kernel/fs/zfs/sys/spa_impl.h

   1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  23  * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
  24  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  25  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  26  * Copyright 2013 Saso Kiselkov. All rights reserved.
  27  * Copyright (c) 2017 Datto Inc.
  28  */
  29
  30 #ifndef _SYS_SPA_IMPL_H
  31 #define _SYS_SPA_IMPL_H
  32
  33 #include <sys/spa.h>
  34 #include <sys/vdev.h>
  35 #include <sys/vdev_removal.h>
  36 #include <sys/metaslab.h>
  37 #include <sys/dmu.h>
  38 #include <sys/dsl_pool.h>
  39 #include <sys/uberblock_impl.h>
  40 #include <sys/zfs_context.h>
  41 #include <sys/avl.h>
  42 #include <sys/refcount.h>
  43 #include <sys/bplist.h>
  44 #include <sys/bpobj.h>
  45 #include <sys/zfeature.h>
  46 #include <sys/zthr.h>
  47 #include <zfeature_common.h>
  48
  49 #ifdef  __cplusplus
  50 extern "C" {
  51 #endif
  52
  53 typedef struct spa_error_entry {
  54         zbookmark_phys_t        se_bookmark;
  55         char                    *se_name;
  56         avl_node_t              se_avl;
  57 } spa_error_entry_t;
  58
  59 typedef struct spa_history_phys {
  60         uint64_t sh_pool_create_len;    /* ending offset of zpool create */
  61         uint64_t sh_phys_max_off;       /* physical EOF */
  62         uint64_t sh_bof;                /* logical BOF */
  63         uint64_t sh_eof;                /* logical EOF */
  64         uint64_t sh_records_lost;       /* num of records overwritten */
  65 } spa_history_phys_t;
  66
  67 /*
  68  * All members must be uint64_t, for byteswap purposes.
  69  */
  70 typedef struct spa_removing_phys {
  71         uint64_t sr_state; /* dsl_scan_state_t */
  72
  73         /*
  74          * The vdev ID that we most recently attempted to remove,
  75          * or -1 if no removal has been attempted.
  76          */
  77         uint64_t sr_removing_vdev;
  78
  79         /*
  80          * The vdev ID that we most recently successfully removed,
  81          * or -1 if no devices have been removed.
  82          */
  83         uint64_t sr_prev_indirect_vdev;
  84
  85         uint64_t sr_start_time;
  86         uint64_t sr_end_time;
  87
  88         /*
  89          * Note that we can not use the space map's or indirect mapping's
  90          * accounting as a substitute for these values, because we need to
  91          * count frees of not-yet-copied data as though it did the copy.
  92          * Otherwise, we could get into a situation where copied > to_copy,
  93          * or we complete before copied == to_copy.
  94          */
  95         uint64_t sr_to_copy; /* bytes that need to be copied */
  96         uint64_t sr_copied; /* bytes that have been copied or freed */
  97 } spa_removing_phys_t;
  98
  99 /*
 100  * This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
 101  * (with key DMU_POOL_CONDENSING_INDIRECT).  It is present if a condense
 102  * of an indirect vdev's mapping object is in progress.
 103  */
 104 typedef struct spa_condensing_indirect_phys {
 105         /*
 106          * The vdev ID of the indirect vdev whose indirect mapping is
 107          * being condensed.
 108          */
 109         uint64_t        scip_vdev;
 110
 111         /*
 112          * The vdev's old obsolete spacemap.  This spacemap's contents are
 113          * being integrated into the new mapping.
 114          */
 115         uint64_t        scip_prev_obsolete_sm_object;
 116
 117         /*
 118          * The new mapping object that is being created.
 119          */
 120         uint64_t        scip_next_mapping_object;
 121 } spa_condensing_indirect_phys_t;
 122
 123 struct spa_aux_vdev {
 124         uint64_t        sav_object;             /* MOS object for device list */
 125         nvlist_t        *sav_config;            /* cached device config */
 126         vdev_t          **sav_vdevs;            /* devices */
 127         int             sav_count;              /* number devices */
 128         boolean_t       sav_sync;               /* sync the device list */
 129         nvlist_t        **sav_pending;          /* pending device additions */
 130         uint_t          sav_npending;           /* # pending devices */
 131 };
 132
 133 typedef struct spa_config_lock {
 134         kmutex_t        scl_lock;
 135         kthread_t       *scl_writer;
 136         int             scl_write_wanted;
 137         kcondvar_t      scl_cv;
 138         refcount_t      scl_count;
 139 } spa_config_lock_t;
 140
 141 typedef struct spa_config_dirent {
 142         list_node_t     scd_link;
 143         char            *scd_path;
 144 } spa_config_dirent_t;
 145
 146 typedef enum zio_taskq_type {
 147         ZIO_TASKQ_ISSUE = 0,
 148         ZIO_TASKQ_ISSUE_HIGH,
 149         ZIO_TASKQ_INTERRUPT,
 150         ZIO_TASKQ_INTERRUPT_HIGH,
 151         ZIO_TASKQ_TYPES
 152 } zio_taskq_type_t;
 153
 154 /*
 155  * State machine for the zpool-poolname process.  The states transitions
 156  * are done as follows:
 157  *
 158  *      From               To                   Routine
 159  *      PROC_NONE       -> PROC_CREATED         spa_activate()
 160  *      PROC_CREATED    -> PROC_ACTIVE          spa_thread()
 161  *      PROC_ACTIVE     -> PROC_DEACTIVATE      spa_deactivate()
 162  *      PROC_DEACTIVATE -> PROC_GONE            spa_thread()
 163  *      PROC_GONE       -> PROC_NONE            spa_deactivate()
 164  */
 165 typedef enum spa_proc_state {
 166         SPA_PROC_NONE,          /* spa_proc = &p0, no process created */
 167         SPA_PROC_CREATED,       /* spa_activate() has proc, is waiting */
 168         SPA_PROC_ACTIVE,        /* taskqs created, spa_proc set */
 169         SPA_PROC_DEACTIVATE,    /* spa_deactivate() requests process exit */
 170         SPA_PROC_GONE           /* spa_thread() is exiting, spa_proc = &p0 */
 171 } spa_proc_state_t;
 172
 173 typedef struct spa_taskqs {
 174         uint_t stqs_count;
 175         taskq_t **stqs_taskq;
 176 } spa_taskqs_t;
 177
 178 typedef enum spa_all_vdev_zap_action {
 179         AVZ_ACTION_NONE = 0,
 180         AVZ_ACTION_DESTROY,     /* Destroy all per-vdev ZAPs and the AVZ. */
 181         AVZ_ACTION_REBUILD,     /* Populate the new AVZ, see spa_avz_rebuild */
 182         AVZ_ACTION_INITIALIZE
 183 } spa_avz_action_t;
 184
 185 typedef enum spa_config_source {
 186         SPA_CONFIG_SRC_NONE = 0,
 187         SPA_CONFIG_SRC_SCAN,            /* scan of path (default: /dev/dsk) */
 188         SPA_CONFIG_SRC_CACHEFILE,       /* any cachefile */
 189         SPA_CONFIG_SRC_TRYIMPORT,       /* returned from call to tryimport */
 190         SPA_CONFIG_SRC_SPLIT,           /* new pool in a pool split */
 191         SPA_CONFIG_SRC_MOS              /* MOS, but not always from right txg */
 192 } spa_config_source_t;
 193
 194 struct spa {
 195         /*
 196          * Fields protected by spa_namespace_lock.
 197          */
 198         char            spa_name[ZFS_MAX_DATASET_NAME_LEN];     /* pool name */
 199         char            *spa_comment;           /* comment */
 200         avl_node_t      spa_avl;                /* node in spa_namespace_avl */
 201         nvlist_t        *spa_config;            /* last synced config */
 202         nvlist_t        *spa_config_syncing;    /* currently syncing config */
 203         nvlist_t        *spa_config_splitting;  /* config for splitting */
 204         nvlist_t        *spa_load_info;         /* info and errors from load */
 205         uint64_t        spa_config_txg;         /* txg of last config change */
 206         int             spa_sync_pass;          /* iterate-to-convergence */
 207         pool_state_t    spa_state;              /* pool state */
 208         int             spa_inject_ref;         /* injection references */
 209         uint8_t         spa_sync_on;            /* sync threads are running */
 210         spa_load_state_t spa_load_state;        /* current load operation */
 211         boolean_t       spa_indirect_vdevs_loaded; /* mappings loaded? */
 212         boolean_t       spa_trust_config;       /* do we trust vdev tree? */
 213         spa_config_source_t spa_config_source;  /* where config comes from? */
 214         uint64_t        spa_import_flags;       /* import specific flags */
 215         spa_taskqs_t    spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
 216         dsl_pool_t      *spa_dsl_pool;
 217         boolean_t       spa_is_initializing;    /* true while opening pool */
 218         metaslab_class_t *spa_normal_class;     /* normal data class */
 219         metaslab_class_t *spa_log_class;        /* intent log data class */
 220         uint64_t        spa_first_txg;          /* first txg after spa_open() */
 221         uint64_t        spa_final_txg;          /* txg of export/destroy */
 222         uint64_t        spa_freeze_txg;         /* freeze pool at this txg */
 223         uint64_t        spa_load_max_txg;       /* best initial ub_txg */
 224         uint64_t        spa_claim_max_txg;      /* highest claimed birth txg */
 225         timespec_t      spa_loaded_ts;          /* 1st successful open time */
 226         objset_t        *spa_meta_objset;       /* copy of dp->dp_meta_objset */
 227         kmutex_t        spa_evicting_os_lock;   /* Evicting objset list lock */
 228         list_t          spa_evicting_os_list;   /* Objsets being evicted. */
 229         kcondvar_t      spa_evicting_os_cv;     /* Objset Eviction Completion */
 230         txg_list_t      spa_vdev_txg_list;      /* per-txg dirty vdev list */
 231         vdev_t          *spa_root_vdev;         /* top-level vdev container */
 232         int             spa_min_ashift;         /* of vdevs in normal class */
 233         int             spa_max_ashift;         /* of vdevs in normal class */
 234         uint64_t        spa_config_guid;        /* config pool guid */
 235         uint64_t        spa_load_guid;          /* spa_load initialized guid */
 236         uint64_t        spa_last_synced_guid;   /* last synced guid */
 237         list_t          spa_config_dirty_list;  /* vdevs with dirty config */
 238         list_t          spa_state_dirty_list;   /* vdevs with dirty state */
 239         kmutex_t        spa_alloc_lock;
 240         avl_tree_t      spa_alloc_tree;
 241         spa_aux_vdev_t  spa_spares;             /* hot spares */
 242         spa_aux_vdev_t  spa_l2cache;            /* L2ARC cache devices */
 243         nvlist_t        *spa_label_features;    /* Features for reading MOS */
 244         uint64_t        spa_config_object;      /* MOS object for pool config */
 245         uint64_t        spa_config_generation;  /* config generation number */
 246         uint64_t        spa_syncing_txg;        /* txg currently syncing */
 247         bpobj_t         spa_deferred_bpobj;     /* deferred-free bplist */
 248         bplist_t        spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
 249         zio_cksum_salt_t spa_cksum_salt;        /* secret salt for cksum */
 250         /* checksum context templates */
 251         kmutex_t        spa_cksum_tmpls_lock;
 252         void            *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
 253         uberblock_t     spa_ubsync;             /* last synced uberblock */
 254         uberblock_t     spa_uberblock;          /* current uberblock */
 255         boolean_t       spa_extreme_rewind;     /* rewind past deferred frees */
 256         uint64_t        spa_last_io;            /* lbolt of last non-scan I/O */
 257         kmutex_t        spa_scrub_lock;         /* resilver/scrub lock */
 258         uint64_t        spa_scrub_inflight;     /* in-flight scrub I/Os */
 259         kcondvar_t      spa_scrub_io_cv;        /* scrub I/O completion */
 260         uint8_t         spa_scrub_active;       /* active or suspended? */
 261         uint8_t         spa_scrub_type;         /* type of scrub we're doing */
 262         uint8_t         spa_scrub_finished;     /* indicator to rotate logs */
 263         uint8_t         spa_scrub_started;      /* started since last boot */
 264         uint8_t         spa_scrub_reopen;       /* scrub doing vdev_reopen */
 265         uint64_t        spa_scan_pass_start;    /* start time per pass/reboot */
 266         uint64_t        spa_scan_pass_scrub_pause; /* scrub pause time */
 267         uint64_t        spa_scan_pass_scrub_spent_paused; /* total paused */
 268         uint64_t        spa_scan_pass_exam;     /* examined bytes per pass */
 269         kmutex_t        spa_async_lock;         /* protect async state */
 270         kthread_t       *spa_async_thread;      /* thread doing async task */
 271         int             spa_async_suspended;    /* async tasks suspended */
 272         kcondvar_t      spa_async_cv;           /* wait for thread_exit() */
 273         uint16_t        spa_async_tasks;        /* async task mask */
 274         uint64_t        spa_missing_tvds;       /* unopenable tvds on load */
 275         uint64_t        spa_missing_tvds_allowed; /* allow loading spa? */
 276
 277         spa_removing_phys_t spa_removing_phys;
 278         spa_vdev_removal_t *spa_vdev_removal;
 279
 280         spa_condensing_indirect_phys_t  spa_condensing_indirect_phys;
 281         spa_condensing_indirect_t       *spa_condensing_indirect;
 282         zthr_t          *spa_condense_zthr;     /* zthr doing condense. */
 283
 284         char            *spa_root;              /* alternate root directory */
 285         uint64_t        spa_ena;                /* spa-wide ereport ENA */
 286         int             spa_last_open_failed;   /* error if last open failed */
 287         uint64_t        spa_last_ubsync_txg;    /* "best" uberblock txg */
 288         uint64_t        spa_last_ubsync_txg_ts; /* timestamp from that ub */
 289         uint64_t        spa_load_txg;           /* ub txg that loaded */
 290         uint64_t        spa_load_txg_ts;        /* timestamp from that ub */
 291         uint64_t        spa_load_meta_errors;   /* verify metadata err count */
 292         uint64_t        spa_load_data_errors;   /* verify data err count */
 293         uint64_t        spa_verify_min_txg;     /* start txg of verify scrub */
 294         kmutex_t        spa_errlog_lock;        /* error log lock */
 295         uint64_t        spa_errlog_last;        /* last error log object */
 296         uint64_t        spa_errlog_scrub;       /* scrub error log object */
 297         kmutex_t        spa_errlist_lock;       /* error list/ereport lock */
 298         avl_tree_t      spa_errlist_last;       /* last error list */
 299         avl_tree_t      spa_errlist_scrub;      /* scrub error list */
 300         uint64_t        spa_deflate;            /* should we deflate? */
 301         uint64_t        spa_history;            /* history object */
 302         kmutex_t        spa_history_lock;       /* history lock */
 303         vdev_t          *spa_pending_vdev;      /* pending vdev additions */
 304         kmutex_t        spa_props_lock;         /* property lock */
 305         uint64_t        spa_pool_props_object;  /* object for properties */
 306         uint64_t        spa_bootfs;             /* default boot filesystem */
 307         uint64_t        spa_failmode;           /* failure mode for the pool */
 308         uint64_t        spa_delegation;         /* delegation on/off */
 309         list_t          spa_config_list;        /* previous cache file(s) */
 310         /* per-CPU array of root of async I/O: */
 311         zio_t           **spa_async_zio_root;
 312         zio_t           *spa_suspend_zio_root;  /* root of all suspended I/O */
 313         zio_t           *spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */
 314         kmutex_t        spa_suspend_lock;       /* protects suspend_zio_root */
 315         kcondvar_t      spa_suspend_cv;         /* notification of resume */
 316         uint8_t         spa_suspended;          /* pool is suspended */
 317         uint8_t         spa_claiming;           /* pool is doing zil_claim() */
 318         boolean_t       spa_debug;              /* debug enabled? */
 319         boolean_t       spa_is_root;            /* pool is root */
 320         int             spa_minref;             /* num refs when first opened */
 321         int             spa_mode;               /* FREAD | FWRITE */
 322         spa_log_state_t spa_log_state;          /* log state */
 323         uint64_t        spa_autoexpand;         /* lun expansion on/off */
 324         uint64_t        spa_bootsize;           /* efi system partition size */
 325         ddt_t           *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */
 326         uint64_t        spa_ddt_stat_object;    /* DDT statistics */
 327         uint64_t        spa_dedup_ditto;        /* dedup ditto threshold */
 328         uint64_t        spa_dedup_checksum;     /* default dedup checksum */
 329         uint64_t        spa_dspace;             /* dspace in normal class */
 330         kmutex_t        spa_vdev_top_lock;      /* dueling offline/remove */
 331         kmutex_t        spa_proc_lock;          /* protects spa_proc* */
 332         kcondvar_t      spa_proc_cv;            /* spa_proc_state transitions */
 333         spa_proc_state_t spa_proc_state;        /* see definition */
 334         struct proc     *spa_proc;              /* "zpool-poolname" process */
 335         uint64_t        spa_did;                /* if procp != p0, did of t1 */
 336         boolean_t       spa_autoreplace;        /* autoreplace set in open */
 337         int             spa_vdev_locks;         /* locks grabbed */
 338         uint64_t        spa_creation_version;   /* version at pool creation */
 339         uint64_t        spa_prev_software_version; /* See ub_software_version */
 340         uint64_t        spa_feat_for_write_obj; /* required to write to pool */
 341         uint64_t        spa_feat_for_read_obj;  /* required to read from pool */
 342         uint64_t        spa_feat_desc_obj;      /* Feature descriptions */
 343         uint64_t        spa_feat_enabled_txg_obj; /* Feature enabled txg */
 344         /* cache feature refcounts */
 345         uint64_t        spa_feat_refcount_cache[SPA_FEATURES];
 346         cyclic_id_t     spa_deadman_cycid;      /* cyclic id */
 347         uint64_t        spa_deadman_calls;      /* number of deadman calls */
 348         hrtime_t        spa_sync_starttime;     /* starting time fo spa_sync */
 349         uint64_t        spa_deadman_synctime;   /* deadman expiration timer */
 350         uint64_t        spa_all_vdev_zaps;      /* ZAP of per-vd ZAP obj #s */
 351         spa_avz_action_t        spa_avz_action; /* destroy/rebuild AVZ? */
 352
 353         /*
 354          * spa_iokstat_lock protects spa_iokstat and
 355          * spa_queue_stats[].
 356          */
 357         kmutex_t        spa_iokstat_lock;
 358         struct kstat    *spa_iokstat;           /* kstat of io to this pool */
 359         struct {
 360                 int spa_active;
 361                 int spa_queued;
 362         } spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE];
 363
 364         hrtime_t        spa_ccw_fail_time;      /* Conf cache write fail time */
 365
 366         /*
 367          * spa_refcount & spa_config_lock must be the last elements
 368          * because refcount_t changes size based on compilation options.
 369          * In order for the MDB module to function correctly, the other
 370          * fields must remain in the same location.
 371          */
 372         spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
 373         refcount_t      spa_refcount;           /* number of opens */
 374 };
 375
 376 extern const char *spa_config_path;
 377
 378 extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
 379     task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent);
 380 extern void spa_load_spares(spa_t *spa);
 381 extern void spa_load_l2cache(spa_t *spa);
 382
 383 #ifdef  __cplusplus
 384 }
 385 #endif
 386
 387 #endif  /* _SYS_SPA_IMPL_H */