c478bd9Sstevel@tonic-gate switch (fcn) { 65037c478bd9Sstevel@tonic-gate case AD_HALT: 65047c478bd9Sstevel@tonic-gate case AD_POWEROFF: 65057c478bd9Sstevel@tonic-gate zcmd = Z_HALT; 65067c478bd9Sstevel@tonic-gate break; 65077c478bd9Sstevel@tonic-gate case AD_BOOT: 65087c478bd9Sstevel@tonic-gate zcmd = Z_REBOOT; 65097c478bd9Sstevel@tonic-gate break; 65107c478bd9Sstevel@tonic-gate case AD_IBOOT: 65117c478bd9Sstevel@tonic-gate case AD_SBOOT: 65127c478bd9Sstevel@tonic-gate case AD_SIBOOT: 65137c478bd9Sstevel@tonic-gate case AD_NOSYNC: 65147c478bd9Sstevel@tonic-gate return (ENOTSUP); 65157c478bd9Sstevel@tonic-gate default: 65167c478bd9Sstevel@tonic-gate return (EINVAL); 65177c478bd9Sstevel@tonic-gate } 65187c478bd9Sstevel@tonic-gate break; 65197c478bd9Sstevel@tonic-gate case A_REBOOT: 65207c478bd9Sstevel@tonic-gate zcmd = Z_REBOOT; 65217c478bd9Sstevel@tonic-gate break; 65227c478bd9Sstevel@tonic-gate case A_FTRACE: 65237c478bd9Sstevel@tonic-gate case A_REMOUNT: 65247c478bd9Sstevel@tonic-gate case A_FREEZE: 65257c478bd9Sstevel@tonic-gate case A_DUMP: 6526753a6d45SSherry Moore case A_CONFIG: 65277c478bd9Sstevel@tonic-gate return (ENOTSUP); 65287c478bd9Sstevel@tonic-gate default: 65297c478bd9Sstevel@tonic-gate ASSERT(cmd != A_SWAPCTL); /* handled by uadmin() */ 65307c478bd9Sstevel@tonic-gate return (EINVAL); 65317c478bd9Sstevel@tonic-gate } 65327c478bd9Sstevel@tonic-gate 65337c478bd9Sstevel@tonic-gate if (secpolicy_zone_admin(credp, B_FALSE)) 65347c478bd9Sstevel@tonic-gate return (EPERM); 65357c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 65363f2f09c1Sdp 65377c478bd9Sstevel@tonic-gate /* 65387c478bd9Sstevel@tonic-gate * zone_status can't be ZONE_IS_EMPTY or higher since curproc 65397c478bd9Sstevel@tonic-gate * is in the zone. 65407c478bd9Sstevel@tonic-gate */ 65417c478bd9Sstevel@tonic-gate ASSERT(zone_status_get(zone) < ZONE_IS_EMPTY); 65427c478bd9Sstevel@tonic-gate if (zone_status_get(zone) > ZONE_IS_RUNNING) { 65437c478bd9Sstevel@tonic-gate /* 65447c478bd9Sstevel@tonic-gate * This zone is already on its way down. 65457c478bd9Sstevel@tonic-gate */ 65467c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 65477c478bd9Sstevel@tonic-gate return (0); 65487c478bd9Sstevel@tonic-gate } 65497c478bd9Sstevel@tonic-gate /* 65507c478bd9Sstevel@tonic-gate * Prevent future zone_enter()s 65517c478bd9Sstevel@tonic-gate */ 65527c478bd9Sstevel@tonic-gate zone_status_set(zone, ZONE_IS_SHUTTING_DOWN); 65537c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 65547c478bd9Sstevel@tonic-gate 65557c478bd9Sstevel@tonic-gate /* 65567c478bd9Sstevel@tonic-gate * Kill everyone now and call zoneadmd later. 65577c478bd9Sstevel@tonic-gate * zone_ki_call_zoneadmd() will do a more thorough job of this 65587c478bd9Sstevel@tonic-gate * later. 65597c478bd9Sstevel@tonic-gate */ 65607c478bd9Sstevel@tonic-gate killall(zone->zone_id); 65617c478bd9Sstevel@tonic-gate /* 65627c478bd9Sstevel@tonic-gate * Now, create the thread to contact zoneadmd and do the rest of the 65637c478bd9Sstevel@tonic-gate * work. This thread can't be created in our zone otherwise 65647c478bd9Sstevel@tonic-gate * zone_destroy() would deadlock. 65657c478bd9Sstevel@tonic-gate */ 65663f2f09c1Sdp zargp = kmem_zalloc(sizeof (*zargp), KM_SLEEP); 65677c478bd9Sstevel@tonic-gate zargp->arg.cmd = zcmd; 65687c478bd9Sstevel@tonic-gate zargp->arg.uniqid = zone->zone_uniqid; 65693f2f09c1Sdp zargp->zone = zone; 65707c478bd9Sstevel@tonic-gate (void) strcpy(zargp->arg.locale, "C"); 65713f2f09c1Sdp /* mdep was already copied in for us by uadmin */ 65723f2f09c1Sdp if (mdep != NULL) 65733f2f09c1Sdp (void) strlcpy(zargp->arg.bootbuf, mdep, 65743f2f09c1Sdp sizeof (zargp->arg.bootbuf)); 65753f2f09c1Sdp zone_hold(zone); 65767c478bd9Sstevel@tonic-gate 65777c478bd9Sstevel@tonic-gate (void) thread_create(NULL, 0, zone_ki_call_zoneadmd, zargp, 0, &p0, 65787c478bd9Sstevel@tonic-gate TS_RUN, minclsyspri); 65797c478bd9Sstevel@tonic-gate exit(CLD_EXITED, 0); 65807c478bd9Sstevel@tonic-gate 65817c478bd9Sstevel@tonic-gate return (EINVAL); 65827c478bd9Sstevel@tonic-gate } 65837c478bd9Sstevel@tonic-gate 65847c478bd9Sstevel@tonic-gate /* 65857c478bd9Sstevel@tonic-gate * Entry point so kadmin(A_SHUTDOWN, ...) can set the global zone's 65867c478bd9Sstevel@tonic-gate * status to ZONE_IS_SHUTTING_DOWN. 65878f983ab3Sjv * 65888f983ab3Sjv * This function also shuts down all running zones to ensure that they won't 65898f983ab3Sjv * fork new processes. 65907c478bd9Sstevel@tonic-gate */ 65917c478bd9Sstevel@tonic-gate void 65927c478bd9Sstevel@tonic-gate zone_shutdown_global(void) 65937c478bd9Sstevel@tonic-gate { 65948f983ab3Sjv zone_t *current_zonep; 65957c478bd9Sstevel@tonic-gate 65968f983ab3Sjv ASSERT(INGLOBALZONE(curproc)); 65978f983ab3Sjv mutex_enter(&zonehash_lock); 65987c478bd9Sstevel@tonic-gate mutex_enter(&zone_status_lock); 65998f983ab3Sjv 66008f983ab3Sjv /* Modify the global zone's status first. */ 66017c478bd9Sstevel@tonic-gate ASSERT(zone_status_get(global_zone) == ZONE_IS_RUNNING); 66027c478bd9Sstevel@tonic-gate zone_status_set(global_zone, ZONE_IS_SHUTTING_DOWN); 66038f983ab3Sjv 66048f983ab3Sjv /* 66058f983ab3Sjv * Now change the states of all running zones to ZONE_IS_SHUTTING_DOWN. 66068f983ab3Sjv * We don't mark all zones with ZONE_IS_SHUTTING_DOWN because doing so 66078f983ab3Sjv * could cause assertions to fail (e.g., assertions about a zone's 66088f983ab3Sjv * state during initialization, readying, or booting) or produce races. 66098f983ab3Sjv * We'll let threads continue to initialize and ready new zones: they'll 66108f983ab3Sjv * fail to boot the new zones when they see that the global zone is 66118f983ab3Sjv * shutting down. 66128f983ab3Sjv */ 66138f983ab3Sjv for (current_zonep = list_head(&zone_active); current_zonep != NULL; 66148f983ab3Sjv current_zonep = list_next(&zone_active, current_zonep)) { 66158f983ab3Sjv if (zone_status_get(current_zonep) == ZONE_IS_RUNNING) 66168f983ab3Sjv zone_status_set(current_zonep, ZONE_IS_SHUTTING_DOWN); 66178f983ab3Sjv } 66187c478bd9Sstevel@tonic-gate mutex_exit(&zone_status_lock); 66198f983ab3Sjv mutex_exit(&zonehash_lock); 66207c478bd9Sstevel@tonic-gate } 6621fa9e4066Sahrens 6622fa9e4066Sahrens /* 6623fa9e4066Sahrens * Returns true if the named dataset is visible in the current zone. 6624fa9e4066Sahrens * The 'write' parameter is set to 1 if the dataset is also writable. 6625fa9e4066Sahrens */ 6626fa9e4066Sahrens int 6627fa9e4066Sahrens zone_dataset_visible(const char *dataset, int *write) 6628fa9e4066Sahrens { 6629658a1dc7SSanjeev Bagewadi static int zfstype = -1; 6630fa9e4066Sahrens zone_dataset_t *zd; 6631fa9e4066Sahrens size_t len; 6632fa9e4066Sahrens zone_t *zone = curproc->p_zone; 6633658a1dc7SSanjeev Bagewadi const char *name = NULL; 6634658a1dc7SSanjeev Bagewadi vfs_t *vfsp = NULL; 6635fa9e4066Sahrens 6636fa9e4066Sahrens if (dataset[0] == '\0') 6637fa9e4066Sahrens return (0); 6638fa9e4066Sahrens 6639fa9e4066Sahrens /* 6640fa9e4066Sahrens * Walk the list once, looking for datasets which match exactly, or 6641fa9e4066Sahrens * specify a dataset underneath an exported dataset. If found, return 6642fa9e4066Sahrens * true and note that it is writable. 6643fa9e4066Sahrens */ 6644fa9e4066Sahrens for (zd = list_head(&zone->zone_datasets); zd != NULL; 6645fa9e4066Sahrens zd = list_next(&zone->zone_datasets, zd)) { 6646fa9e4066Sahrens 6647fa9e4066Sahrens len = strlen(zd->zd_dataset); 6648fa9e4066Sahrens if (strlen(dataset) >= len && 6649fa9e4066Sahrens bcmp(dataset, zd->zd_dataset, len) == 0 && 665095c9592aSmaybee (dataset[len] == '\0' || dataset[len] == '/' || 665195c9592aSmaybee dataset[len] == '@')) { 6652fa9e4066Sahrens if (write) 6653fa9e4066Sahrens *write = 1; 6654fa9e4066Sahrens return (1); 6655fa9e4066Sahrens } 6656fa9e4066Sahrens } 6657fa9e4066Sahrens 6658fa9e4066Sahrens /* 6659fa9e4066Sahrens * Walk the list a second time, searching for datasets which are parents 6660fa9e4066Sahrens * of exported datasets. These should be visible, but read-only. 6661fa9e4066Sahrens * 6662fa9e4066Sahrens * Note that we also have to support forms such as 'pool/dataset/', with 6663fa9e4066Sahrens * a trailing slash. 6664fa9e4066Sahrens */ 6665fa9e4066Sahrens for (zd = list_head(&zone->zone_datasets); zd != NULL; 6666fa9e4066Sahrens zd = list_next(&zone->zone_datasets, zd)) { 6667fa9e4066Sahrens 6668fa9e4066Sahrens len = strlen(dataset); 6669fa9e4066Sahrens if (dataset[len - 1] == '/') 6670fa9e4066Sahrens len--; /* Ignore trailing slash */ 6671fa9e4066Sahrens if (len < strlen(zd->zd_dataset) && 6672fa9e4066Sahrens bcmp(dataset, zd->zd_dataset, len) == 0 && 6673fa9e4066Sahrens zd->zd_dataset[len] == '/') { 6674fa9e4066Sahrens if (write) 6675fa9e4066Sahrens *write = 0; 6676fa9e4066Sahrens return (1); 6677fa9e4066Sahrens } 6678fa9e4066Sahrens } 6679fa9e4066Sahrens 6680658a1dc7SSanjeev Bagewadi /* 6681658a1dc7SSanjeev Bagewadi * We reach here if the given dataset is not found in the zone_dataset 6682658a1dc7SSanjeev Bagewadi * list. Check if this dataset was added as a filesystem (ie. "add fs") 6683658a1dc7SSanjeev Bagewadi * instead of delegation. For this we search for the dataset in the 6684658a1dc7SSanjeev Bagewadi * zone_vfslist of this zone. If found, return true and note that it is 6685658a1dc7SSanjeev Bagewadi * not writable. 6686658a1dc7SSanjeev Bagewadi */ 6687658a1dc7SSanjeev Bagewadi 6688658a1dc7SSanjeev Bagewadi /* 6689658a1dc7SSanjeev Bagewadi * Initialize zfstype if it is not initialized yet. 6690658a1dc7SSanjeev Bagewadi */ 6691658a1dc7SSanjeev Bagewadi if (zfstype == -1) { 6692658a1dc7SSanjeev Bagewadi struct vfssw *vswp = vfs_getvfssw("zfs"); 6693658a1dc7SSanjeev Bagewadi zfstype = vswp - vfssw; 6694658a1dc7SSanjeev Bagewadi vfs_unrefvfssw(vswp); 6695658a1dc7SSanjeev Bagewadi } 6696658a1dc7SSanjeev Bagewadi 6697658a1dc7SSanjeev Bagewadi vfs_list_read_lock(); 6698658a1dc7SSanjeev Bagewadi vfsp = zone->zone_vfslist; 6699658a1dc7SSanjeev Bagewadi do { 6700658a1dc7SSanjeev Bagewadi ASSERT(vfsp); 6701658a1dc7SSanjeev Bagewadi if (vfsp->vfs_fstype == zfstype) { 6702658a1dc7SSanjeev Bagewadi name = refstr_value(vfsp->vfs_resource); 6703658a1dc7SSanjeev Bagewadi 6704658a1dc7SSanjeev Bagewadi /* 6705658a1dc7SSanjeev Bagewadi * Check if we have an exact match. 6706658a1dc7SSanjeev Bagewadi */ 6707658a1dc7SSanjeev Bagewadi if (strcmp(dataset, name) == 0) { 6708658a1dc7SSanjeev Bagewadi vfs_list_unlock(); 6709658a1dc7SSanjeev Bagewadi if (write) 6710658a1dc7SSanjeev Bagewadi *write = 0; 6711658a1dc7SSanjeev Bagewadi return (1); 6712658a1dc7SSanjeev Bagewadi } 6713658a1dc7SSanjeev Bagewadi /* 6714658a1dc7SSanjeev Bagewadi * We need to check if we are looking for parents of 6715658a1dc7SSanjeev Bagewadi * a dataset. These should be visible, but read-only. 6716658a1dc7SSanjeev Bagewadi */ 6717658a1dc7SSanjeev Bagewadi len = strlen(dataset); 6718658a1dc7SSanjeev Bagewadi if (dataset[len - 1] == '/') 6719658a1dc7SSanjeev Bagewadi len--; 6720658a1dc7SSanjeev Bagewadi 6721658a1dc7SSanjeev Bagewadi if (len < strlen(name) && 6722658a1dc7SSanjeev Bagewadi bcmp(dataset, name, len) == 0 && name[len] == '/') { 6723658a1dc7SSanjeev Bagewadi vfs_list_unlock(); 6724658a1dc7SSanjeev Bagewadi if (write) 6725658a1dc7SSanjeev Bagewadi *write = 0; 6726658a1dc7SSanjeev Bagewadi return (1); 6727658a1dc7SSanjeev Bagewadi } 6728658a1dc7SSanjeev Bagewadi } 6729658a1dc7SSanjeev Bagewadi vfsp = vfsp->vfs_zone_next; 6730658a1dc7SSanjeev Bagewadi } while (vfsp != zone->zone_vfslist); 6731658a1dc7SSanjeev Bagewadi 6732658a1dc7SSanjeev Bagewadi vfs_list_unlock(); 6733fa9e4066Sahrens return (0); 6734fa9e4066Sahrens } 673545916cd2Sjpk 673645916cd2Sjpk /* 673745916cd2Sjpk * zone_find_by_any_path() - 673845916cd2Sjpk * 673945916cd2Sjpk * kernel-private routine similar to zone_find_by_path(), but which 674045916cd2Sjpk * effectively compares against zone paths rather than zonerootpath 674145916cd2Sjpk * (i.e., the last component of zonerootpaths, which should be "root/", 674245916cd2Sjpk * are not compared.) This is done in order to accurately identify all 674345916cd2Sjpk * paths, whether zone-visible or not, including those which are parallel 674445916cd2Sjpk * to /root/, such as /dev/, /home/, etc... 674545916cd2Sjpk * 674645916cd2Sjpk * If the specified path does not fall under any zone path then global 674745916cd2Sjpk * zone is returned. 674845916cd2Sjpk * 674945916cd2Sjpk * The treat_abs parameter indicates whether the path should be treated as 675045916cd2Sjpk * an absolute path although it does not begin with "/". (This supports 675145916cd2Sjpk * nfs mount syntax such as host:any/path.) 675245916cd2Sjpk * 675345916cd2Sjpk * The caller is responsible for zone_rele of the returned zone. 675445916cd2Sjpk */ 675545916cd2Sjpk zone_t * 675645916cd2Sjpk zone_find_by_any_path(const char *path, boolean_t treat_abs) 675745916cd2Sjpk { 675845916cd2Sjpk zone_t *zone; 675945916cd2Sjpk int path_offset = 0; 676045916cd2Sjpk 676145916cd2Sjpk if (path == NULL) { 676245916cd2Sjpk zone_hold(global_zone); 676345916cd2Sjpk return (global_zone); 676445916cd2Sjpk } 676545916cd2Sjpk 676645916cd2Sjpk if (*path != '/') { 676745916cd2Sjpk ASSERT(treat_abs); 676845916cd2Sjpk path_offset = 1; 676945916cd2Sjpk } 677045916cd2Sjpk 677145916cd2Sjpk mutex_enter(&zonehash_lock); 677245916cd2Sjpk for (zone = list_head(&zone_active); zone != NULL; 677345916cd2Sjpk zone = list_next(&zone_active, zone)) { 677445916cd2Sjpk char *c; 677545916cd2Sjpk size_t pathlen; 67760f95d722Smp char *rootpath_start; 677745916cd2Sjpk 677845916cd2Sjpk if (zone == global_zone) /* skip global zone */ 677945916cd2Sjpk continue; 678045916cd2Sjpk 678145916cd2Sjpk /* scan backwards to find start of last component */ 678245916cd2Sjpk c = zone->zone_rootpath + zone->zone_rootpathlen - 2; 678345916cd2Sjpk do { 678445916cd2Sjpk c--; 678545916cd2Sjpk } while (*c != '/'); 678645916cd2Sjpk 67870f95d722Smp pathlen = c - zone->zone_rootpath + 1 - path_offset; 67880f95d722Smp rootpath_start = (zone->zone_rootpath + path_offset); 67890f95d722Smp if (strncmp(path, rootpath_start, pathlen) == 0) 679045916cd2Sjpk break; 679145916cd2Sjpk } 679245916cd2Sjpk if (zone == NULL) 679345916cd2Sjpk zone = global_zone; 679445916cd2Sjpk zone_hold(zone); 679545916cd2Sjpk mutex_exit(&zonehash_lock); 679645916cd2Sjpk return (zone); 679745916cd2Sjpk } 6798f4b3ec61Sdh 6799f4b3ec61Sdh /* 68002b24ab6bSSebastien Roy * Finds a zone_dl_t with the given linkid in the given zone. Returns the 68012b24ab6bSSebastien Roy * zone_dl_t pointer if found, and NULL otherwise. 6802f4b3ec61Sdh */ 68032b24ab6bSSebastien Roy static zone_dl_t * 68042b24ab6bSSebastien Roy zone_find_dl(zone_t *zone, datalink_id_t linkid) 6805f4b3ec61Sdh { 68062b24ab6bSSebastien Roy zone_dl_t *zdl; 6807f4b3ec61Sdh 68082b24ab6bSSebastien Roy ASSERT(mutex_owned(&zone->zone_lock)); 68092b24ab6bSSebastien Roy for (zdl = list_head(&zone->zone_dl_list); zdl != NULL; 68102b24ab6bSSebastien Roy zdl = list_next(&zone->zone_dl_list, zdl)) { 68112b24ab6bSSebastien Roy if (zdl->zdl_id == linkid) 6812f4b3ec61Sdh break; 6813f4b3ec61Sdh } 68142b24ab6bSSebastien Roy return (zdl); 68152b24ab6bSSebastien Roy } 68162b24ab6bSSebastien Roy 68172b24ab6bSSebastien Roy static boolean_t 68182b24ab6bSSebastien Roy zone_dl_exists(zone_t *zone, datalink_id_t linkid) 68192b24ab6bSSebastien Roy { 68202b24ab6bSSebastien Roy boolean_t exists; 68212b24ab6bSSebastien Roy 68222b24ab6bSSebastien Roy mutex_enter(&zone->zone_lock); 68232b24ab6bSSebastien Roy exists = (zone_find_dl(zone, linkid) != NULL); 6824f4b3ec61Sdh mutex_exit(&zone->zone_lock); 68252b24ab6bSSebastien Roy return (exists); 6826f4b3ec61Sdh } 6827f4b3ec61Sdh 6828f4b3ec61Sdh /* 68292b24ab6bSSebastien Roy * Add an data link name for the zone. 6830f4b3ec61Sdh */ 6831f4b3ec61Sdh static int 68322b24ab6bSSebastien Roy zone_add_datalink(zoneid_t zoneid, datalink_id_t linkid) 6833f4b3ec61Sdh { 68342b24ab6bSSebastien Roy zone_dl_t *zdl; 6835f4b3ec61Sdh zone_t *zone; 6836f4b3ec61Sdh zone_t *thiszone; 6837f4b3ec61Sdh 68382b24ab6bSSebastien Roy if ((thiszone = zone_find_by_id(zoneid)) == NULL) 6839f4b3ec61Sdh return (set_errno(ENXIO)); 6840f4b3ec61Sdh 68412b24ab6bSSebastien Roy /* Verify that the datalink ID doesn't already belong to a zone. */ 6842f4b3ec61Sdh mutex_enter(&zonehash_lock); 6843f4b3ec61Sdh for (zone = list_head(&zone_active); zone != NULL; 6844f4b3ec61Sdh zone = list_next(&zone_active, zone)) { 68452b24ab6bSSebastien Roy if (zone_dl_exists(zone, linkid)) { 6846f4b3ec61Sdh mutex_exit(&zonehash_lock); 6847f4b3ec61Sdh zone_rele(thiszone); 68482b24ab6bSSebastien Roy return (set_errno((zone == thiszone) ? EEXIST : EPERM)); 6849f4b3ec61Sdh } 6850f4b3ec61Sdh } 68512b24ab6bSSebastien Roy 68522b24ab6bSSebastien Roy zdl = kmem_zalloc(sizeof (*zdl), KM_SLEEP); 68532b24ab6bSSebastien Roy zdl->zdl_id = linkid; 6854550b6e40SSowmini Varadhan zdl->zdl_net = NULL; 6855f4b3ec61Sdh mutex_enter(&thiszone->zone_lock); 68562b24ab6bSSebastien Roy list_insert_head(&thiszone->zone_dl_list, zdl); 6857f4b3ec61Sdh mutex_exit(&thiszone->zone_lock); 6858f4b3ec61Sdh mutex_exit(&zonehash_lock); 6859f4b3ec61Sdh zone_rele(thiszone); 6860f4b3ec61Sdh return (0); 6861f4b3ec61Sdh } 6862f4b3ec61Sdh 6863f4b3ec61Sdh static int 68642b24ab6bSSebastien Roy zone_remove_datalink(zoneid_t zoneid, datalink_id_t linkid) 6865f4b3ec61Sdh { 68662b24ab6bSSebastien Roy zone_dl_t *zdl; 6867f4b3ec61Sdh zone_t *zone; 68682b24ab6bSSebastien Roy int err = 0; 6869f4b3ec61Sdh 68702b24ab6bSSebastien Roy if ((zone = zone_find_by_id(zoneid)) == NULL) 6871f4b3ec61Sdh return (set_errno(EINVAL)); 6872f4b3ec61Sdh 6873f4b3ec61Sdh mutex_enter(&zone->zone_lock); 68742b24ab6bSSebastien Roy if ((zdl = zone_find_dl(zone, linkid)) == NULL) { 68752b24ab6bSSebastien Roy err = ENXIO; 68762b24ab6bSSebastien Roy } else { 68772b24ab6bSSebastien Roy list_remove(&zone->zone_dl_list, zdl); 6878550b6e40SSowmini Varadhan if (zdl->zdl_net != NULL) 6879550b6e40SSowmini Varadhan nvlist_free(zdl->zdl_net); 68802b24ab6bSSebastien Roy kmem_free(zdl, sizeof (zone_dl_t)); 6881f4b3ec61Sdh } 6882f4b3ec61Sdh mutex_exit(&zone->zone_lock); 6883f4b3ec61Sdh zone_rele(zone); 68842b24ab6bSSebastien Roy return (err == 0 ? 0 : set_errno(err)); 6885f4b3ec61Sdh } 6886f4b3ec61Sdh 6887f4b3ec61Sdh /* 68882b24ab6bSSebastien Roy * Using the zoneidp as ALL_ZONES, we can lookup which zone has been assigned 68892b24ab6bSSebastien Roy * the linkid. Otherwise we just check if the specified zoneidp has been 68902b24ab6bSSebastien Roy * assigned the supplied linkid. 6891f4b3ec61Sdh */ 68922b24ab6bSSebastien Roy int 68932b24ab6bSSebastien Roy zone_check_datalink(zoneid_t *zoneidp, datalink_id_t linkid) 6894f4b3ec61Sdh { 6895f4b3ec61Sdh zone_t *zone; 68962b24ab6bSSebastien Roy int err = ENXIO; 6897f4b3ec61Sdh 68982b24ab6bSSebastien Roy if (*zoneidp != ALL_ZONES) { 68992b24ab6bSSebastien Roy if ((zone = zone_find_by_id(*zoneidp)) != NULL) { 69002b24ab6bSSebastien Roy if (zone_dl_exists(zone, linkid)) 69012b24ab6bSSebastien Roy err = 0; 69022b24ab6bSSebastien Roy zone_rele(zone); 69032b24ab6bSSebastien Roy } 69042b24ab6bSSebastien Roy return (err); 6905f4b3ec61Sdh } 6906f4b3ec61Sdh 6907f4b3ec61Sdh mutex_enter(&zonehash_lock); 6908f4b3ec61Sdh for (zone = list_head(&zone_active); zone != NULL; 6909f4b3ec61Sdh zone = list_next(&zone_active, zone)) { 69102b24ab6bSSebastien Roy if (zone_dl_exists(zone, linkid)) { 69112b24ab6bSSebastien Roy *zoneidp = zone->zone_id; 69122b24ab6bSSebastien Roy err = 0; 69132b24ab6bSSebastien Roy break; 6914f4b3ec61Sdh } 6915f4b3ec61Sdh } 6916f4b3ec61Sdh mutex_exit(&zonehash_lock); 69172b24ab6bSSebastien Roy return (err); 6918f4b3ec61Sdh } 6919f4b3ec61Sdh 6920f4b3ec61Sdh /* 69212b24ab6bSSebastien Roy * Get the list of datalink IDs assigned to a zone. 69222b24ab6bSSebastien Roy * 69232b24ab6bSSebastien Roy * On input, *nump is the number of datalink IDs that can fit in the supplied 69242b24ab6bSSebastien Roy * idarray. Upon return, *nump is either set to the number of datalink IDs 69252b24ab6bSSebastien Roy * that were placed in the array if the array was large enough, or to the 69262b24ab6bSSebastien Roy * number of datalink IDs that the function needs to place in the array if the 69272b24ab6bSSebastien Roy * array is too small. 6928f4b3ec61Sdh */ 6929f4b3ec61Sdh static int 69302b24ab6bSSebastien Roy zone_list_datalink(zoneid_t zoneid, int *nump, datalink_id_t *idarray) 6931f4b3ec61Sdh { 69322b24ab6bSSebastien Roy uint_t num, dlcount; 6933f4b3ec61Sdh zone_t *zone; 69342b24ab6bSSebastien Roy zone_dl_t *zdl; 69352b24ab6bSSebastien Roy datalink_id_t *idptr = idarray; 6936f4b3ec61Sdh 6937f4b3ec61Sdh if (copyin(nump, &dlcount, sizeof (dlcount)) != 0) 6938f4b3ec61Sdh return (set_errno(EFAULT)); 69392b24ab6bSSebastien Roy if ((zone = zone_find_by_id(zoneid)) == NULL) 6940f4b3ec61Sdh return (set_errno(ENXIO)); 6941f4b3ec61Sdh 6942f4b3ec61Sdh num = 0; 6943f4b3ec61Sdh mutex_enter(&zone->zone_lock); 69442b24ab6bSSebastien Roy for (zdl = list_head(&zone->zone_dl_list); zdl != NULL; 69452b24ab6bSSebastien Roy zdl = list_next(&zone->zone_dl_list, zdl)) { 6946f4b3ec61Sdh /* 69472b24ab6bSSebastien Roy * If the list is bigger than what the caller supplied, just 69482b24ab6bSSebastien Roy * count, don't do copyout. 6949f4b3ec61Sdh */ 6950f4b3ec61Sdh if (++num > dlcount) 6951f4b3ec61Sdh continue; 69522b24ab6bSSebastien Roy if (copyout(&zdl->zdl_id, idptr, sizeof (*idptr)) != 0) { 6953f4b3ec61Sdh mutex_exit(&zone->zone_lock); 6954f4b3ec61Sdh zone_rele(zone); 6955f4b3ec61Sdh return (set_errno(EFAULT)); 6956f4b3ec61Sdh } 69572b24ab6bSSebastien Roy idptr++; 6958f4b3ec61Sdh } 6959f4b3ec61Sdh mutex_exit(&zone->zone_lock); 6960f4b3ec61Sdh zone_rele(zone); 6961f4b3ec61Sdh 6962f4b3ec61Sdh /* Increased or decreased, caller should be notified. */ 6963f4b3ec61Sdh if (num != dlcount) { 69642b24ab6bSSebastien Roy if (copyout(&num, nump, sizeof (num)) != 0) 6965f4b3ec61Sdh return (set_errno(EFAULT)); 6966f4b3ec61Sdh } 6967f4b3ec61Sdh return (0); 6968f4b3ec61Sdh } 6969f4b3ec61Sdh 6970f4b3ec61Sdh /* 6971f4b3ec61Sdh * Public interface for looking up a zone by zoneid. It's a customized version 6972bd41d0a8Snordmark * for netstack_zone_create(). It can only be called from the zsd create 6973bd41d0a8Snordmark * callbacks, since it doesn't have reference on the zone structure hence if 6974bd41d0a8Snordmark * it is called elsewhere the zone could disappear after the zonehash_lock 6975bd41d0a8Snordmark * is dropped. 6976bd41d0a8Snordmark * 6977bd41d0a8Snordmark * Furthermore it 6978bd41d0a8Snordmark * 1. Doesn't check the status of the zone. 6979bd41d0a8Snordmark * 2. It will be called even before zone_init is called, in that case the 6980f4b3ec61Sdh * address of zone0 is returned directly, and netstack_zone_create() 6981f4b3ec61Sdh * will only assign a value to zone0.zone_netstack, won't break anything. 6982bd41d0a8Snordmark * 3. Returns without the zone being held. 6983f4b3ec61Sdh */ 6984f4b3ec61Sdh zone_t * 6985f4b3ec61Sdh zone_find_by_id_nolock(zoneid_t zoneid) 6986f4b3ec61Sdh { 6987bd41d0a8Snordmark zone_t *zone; 6988f4b3ec61Sdh 6989bd41d0a8Snordmark mutex_enter(&zonehash_lock); 6990f4b3ec61Sdh if (zonehashbyid == NULL) 6991bd41d0a8Snordmark zone = &zone0; 6992f4b3ec61Sdh else 6993bd41d0a8Snordmark zone = zone_find_all_by_id(zoneid); 6994bd41d0a8Snordmark mutex_exit(&zonehash_lock); 6995bd41d0a8Snordmark return (zone); 6996f4b3ec61Sdh } 6997d62bc4baSyz 6998d62bc4baSyz /* 6999d62bc4baSyz * Walk the datalinks for a given zone 7000d62bc4baSyz */ 7001d62bc4baSyz int 70022b24ab6bSSebastien Roy zone_datalink_walk(zoneid_t zoneid, int (*cb)(datalink_id_t, void *), 70032b24ab6bSSebastien Roy void *data) 7004d62bc4baSyz { 70052b24ab6bSSebastien Roy zone_t *zone; 70062b24ab6bSSebastien Roy zone_dl_t *zdl; 70072b24ab6bSSebastien Roy datalink_id_t *idarray; 70082b24ab6bSSebastien Roy uint_t idcount = 0; 70092b24ab6bSSebastien Roy int i, ret = 0; 7010d62bc4baSyz 7011d62bc4baSyz if ((zone = zone_find_by_id(zoneid)) == NULL) 7012d62bc4baSyz return (ENOENT); 7013d62bc4baSyz 70142b24ab6bSSebastien Roy /* 70152b24ab6bSSebastien Roy * We first build an array of linkid's so that we can walk these and 70162b24ab6bSSebastien Roy * execute the callback with the zone_lock dropped. 70172b24ab6bSSebastien Roy */ 7018d62bc4baSyz mutex_enter(&zone->zone_lock); 70192b24ab6bSSebastien Roy for (zdl = list_head(&zone->zone_dl_list); zdl != NULL; 70202b24ab6bSSebastien Roy zdl = list_next(&zone->zone_dl_list, zdl)) { 70212b24ab6bSSebastien Roy idcount++; 7022d62bc4baSyz } 70232b24ab6bSSebastien Roy 70242b24ab6bSSebastien Roy if (idcount == 0) { 70252b24ab6bSSebastien Roy mutex_exit(&zone->zone_lock); 70262b24ab6bSSebastien Roy zone_rele(zone); 70272b24ab6bSSebastien Roy return (0); 70282b24ab6bSSebastien Roy } 70292b24ab6bSSebastien Roy 70302b24ab6bSSebastien Roy idarray = kmem_alloc(sizeof (datalink_id_t) * idcount, KM_NOSLEEP); 70312b24ab6bSSebastien Roy if (idarray == NULL) { 70322b24ab6bSSebastien Roy mutex_exit(&zone->zone_lock); 70332b24ab6bSSebastien Roy zone_rele(zone); 70342b24ab6bSSebastien Roy return (ENOMEM); 70352b24ab6bSSebastien Roy } 70362b24ab6bSSebastien Roy 70372b24ab6bSSebastien Roy for (i = 0, zdl = list_head(&zone->zone_dl_list); zdl != NULL; 70382b24ab6bSSebastien Roy i++, zdl = list_next(&zone->zone_dl_list, zdl)) { 70392b24ab6bSSebastien Roy idarray[i] = zdl->zdl_id; 70402b24ab6bSSebastien Roy } 70412b24ab6bSSebastien Roy 7042d62bc4baSyz mutex_exit(&zone->zone_lock); 70432b24ab6bSSebastien Roy 70442b24ab6bSSebastien Roy for (i = 0; i < idcount && ret == 0; i++) { 70452b24ab6bSSebastien Roy if ((ret = (*cb)(idarray[i], data)) != 0) 70462b24ab6bSSebastien Roy break; 70472b24ab6bSSebastien Roy } 70482b24ab6bSSebastien Roy 7049d62bc4baSyz zone_rele(zone); 70502b24ab6bSSebastien Roy kmem_free(idarray, sizeof (datalink_id_t) * idcount); 7051d62bc4baSyz return (ret); 7052d62bc4baSyz } 7053550b6e40SSowmini Varadhan 7054550b6e40SSowmini Varadhan static char * 7055550b6e40SSowmini Varadhan zone_net_type2name(int type) 7056550b6e40SSowmini Varadhan { 7057550b6e40SSowmini Varadhan switch (type) { 7058550b6e40SSowmini Varadhan case ZONE_NETWORK_ADDRESS: 7059550b6e40SSowmini Varadhan return (ZONE_NET_ADDRNAME); 7060550b6e40SSowmini Varadhan case ZONE_NETWORK_DEFROUTER: 7061550b6e40SSowmini Varadhan return (ZONE_NET_RTRNAME); 7062550b6e40SSowmini Varadhan default: 7063550b6e40SSowmini Varadhan return (NULL); 7064550b6e40SSowmini Varadhan } 7065550b6e40SSowmini Varadhan } 7066550b6e40SSowmini Varadhan 7067550b6e40SSowmini Varadhan static int 7068550b6e40SSowmini Varadhan zone_set_network(zoneid_t zoneid, zone_net_data_t *znbuf) 7069550b6e40SSowmini Varadhan { 7070550b6e40SSowmini Varadhan zone_t *zone; 7071550b6e40SSowmini Varadhan zone_dl_t *zdl; 7072550b6e40SSowmini Varadhan nvlist_t *nvl; 7073550b6e40SSowmini Varadhan int err = 0; 7074550b6e40SSowmini Varadhan uint8_t *new = NULL; 7075550b6e40SSowmini Varadhan char *nvname; 7076550b6e40SSowmini Varadhan int bufsize; 7077550b6e40SSowmini Varadhan datalink_id_t linkid = znbuf->zn_linkid; 7078550b6e40SSowmini Varadhan 7079550b6e40SSowmini Varadhan if (secpolicy_zone_config(CRED()) != 0) 7080550b6e40SSowmini Varadhan return (set_errno(EPERM)); 7081550b6e40SSowmini Varadhan 7082550b6e40SSowmini Varadhan if (zoneid == GLOBAL_ZONEID) 7083550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7084550b6e40SSowmini Varadhan 7085550b6e40SSowmini Varadhan nvname = zone_net_type2name(znbuf->zn_type); 7086550b6e40SSowmini Varadhan bufsize = znbuf->zn_len; 7087550b6e40SSowmini Varadhan new = znbuf->zn_val; 7088550b6e40SSowmini Varadhan if (nvname == NULL) 7089550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7090550b6e40SSowmini Varadhan 7091550b6e40SSowmini Varadhan if ((zone = zone_find_by_id(zoneid)) == NULL) { 7092550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7093550b6e40SSowmini Varadhan } 7094550b6e40SSowmini Varadhan 7095550b6e40SSowmini Varadhan mutex_enter(&zone->zone_lock); 7096550b6e40SSowmini Varadhan if ((zdl = zone_find_dl(zone, linkid)) == NULL) { 7097550b6e40SSowmini Varadhan err = ENXIO; 7098550b6e40SSowmini Varadhan goto done; 7099550b6e40SSowmini Varadhan } 7100550b6e40SSowmini Varadhan if ((nvl = zdl->zdl_net) == NULL) { 7101550b6e40SSowmini Varadhan if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_SLEEP)) { 7102550b6e40SSowmini Varadhan err = ENOMEM; 7103550b6e40SSowmini Varadhan goto done; 7104550b6e40SSowmini Varadhan } else { 7105550b6e40SSowmini Varadhan zdl->zdl_net = nvl; 7106550b6e40SSowmini Varadhan } 7107550b6e40SSowmini Varadhan } 7108550b6e40SSowmini Varadhan if (nvlist_exists(nvl, nvname)) { 7109550b6e40SSowmini Varadhan err = EINVAL; 7110550b6e40SSowmini Varadhan goto done; 7111550b6e40SSowmini Varadhan } 7112550b6e40SSowmini Varadhan err = nvlist_add_uint8_array(nvl, nvname, new, bufsize); 7113550b6e40SSowmini Varadhan ASSERT(err == 0); 7114550b6e40SSowmini Varadhan done: 7115550b6e40SSowmini Varadhan mutex_exit(&zone->zone_lock); 7116550b6e40SSowmini Varadhan zone_rele(zone); 7117550b6e40SSowmini Varadhan if (err != 0) 7118550b6e40SSowmini Varadhan return (set_errno(err)); 7119550b6e40SSowmini Varadhan else 7120550b6e40SSowmini Varadhan return (0); 7121550b6e40SSowmini Varadhan } 7122550b6e40SSowmini Varadhan 7123550b6e40SSowmini Varadhan static int 7124550b6e40SSowmini Varadhan zone_get_network(zoneid_t zoneid, zone_net_data_t *znbuf) 7125550b6e40SSowmini Varadhan { 7126550b6e40SSowmini Varadhan zone_t *zone; 7127550b6e40SSowmini Varadhan zone_dl_t *zdl; 7128550b6e40SSowmini Varadhan nvlist_t *nvl; 7129550b6e40SSowmini Varadhan uint8_t *ptr; 7130550b6e40SSowmini Varadhan uint_t psize; 7131550b6e40SSowmini Varadhan int err = 0; 7132550b6e40SSowmini Varadhan char *nvname; 7133550b6e40SSowmini Varadhan int bufsize; 7134550b6e40SSowmini Varadhan void *buf; 7135550b6e40SSowmini Varadhan datalink_id_t linkid = znbuf->zn_linkid; 7136550b6e40SSowmini Varadhan 7137550b6e40SSowmini Varadhan if (zoneid == GLOBAL_ZONEID) 7138550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7139550b6e40SSowmini Varadhan 7140550b6e40SSowmini Varadhan nvname = zone_net_type2name(znbuf->zn_type); 7141550b6e40SSowmini Varadhan bufsize = znbuf->zn_len; 7142550b6e40SSowmini Varadhan buf = znbuf->zn_val; 7143550b6e40SSowmini Varadhan 7144550b6e40SSowmini Varadhan if (nvname == NULL) 7145550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7146550b6e40SSowmini Varadhan if ((zone = zone_find_by_id(zoneid)) == NULL) 7147550b6e40SSowmini Varadhan return (set_errno(EINVAL)); 7148550b6e40SSowmini Varadhan 7149550b6e40SSowmini Varadhan mutex_enter(&zone->zone_lock); 7150550b6e40SSowmini Varadhan if ((zdl = zone_find_dl(zone, linkid)) == NULL) { 7151550b6e40SSowmini Varadhan err = ENXIO; 7152550b6e40SSowmini Varadhan goto done; 7153550b6e40SSowmini Varadhan } 7154550b6e40SSowmini Varadhan if ((nvl = zdl->zdl_net) == NULL || !nvlist_exists(nvl, nvname)) { 7155550b6e40SSowmini Varadhan err = ENOENT; 7156550b6e40SSowmini Varadhan goto done; 7157550b6e40SSowmini Varadhan } 7158550b6e40SSowmini Varadhan err = nvlist_lookup_uint8_array(nvl, nvname, &ptr, &psize); 7159550b6e40SSowmini Varadhan ASSERT(err == 0); 7160550b6e40SSowmini Varadhan 7161550b6e40SSowmini Varadhan if (psize > bufsize) { 7162550b6e40SSowmini Varadhan err = ENOBUFS; 7163550b6e40SSowmini Varadhan goto done; 7164550b6e40SSowmini Varadhan } 7165550b6e40SSowmini Varadhan znbuf->zn_len = psize; 7166550b6e40SSowmini Varadhan bcopy(ptr, buf, psize); 7167550b6e40SSowmini Varadhan done: 7168550b6e40SSowmini Varadhan mutex_exit(&zone->zone_lock); 7169550b6e40SSowmini Varadhan zone_rele(zone); 7170550b6e40SSowmini Varadhan if (err != 0) 7171550b6e40SSowmini Varadhan return (set_errno(err)); 7172550b6e40SSowmini Varadhan else 7173550b6e40SSowmini Varadhan return (0); 7174550b6e40SSowmini Varadhan } 7175