1fa9e4066Sahrens /* 2fa9e4066Sahrens * CDDL HEADER START 3fa9e4066Sahrens * 4fa9e4066Sahrens * The contents of this file are subject to the terms of the 5441d80aaSlling * Common Development and Distribution License (the "License"). 6441d80aaSlling * You may not use this file except in compliance with the License. 7fa9e4066Sahrens * 8fa9e4066Sahrens * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9fa9e4066Sahrens * or http://www.opensolaris.org/os/licensing. 10fa9e4066Sahrens * See the License for the specific language governing permissions 11fa9e4066Sahrens * and limitations under the License. 12fa9e4066Sahrens * 13fa9e4066Sahrens * When distributing Covered Code, include this CDDL HEADER in each 14fa9e4066Sahrens * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15fa9e4066Sahrens * If applicable, add the following below this CDDL HEADER, with the 16fa9e4066Sahrens * fields enclosed by brackets "[]" replaced with your own identifying 17fa9e4066Sahrens * information: Portions Copyright [yyyy] [name of copyright owner] 18fa9e4066Sahrens * 19fa9e4066Sahrens * CDDL HEADER END 20fa9e4066Sahrens */ 21fa9e4066Sahrens /* 2298d1cbfeSGeorge Wilson * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. 23f78cdc34SPaul Dagnelie * Copyright (c) 2011, 2018 by Delphix. All rights reserved. 24fa9e4066Sahrens */ 25fa9e4066Sahrens 26fa9e4066Sahrens #ifndef _SYS_VDEV_IMPL_H 27fa9e4066Sahrens #define _SYS_VDEV_IMPL_H 28fa9e4066Sahrens 29fa9e4066Sahrens #include <sys/avl.h> 305cabbc6bSPrashanth Sreenivasa #include <sys/bpobj.h> 31fa9e4066Sahrens #include <sys/dmu.h> 32fa9e4066Sahrens #include <sys/metaslab.h> 33fa9e4066Sahrens #include <sys/nvpair.h> 34fa9e4066Sahrens #include <sys/space_map.h> 35fa9e4066Sahrens #include <sys/vdev.h> 36fa9e4066Sahrens #include <sys/dkio.h> 37fa9e4066Sahrens #include <sys/uberblock_impl.h> 385cabbc6bSPrashanth Sreenivasa #include <sys/vdev_indirect_mapping.h> 395cabbc6bSPrashanth Sreenivasa #include <sys/vdev_indirect_births.h> 405cabbc6bSPrashanth Sreenivasa #include <sys/vdev_removal.h> 41fa9e4066Sahrens 42fa9e4066Sahrens #ifdef __cplusplus 43fa9e4066Sahrens extern "C" { 44fa9e4066Sahrens #endif 45fa9e4066Sahrens 46fa9e4066Sahrens /* 47fa9e4066Sahrens * Virtual device descriptors. 48fa9e4066Sahrens * 49fa9e4066Sahrens * All storage pool operations go through the virtual device framework, 50fa9e4066Sahrens * which provides data replication and I/O scheduling. 51fa9e4066Sahrens */ 52fa9e4066Sahrens 53fa9e4066Sahrens /* 54fa9e4066Sahrens * Forward declarations that lots of things need. 55fa9e4066Sahrens */ 56fa9e4066Sahrens typedef struct vdev_queue vdev_queue_t; 57fa9e4066Sahrens typedef struct vdev_cache vdev_cache_t; 58fa9e4066Sahrens typedef struct vdev_cache_entry vdev_cache_entry_t; 59770499e1SDan Kimmel struct abd; 60fa9e4066Sahrens 610f7643c7SGeorge Wilson extern int zfs_vdev_queue_depth_pct; 62f78cdc34SPaul Dagnelie extern int zfs_vdev_def_queue_depth; 630f7643c7SGeorge Wilson extern uint32_t zfs_vdev_async_write_max_active; 640f7643c7SGeorge Wilson 65fa9e4066Sahrens /* 66fa9e4066Sahrens * Virtual device operations 67fa9e4066Sahrens */ 684263d13fSGeorge Wilson typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size, 694263d13fSGeorge Wilson uint64_t *ashift); 70fa9e4066Sahrens typedef void vdev_close_func_t(vdev_t *vd); 71fa9e4066Sahrens typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize); 72738f37bcSGeorge Wilson typedef void vdev_io_start_func_t(zio_t *zio); 73e14bb325SJeff Bonwick typedef void vdev_io_done_func_t(zio_t *zio); 74fa9e4066Sahrens typedef void vdev_state_change_func_t(vdev_t *vd, int, int); 75dcba9f3fSGeorge Wilson typedef void vdev_hold_func_t(vdev_t *vd); 76dcba9f3fSGeorge Wilson typedef void vdev_rele_func_t(vdev_t *vd); 77fa9e4066Sahrens 785cabbc6bSPrashanth Sreenivasa typedef void vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd, 795cabbc6bSPrashanth Sreenivasa uint64_t offset, uint64_t size, void *arg); 805cabbc6bSPrashanth Sreenivasa typedef void vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size, 815cabbc6bSPrashanth Sreenivasa vdev_remap_cb_t callback, void *arg); 82*094e47e9SGeorge Wilson /* 83*094e47e9SGeorge Wilson * Given a target vdev, translates the logical range "in" to the physical 84*094e47e9SGeorge Wilson * range "res" 85*094e47e9SGeorge Wilson */ 86*094e47e9SGeorge Wilson typedef void vdev_xlation_func_t(vdev_t *cvd, const range_seg_t *in, 87*094e47e9SGeorge Wilson range_seg_t *res); 885cabbc6bSPrashanth Sreenivasa 89fa9e4066Sahrens typedef struct vdev_ops { 90fa9e4066Sahrens vdev_open_func_t *vdev_op_open; 91fa9e4066Sahrens vdev_close_func_t *vdev_op_close; 92fa9e4066Sahrens vdev_asize_func_t *vdev_op_asize; 93fa9e4066Sahrens vdev_io_start_func_t *vdev_op_io_start; 94fa9e4066Sahrens vdev_io_done_func_t *vdev_op_io_done; 95fa9e4066Sahrens vdev_state_change_func_t *vdev_op_state_change; 96dcba9f3fSGeorge Wilson vdev_hold_func_t *vdev_op_hold; 97dcba9f3fSGeorge Wilson vdev_rele_func_t *vdev_op_rele; 985cabbc6bSPrashanth Sreenivasa vdev_remap_func_t *vdev_op_remap; 99*094e47e9SGeorge Wilson /* 100*094e47e9SGeorge Wilson * For translating ranges from non-leaf vdevs (e.g. raidz) to leaves. 101*094e47e9SGeorge Wilson * Used when initializing vdevs. Isn't used by leaf ops. 102*094e47e9SGeorge Wilson */ 103*094e47e9SGeorge Wilson vdev_xlation_func_t *vdev_op_xlate; 104fa9e4066Sahrens char vdev_op_type[16]; 105fa9e4066Sahrens boolean_t vdev_op_leaf; 106fa9e4066Sahrens } vdev_ops_t; 107fa9e4066Sahrens 108fa9e4066Sahrens /* 109fa9e4066Sahrens * Virtual device properties 110fa9e4066Sahrens */ 111fa9e4066Sahrens struct vdev_cache_entry { 112770499e1SDan Kimmel struct abd *ve_abd; 113fa9e4066Sahrens uint64_t ve_offset; 114fa9e4066Sahrens uint64_t ve_lastused; 115fa9e4066Sahrens avl_node_t ve_offset_node; 116fa9e4066Sahrens avl_node_t ve_lastused_node; 117fa9e4066Sahrens uint32_t ve_hits; 118fa9e4066Sahrens uint16_t ve_missed_update; 119fa9e4066Sahrens zio_t *ve_fill_io; 120fa9e4066Sahrens }; 121fa9e4066Sahrens 122fa9e4066Sahrens struct vdev_cache { 123fa9e4066Sahrens avl_tree_t vc_offset_tree; 124fa9e4066Sahrens avl_tree_t vc_lastused_tree; 125fa9e4066Sahrens kmutex_t vc_lock; 126fa9e4066Sahrens }; 127fa9e4066Sahrens 12869962b56SMatthew Ahrens typedef struct vdev_queue_class { 12969962b56SMatthew Ahrens uint32_t vqc_active; 13069962b56SMatthew Ahrens 13169962b56SMatthew Ahrens /* 13269962b56SMatthew Ahrens * Sorted by offset or timestamp, depending on if the queue is 13369962b56SMatthew Ahrens * LBA-ordered vs FIFO. 13469962b56SMatthew Ahrens */ 13569962b56SMatthew Ahrens avl_tree_t vqc_queued_tree; 13669962b56SMatthew Ahrens } vdev_queue_class_t; 13769962b56SMatthew Ahrens 138fa9e4066Sahrens struct vdev_queue { 13969962b56SMatthew Ahrens vdev_t *vq_vdev; 14069962b56SMatthew Ahrens vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE]; 14169962b56SMatthew Ahrens avl_tree_t vq_active_tree; 142fe319232SJustin T. Gibbs avl_tree_t vq_read_offset_tree; 143fe319232SJustin T. Gibbs avl_tree_t vq_write_offset_tree; 14469962b56SMatthew Ahrens uint64_t vq_last_offset; 14569962b56SMatthew Ahrens hrtime_t vq_io_complete_ts; /* time last i/o completed */ 146fa9e4066Sahrens kmutex_t vq_lock; 147fa9e4066Sahrens }; 148fa9e4066Sahrens 1495cabbc6bSPrashanth Sreenivasa /* 1505cabbc6bSPrashanth Sreenivasa * On-disk indirect vdev state. 1515cabbc6bSPrashanth Sreenivasa * 1525cabbc6bSPrashanth Sreenivasa * An indirect vdev is described exclusively in the MOS config of a pool. 1535cabbc6bSPrashanth Sreenivasa * The config for an indirect vdev includes several fields, which are 1545cabbc6bSPrashanth Sreenivasa * accessed in memory by a vdev_indirect_config_t. 1555cabbc6bSPrashanth Sreenivasa */ 1565cabbc6bSPrashanth Sreenivasa typedef struct vdev_indirect_config { 1575cabbc6bSPrashanth Sreenivasa /* 1585cabbc6bSPrashanth Sreenivasa * Object (in MOS) which contains the indirect mapping. This object 1595cabbc6bSPrashanth Sreenivasa * contains an array of vdev_indirect_mapping_entry_phys_t ordered by 1605cabbc6bSPrashanth Sreenivasa * vimep_src. The bonus buffer for this object is a 1615cabbc6bSPrashanth Sreenivasa * vdev_indirect_mapping_phys_t. This object is allocated when a vdev 1625cabbc6bSPrashanth Sreenivasa * removal is initiated. 1635cabbc6bSPrashanth Sreenivasa * 1645cabbc6bSPrashanth Sreenivasa * Note that this object can be empty if none of the data on the vdev 1655cabbc6bSPrashanth Sreenivasa * has been copied yet. 1665cabbc6bSPrashanth Sreenivasa */ 1675cabbc6bSPrashanth Sreenivasa uint64_t vic_mapping_object; 1685cabbc6bSPrashanth Sreenivasa 1695cabbc6bSPrashanth Sreenivasa /* 1705cabbc6bSPrashanth Sreenivasa * Object (in MOS) which contains the birth times for the mapping 1715cabbc6bSPrashanth Sreenivasa * entries. This object contains an array of 1725cabbc6bSPrashanth Sreenivasa * vdev_indirect_birth_entry_phys_t sorted by vibe_offset. The bonus 1735cabbc6bSPrashanth Sreenivasa * buffer for this object is a vdev_indirect_birth_phys_t. This object 1745cabbc6bSPrashanth Sreenivasa * is allocated when a vdev removal is initiated. 1755cabbc6bSPrashanth Sreenivasa * 1765cabbc6bSPrashanth Sreenivasa * Note that this object can be empty if none of the vdev has yet been 1775cabbc6bSPrashanth Sreenivasa * copied. 1785cabbc6bSPrashanth Sreenivasa */ 1795cabbc6bSPrashanth Sreenivasa uint64_t vic_births_object; 1805cabbc6bSPrashanth Sreenivasa 1815cabbc6bSPrashanth Sreenivasa /* 1825cabbc6bSPrashanth Sreenivasa * This is the vdev ID which was removed previous to this vdev, or 1835cabbc6bSPrashanth Sreenivasa * UINT64_MAX if there are no previously removed vdevs. 1845cabbc6bSPrashanth Sreenivasa */ 1855cabbc6bSPrashanth Sreenivasa uint64_t vic_prev_indirect_vdev; 1865cabbc6bSPrashanth Sreenivasa } vdev_indirect_config_t; 1875cabbc6bSPrashanth Sreenivasa 188fa9e4066Sahrens /* 189fa9e4066Sahrens * Virtual device descriptor 190fa9e4066Sahrens */ 191fa9e4066Sahrens struct vdev { 192fa9e4066Sahrens /* 193fa9e4066Sahrens * Common to all vdev types. 194fa9e4066Sahrens */ 195fa9e4066Sahrens uint64_t vdev_id; /* child number in vdev parent */ 196fa9e4066Sahrens uint64_t vdev_guid; /* unique ID for this vdev */ 197fa9e4066Sahrens uint64_t vdev_guid_sum; /* self guid + all child guids */ 1981195e687SMark J Musante uint64_t vdev_orig_guid; /* orig. guid prior to remove */ 199fa9e4066Sahrens uint64_t vdev_asize; /* allocatable device capacity */ 200573ca77eSGeorge Wilson uint64_t vdev_min_asize; /* min acceptable asize */ 2014263d13fSGeorge Wilson uint64_t vdev_max_asize; /* max acceptable asize */ 202fa9e4066Sahrens uint64_t vdev_ashift; /* block alignment shift */ 203fa9e4066Sahrens uint64_t vdev_state; /* see VDEV_STATE_* #defines */ 204560e6e96Seschrock uint64_t vdev_prevstate; /* used when reopening a vdev */ 205fa9e4066Sahrens vdev_ops_t *vdev_ops; /* vdev operations */ 206fa9e4066Sahrens spa_t *vdev_spa; /* spa for this vdev */ 207fa9e4066Sahrens void *vdev_tsd; /* type-specific data */ 208dcba9f3fSGeorge Wilson vnode_t *vdev_name_vp; /* vnode for pathname */ 209dcba9f3fSGeorge Wilson vnode_t *vdev_devid_vp; /* vnode for devid */ 210fa9e4066Sahrens vdev_t *vdev_top; /* top-level vdev */ 211fa9e4066Sahrens vdev_t *vdev_parent; /* parent vdev */ 212fa9e4066Sahrens vdev_t **vdev_child; /* array of children */ 213fa9e4066Sahrens uint64_t vdev_children; /* number of children */ 214fa9e4066Sahrens vdev_stat_t vdev_stat; /* virtual device statistics */ 215573ca77eSGeorge Wilson boolean_t vdev_expanding; /* expand the vdev? */ 216095bcd66SGeorge Wilson boolean_t vdev_reopening; /* reopen in progress? */ 217f64c0e34SEric Taylor int vdev_open_error; /* error on last open */ 218f64c0e34SEric Taylor kthread_t *vdev_open_thread; /* thread opening children */ 21988ecc943SGeorge Wilson uint64_t vdev_crtxg; /* txg when top-level was added */ 220fa9e4066Sahrens 221fa9e4066Sahrens /* 222fa9e4066Sahrens * Top-level vdev state. 223fa9e4066Sahrens */ 224fa9e4066Sahrens uint64_t vdev_ms_array; /* metaslab array object */ 225fa9e4066Sahrens uint64_t vdev_ms_shift; /* metaslab size shift */ 226fa9e4066Sahrens uint64_t vdev_ms_count; /* number of metaslabs */ 227fa9e4066Sahrens metaslab_group_t *vdev_mg; /* metaslab group */ 228fa9e4066Sahrens metaslab_t **vdev_ms; /* metaslab array */ 229fa9e4066Sahrens txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */ 230fa9e4066Sahrens txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */ 231fa9e4066Sahrens txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */ 2323d7072f8Seschrock boolean_t vdev_remove_wanted; /* async remove wanted? */ 233e14bb325SJeff Bonwick boolean_t vdev_probe_wanted; /* async probe wanted? */ 234e14bb325SJeff Bonwick list_node_t vdev_config_dirty_node; /* config dirty list */ 235e14bb325SJeff Bonwick list_node_t vdev_state_dirty_node; /* state dirty list */ 23699653d4eSeschrock uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */ 2378654d025Sperrin uint64_t vdev_islog; /* is an intent log device */ 2380713e232SGeorge Wilson uint64_t vdev_removing; /* device is being removed? */ 2390f7643c7SGeorge Wilson boolean_t vdev_ishole; /* is a hole in the namespace */ 2400f7643c7SGeorge Wilson kmutex_t vdev_queue_lock; /* protects vdev_queue_depth */ 241215198a6SJoe Stein uint64_t vdev_top_zap; 242fa9e4066Sahrens 24386714001SSerapheim Dimitropoulos /* pool checkpoint related */ 24486714001SSerapheim Dimitropoulos space_map_t *vdev_checkpoint_sm; /* contains reserved blocks */ 245*094e47e9SGeorge Wilson 246*094e47e9SGeorge Wilson boolean_t vdev_initialize_exit_wanted; 247*094e47e9SGeorge Wilson vdev_initializing_state_t vdev_initialize_state; 248*094e47e9SGeorge Wilson kthread_t *vdev_initialize_thread; 249*094e47e9SGeorge Wilson /* Protects vdev_initialize_thread and vdev_initialize_state. */ 250*094e47e9SGeorge Wilson kmutex_t vdev_initialize_lock; 251*094e47e9SGeorge Wilson kcondvar_t vdev_initialize_cv; 252*094e47e9SGeorge Wilson uint64_t vdev_initialize_offset[TXG_SIZE]; 253*094e47e9SGeorge Wilson uint64_t vdev_initialize_last_offset; 254*094e47e9SGeorge Wilson range_tree_t *vdev_initialize_tree; /* valid while initializing */ 255*094e47e9SGeorge Wilson uint64_t vdev_initialize_bytes_est; 256*094e47e9SGeorge Wilson uint64_t vdev_initialize_bytes_done; 257*094e47e9SGeorge Wilson time_t vdev_initialize_action_time; /* start and end time */ 258*094e47e9SGeorge Wilson 259*094e47e9SGeorge Wilson /* for limiting outstanding I/Os */ 260*094e47e9SGeorge Wilson kmutex_t vdev_initialize_io_lock; 261*094e47e9SGeorge Wilson kcondvar_t vdev_initialize_io_cv; 262*094e47e9SGeorge Wilson uint64_t vdev_initialize_inflight; 26386714001SSerapheim Dimitropoulos 2645cabbc6bSPrashanth Sreenivasa /* 2655cabbc6bSPrashanth Sreenivasa * Values stored in the config for an indirect or removing vdev. 2665cabbc6bSPrashanth Sreenivasa */ 2675cabbc6bSPrashanth Sreenivasa vdev_indirect_config_t vdev_indirect_config; 2685cabbc6bSPrashanth Sreenivasa 2695cabbc6bSPrashanth Sreenivasa /* 2705cabbc6bSPrashanth Sreenivasa * The vdev_indirect_rwlock protects the vdev_indirect_mapping 2715cabbc6bSPrashanth Sreenivasa * pointer from changing on indirect vdevs (when it is condensed). 2725cabbc6bSPrashanth Sreenivasa * Note that removing (not yet indirect) vdevs have different 2735cabbc6bSPrashanth Sreenivasa * access patterns (the mapping is not accessed from open context, 2745cabbc6bSPrashanth Sreenivasa * e.g. from zio_read) and locking strategy (e.g. svr_lock). 2755cabbc6bSPrashanth Sreenivasa */ 2765cabbc6bSPrashanth Sreenivasa krwlock_t vdev_indirect_rwlock; 2775cabbc6bSPrashanth Sreenivasa vdev_indirect_mapping_t *vdev_indirect_mapping; 2785cabbc6bSPrashanth Sreenivasa vdev_indirect_births_t *vdev_indirect_births; 2795cabbc6bSPrashanth Sreenivasa 2805cabbc6bSPrashanth Sreenivasa /* 2815cabbc6bSPrashanth Sreenivasa * In memory data structures used to manage the obsolete sm, for 2825cabbc6bSPrashanth Sreenivasa * indirect or removing vdevs. 2835cabbc6bSPrashanth Sreenivasa * 2845cabbc6bSPrashanth Sreenivasa * The vdev_obsolete_segments is the in-core record of the segments 2855cabbc6bSPrashanth Sreenivasa * that are no longer referenced anywhere in the pool (due to 2865cabbc6bSPrashanth Sreenivasa * being freed or remapped and not referenced by any snapshots). 2875cabbc6bSPrashanth Sreenivasa * During a sync, segments are added to vdev_obsolete_segments 2885cabbc6bSPrashanth Sreenivasa * via vdev_indirect_mark_obsolete(); at the end of each sync 2895cabbc6bSPrashanth Sreenivasa * pass, this is appended to vdev_obsolete_sm via 2905cabbc6bSPrashanth Sreenivasa * vdev_indirect_sync_obsolete(). The vdev_obsolete_lock 2915cabbc6bSPrashanth Sreenivasa * protects against concurrent modifications of vdev_obsolete_segments 2925cabbc6bSPrashanth Sreenivasa * from multiple zio threads. 2935cabbc6bSPrashanth Sreenivasa */ 2945cabbc6bSPrashanth Sreenivasa kmutex_t vdev_obsolete_lock; 2955cabbc6bSPrashanth Sreenivasa range_tree_t *vdev_obsolete_segments; 2965cabbc6bSPrashanth Sreenivasa space_map_t *vdev_obsolete_sm; 2975cabbc6bSPrashanth Sreenivasa 2980f7643c7SGeorge Wilson /* 2990f7643c7SGeorge Wilson * The queue depth parameters determine how many async writes are 3000f7643c7SGeorge Wilson * still pending (i.e. allocated by net yet issued to disk) per 3010f7643c7SGeorge Wilson * top-level (vdev_async_write_queue_depth) and the maximum allowed 3020f7643c7SGeorge Wilson * (vdev_max_async_write_queue_depth). These values only apply to 3030f7643c7SGeorge Wilson * top-level vdevs. 3040f7643c7SGeorge Wilson */ 3050f7643c7SGeorge Wilson uint64_t vdev_async_write_queue_depth; 3060f7643c7SGeorge Wilson uint64_t vdev_max_async_write_queue_depth; 3070f7643c7SGeorge Wilson 308fa9e4066Sahrens /* 309fa9e4066Sahrens * Leaf vdev state. 310fa9e4066Sahrens */ 3110713e232SGeorge Wilson range_tree_t *vdev_dtl[DTL_TYPES]; /* dirty time logs */ 3120713e232SGeorge Wilson space_map_t *vdev_dtl_sm; /* dirty time log space map */ 313fa9e4066Sahrens txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */ 3140713e232SGeorge Wilson uint64_t vdev_dtl_object; /* DTL object */ 3150713e232SGeorge Wilson uint64_t vdev_psize; /* physical device capacity */ 316afefbcddSeschrock uint64_t vdev_wholedisk; /* true if this is a whole disk */ 3173d7072f8Seschrock uint64_t vdev_offline; /* persistent offline state */ 3183d7072f8Seschrock uint64_t vdev_faulted; /* persistent faulted state */ 3193d7072f8Seschrock uint64_t vdev_degraded; /* persistent degraded state */ 3203d7072f8Seschrock uint64_t vdev_removed; /* persistent removed state */ 321b4952e17SGeorge Wilson uint64_t vdev_resilver_txg; /* persistent resilvering state */ 32299653d4eSeschrock uint64_t vdev_nparity; /* number of parity devices for raidz */ 323fa9e4066Sahrens char *vdev_path; /* vdev path (if any) */ 324fa9e4066Sahrens char *vdev_devid; /* vdev devid (if any) */ 3253d7072f8Seschrock char *vdev_physpath; /* vdev device path (if any) */ 3266809eb4eSEric Schrock char *vdev_fru; /* physical FRU location */ 327e14bb325SJeff Bonwick uint64_t vdev_not_present; /* not present during import */ 328e14bb325SJeff Bonwick uint64_t vdev_unspare; /* unspare when resilvering done */ 329e14bb325SJeff Bonwick boolean_t vdev_nowritecache; /* true if flushwritecache failed */ 330e14bb325SJeff Bonwick boolean_t vdev_checkremove; /* temporary online test */ 331e14bb325SJeff Bonwick boolean_t vdev_forcefault; /* force online fault */ 3321195e687SMark J Musante boolean_t vdev_splitting; /* split or repair in progress */ 33398d1cbfeSGeorge Wilson boolean_t vdev_delayed_close; /* delayed device close? */ 3340713e232SGeorge Wilson boolean_t vdev_tmpoffline; /* device taken offline temporarily? */ 3350713e232SGeorge Wilson boolean_t vdev_detached; /* device detached? */ 3360713e232SGeorge Wilson boolean_t vdev_cant_read; /* vdev is failing all reads */ 3370713e232SGeorge Wilson boolean_t vdev_cant_write; /* vdev is failing all writes */ 3380713e232SGeorge Wilson boolean_t vdev_isspare; /* was a hot spare */ 3390713e232SGeorge Wilson boolean_t vdev_isl2cache; /* was a l2cache device */ 340fa9e4066Sahrens vdev_queue_t vdev_queue; /* I/O deadline schedule queue */ 341fa9e4066Sahrens vdev_cache_t vdev_cache; /* physical block cache */ 34281cd5c55SMatthew Ahrens spa_aux_vdev_t *vdev_aux; /* for l2cache and spares vdevs */ 343e14bb325SJeff Bonwick zio_t *vdev_probe_zio; /* root of current probe */ 344069f55e2SEric Schrock vdev_aux_t vdev_label_aux; /* on-disk aux state */ 345215198a6SJoe Stein uint64_t vdev_leaf_zap; 346fa9e4066Sahrens 347fa9e4066Sahrens /* 348fa9e4066Sahrens * For DTrace to work in userland (libzpool) context, these fields must 349fa9e4066Sahrens * remain at the end of the structure. DTrace will use the kernel's 350fa9e4066Sahrens * CTF definition for 'struct vdev', and since the size of a kmutex_t is 351ad135b5dSChristopher Siden * larger in userland, the offsets for the rest of the fields would be 352fa9e4066Sahrens * incorrect. 353fa9e4066Sahrens */ 354fa9e4066Sahrens kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */ 355fa9e4066Sahrens kmutex_t vdev_stat_lock; /* vdev_stat */ 356e14bb325SJeff Bonwick kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */ 357fa9e4066Sahrens }; 358fa9e4066Sahrens 359b24ab676SJeff Bonwick #define VDEV_RAIDZ_MAXPARITY 3 360b24ab676SJeff Bonwick 361f83ffe1aSLin Ling #define VDEV_PAD_SIZE (8 << 10) 362f83ffe1aSLin Ling /* 2 padding areas (vl_pad1 and vl_pad2) to skip */ 363f83ffe1aSLin Ling #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2 364fa9e4066Sahrens #define VDEV_PHYS_SIZE (112 << 10) 365ecc2d604Sbonwick #define VDEV_UBERBLOCK_RING (128 << 10) 366ecc2d604Sbonwick 3672a104a52SAlex Reece /* The largest uberblock we support is 8k. */ 3682a104a52SAlex Reece #define MAX_UBERBLOCK_SHIFT (13) 369ecc2d604Sbonwick #define VDEV_UBERBLOCK_SHIFT(vd) \ 3702a104a52SAlex Reece MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \ 3712a104a52SAlex Reece MAX_UBERBLOCK_SHIFT) 372ecc2d604Sbonwick #define VDEV_UBERBLOCK_COUNT(vd) \ 373ecc2d604Sbonwick (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd)) 374ecc2d604Sbonwick #define VDEV_UBERBLOCK_OFFSET(vd, n) \ 375ecc2d604Sbonwick offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)]) 376ecc2d604Sbonwick #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd)) 377fa9e4066Sahrens 378fa9e4066Sahrens typedef struct vdev_phys { 3796e1f5caaSNeil Perrin char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)]; 3806e1f5caaSNeil Perrin zio_eck_t vp_zbt; 381fa9e4066Sahrens } vdev_phys_t; 382fa9e4066Sahrens 383fa9e4066Sahrens typedef struct vdev_label { 384f83ffe1aSLin Ling char vl_pad1[VDEV_PAD_SIZE]; /* 8K */ 385f83ffe1aSLin Ling char vl_pad2[VDEV_PAD_SIZE]; /* 8K */ 386ecc2d604Sbonwick vdev_phys_t vl_vdev_phys; /* 112K */ 387ecc2d604Sbonwick char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */ 388fa9e4066Sahrens } vdev_label_t; /* 256K total */ 389fa9e4066Sahrens 390ecc2d604Sbonwick /* 391ecc2d604Sbonwick * vdev_dirty() flags 392ecc2d604Sbonwick */ 393ecc2d604Sbonwick #define VDD_METASLAB 0x01 394ecc2d604Sbonwick #define VDD_DTL 0x02 395ecc2d604Sbonwick 396f7170741SWill Andrews /* Offset of embedded boot loader region on each label */ 397f7170741SWill Andrews #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t)) 398fa9e4066Sahrens /* 399f7170741SWill Andrews * Size of embedded boot loader region on each label. 400fa9e4066Sahrens * The total size of the first two labels plus the boot area is 4MB. 401fa9e4066Sahrens */ 402f7170741SWill Andrews #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */ 403fa9e4066Sahrens 404fa9e4066Sahrens /* 405fa9e4066Sahrens * Size of label regions at the start and end of each leaf device. 406fa9e4066Sahrens */ 407fa9e4066Sahrens #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE) 408fa9e4066Sahrens #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t)) 409fa9e4066Sahrens #define VDEV_LABELS 4 410ad135b5dSChristopher Siden #define VDEV_BEST_LABEL VDEV_LABELS 411fa9e4066Sahrens 412fa9e4066Sahrens #define VDEV_ALLOC_LOAD 0 413fa9e4066Sahrens #define VDEV_ALLOC_ADD 1 41499653d4eSeschrock #define VDEV_ALLOC_SPARE 2 415fa94a07fSbrendan #define VDEV_ALLOC_L2CACHE 3 41621ecdf64SLin Ling #define VDEV_ALLOC_ROOTPOOL 4 4171195e687SMark J Musante #define VDEV_ALLOC_SPLIT 5 418cd0837ccSGeorge Wilson #define VDEV_ALLOC_ATTACH 6 419fa9e4066Sahrens 420fa9e4066Sahrens /* 421fa9e4066Sahrens * Allocate or free a vdev 422fa9e4066Sahrens */ 42388ecc943SGeorge Wilson extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, 42488ecc943SGeorge Wilson vdev_ops_t *ops); 42599653d4eSeschrock extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config, 42699653d4eSeschrock vdev_t *parent, uint_t id, int alloctype); 427fa9e4066Sahrens extern void vdev_free(vdev_t *vd); 428fa9e4066Sahrens 429fa9e4066Sahrens /* 430fa9e4066Sahrens * Add or remove children and parents 431fa9e4066Sahrens */ 432fa9e4066Sahrens extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd); 433fa9e4066Sahrens extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd); 434fa9e4066Sahrens extern void vdev_compact_children(vdev_t *pvd); 435fa9e4066Sahrens extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops); 436fa9e4066Sahrens extern void vdev_remove_parent(vdev_t *cvd); 437fa9e4066Sahrens 438fa9e4066Sahrens /* 439fa9e4066Sahrens * vdev sync load and sync 440fa9e4066Sahrens */ 4414b964adaSGeorge Wilson extern boolean_t vdev_log_state_valid(vdev_t *vd); 4425cabbc6bSPrashanth Sreenivasa extern int vdev_load(vdev_t *vd); 4430713e232SGeorge Wilson extern int vdev_dtl_load(vdev_t *vd); 444fa9e4066Sahrens extern void vdev_sync(vdev_t *vd, uint64_t txg); 445fa9e4066Sahrens extern void vdev_sync_done(vdev_t *vd, uint64_t txg); 446ecc2d604Sbonwick extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg); 4470713e232SGeorge Wilson extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg); 448fa9e4066Sahrens 449fa9e4066Sahrens /* 450fa9e4066Sahrens * Available vdev types. 451fa9e4066Sahrens */ 452fa9e4066Sahrens extern vdev_ops_t vdev_root_ops; 453fa9e4066Sahrens extern vdev_ops_t vdev_mirror_ops; 454fa9e4066Sahrens extern vdev_ops_t vdev_replacing_ops; 455fa9e4066Sahrens extern vdev_ops_t vdev_raidz_ops; 456fa9e4066Sahrens extern vdev_ops_t vdev_disk_ops; 457fa9e4066Sahrens extern vdev_ops_t vdev_file_ops; 458fa9e4066Sahrens extern vdev_ops_t vdev_missing_ops; 45988ecc943SGeorge Wilson extern vdev_ops_t vdev_hole_ops; 46099653d4eSeschrock extern vdev_ops_t vdev_spare_ops; 4615cabbc6bSPrashanth Sreenivasa extern vdev_ops_t vdev_indirect_ops; 462fa9e4066Sahrens 463fa9e4066Sahrens /* 4642a79c5feSlling * Common size functions 465fa9e4066Sahrens */ 466*094e47e9SGeorge Wilson extern void vdev_default_xlate(vdev_t *vd, const range_seg_t *in, 467*094e47e9SGeorge Wilson range_seg_t *out); 468fa9e4066Sahrens extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize); 469573ca77eSGeorge Wilson extern uint64_t vdev_get_min_asize(vdev_t *vd); 470573ca77eSGeorge Wilson extern void vdev_set_min_asize(vdev_t *vd); 471fa9e4066Sahrens 472e81c57b1Sahrens /* 473f7170741SWill Andrews * Global variables 474e81c57b1Sahrens */ 47586714001SSerapheim Dimitropoulos extern int vdev_standard_sm_blksz; 476f7170741SWill Andrews /* zdb uses this tunable, so it must be declared here to make lint happy. */ 477e81c57b1Sahrens extern int zfs_vdev_cache_size; 478e81c57b1Sahrens 4795cabbc6bSPrashanth Sreenivasa /* 4805cabbc6bSPrashanth Sreenivasa * Functions from vdev_indirect.c 4815cabbc6bSPrashanth Sreenivasa */ 4825cabbc6bSPrashanth Sreenivasa extern void vdev_indirect_sync_obsolete(vdev_t *vd, dmu_tx_t *tx); 4835cabbc6bSPrashanth Sreenivasa extern boolean_t vdev_indirect_should_condense(vdev_t *vd); 4845cabbc6bSPrashanth Sreenivasa extern void spa_condense_indirect_start_sync(vdev_t *vd, dmu_tx_t *tx); 4855cabbc6bSPrashanth Sreenivasa extern int vdev_obsolete_sm_object(vdev_t *vd); 4865cabbc6bSPrashanth Sreenivasa extern boolean_t vdev_obsolete_counts_are_precise(vdev_t *vd); 4875cabbc6bSPrashanth Sreenivasa 48886714001SSerapheim Dimitropoulos /* 48986714001SSerapheim Dimitropoulos * Other miscellaneous functions 49086714001SSerapheim Dimitropoulos */ 49186714001SSerapheim Dimitropoulos int vdev_checkpoint_sm_object(vdev_t *vd); 49286714001SSerapheim Dimitropoulos 49331d7e8faSGeorge Wilson /* 49431d7e8faSGeorge Wilson * The vdev_buf_t is used to translate between zio_t and buf_t, and back again. 49531d7e8faSGeorge Wilson */ 49631d7e8faSGeorge Wilson typedef struct vdev_buf { 49731d7e8faSGeorge Wilson buf_t vb_buf; /* buffer that describes the io */ 49831d7e8faSGeorge Wilson zio_t *vb_io; /* pointer back to the original zio_t */ 49931d7e8faSGeorge Wilson } vdev_buf_t; 50031d7e8faSGeorge Wilson 501fa9e4066Sahrens #ifdef __cplusplus 502fa9e4066Sahrens } 503fa9e4066Sahrens #endif 504fa9e4066Sahrens 505fa9e4066Sahrens #endif /* _SYS_VDEV_IMPL_H */ 506