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, 2020 by Delphix. All rights reserved. 24 * Copyright (c) 2017, Intel Corporation. 25 */ 26 27 #ifndef _SYS_VDEV_IMPL_H 28 #define _SYS_VDEV_IMPL_H 29 30 #include <sys/avl.h> 31 #include <sys/bpobj.h> 32 #include <sys/dmu.h> 33 #include <sys/metaslab.h> 34 #include <sys/nvpair.h> 35 #include <sys/space_map.h> 36 #include <sys/vdev.h> 37 #include <sys/dkio.h> 38 #include <sys/uberblock_impl.h> 39 #include <sys/vdev_indirect_mapping.h> 40 #include <sys/vdev_indirect_births.h> 41 #include <sys/vdev_rebuild.h> 42 #include <sys/vdev_removal.h> 43 #include <sys/zfs_ratelimit.h> 44 45 #ifdef __cplusplus 46 extern "C" { 47 #endif 48 49 /* 50 * Virtual device descriptors. 51 * 52 * All storage pool operations go through the virtual device framework, 53 * which provides data replication and I/O scheduling. 54 */ 55 56 /* 57 * Forward declarations that lots of things need. 58 */ 59 typedef struct vdev_queue vdev_queue_t; 60 typedef struct vdev_cache vdev_cache_t; 61 typedef struct vdev_cache_entry vdev_cache_entry_t; 62 struct abd; 63 64 extern int zfs_vdev_queue_depth_pct; 65 extern int zfs_vdev_def_queue_depth; 66 extern uint32_t zfs_vdev_async_write_max_active; 67 68 /* 69 * Virtual device operations 70 */ 71 typedef int vdev_init_func_t(spa_t *spa, nvlist_t *nv, void **tsd); 72 typedef void vdev_fini_func_t(vdev_t *vd); 73 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size, 74 uint64_t *ashift, uint64_t *pshift); 75 typedef void vdev_close_func_t(vdev_t *vd); 76 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize); 77 typedef uint64_t vdev_min_asize_func_t(vdev_t *vd); 78 typedef uint64_t vdev_min_alloc_func_t(vdev_t *vd); 79 typedef void vdev_io_start_func_t(zio_t *zio); 80 typedef void vdev_io_done_func_t(zio_t *zio); 81 typedef void vdev_state_change_func_t(vdev_t *vd, int, int); 82 typedef boolean_t vdev_need_resilver_func_t(vdev_t *vd, const dva_t *dva, 83 size_t psize, uint64_t phys_birth); 84 typedef void vdev_hold_func_t(vdev_t *vd); 85 typedef void vdev_rele_func_t(vdev_t *vd); 86 87 typedef void vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd, 88 uint64_t offset, uint64_t size, void *arg); 89 typedef void vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size, 90 vdev_remap_cb_t callback, void *arg); 91 /* 92 * Given a target vdev, translates the logical range "in" to the physical 93 * range "res" 94 */ 95 typedef void vdev_xlation_func_t(vdev_t *cvd, const range_seg64_t *logical, 96 range_seg64_t *physical, range_seg64_t *remain); 97 typedef uint64_t vdev_rebuild_asize_func_t(vdev_t *vd, uint64_t start, 98 uint64_t size, uint64_t max_segment); 99 typedef void vdev_metaslab_init_func_t(vdev_t *vd, uint64_t *startp, 100 uint64_t *sizep); 101 typedef void vdev_config_generate_func_t(vdev_t *vd, nvlist_t *nv); 102 typedef uint64_t vdev_nparity_func_t(vdev_t *vd); 103 typedef uint64_t vdev_ndisks_func_t(vdev_t *vd); 104 105 typedef const struct vdev_ops { 106 vdev_init_func_t *vdev_op_init; 107 vdev_fini_func_t *vdev_op_fini; 108 vdev_open_func_t *vdev_op_open; 109 vdev_close_func_t *vdev_op_close; 110 vdev_asize_func_t *vdev_op_asize; 111 vdev_min_asize_func_t *vdev_op_min_asize; 112 vdev_min_alloc_func_t *vdev_op_min_alloc; 113 vdev_io_start_func_t *vdev_op_io_start; 114 vdev_io_done_func_t *vdev_op_io_done; 115 vdev_state_change_func_t *vdev_op_state_change; 116 vdev_need_resilver_func_t *vdev_op_need_resilver; 117 vdev_hold_func_t *vdev_op_hold; 118 vdev_rele_func_t *vdev_op_rele; 119 vdev_remap_func_t *vdev_op_remap; 120 vdev_xlation_func_t *vdev_op_xlate; 121 vdev_rebuild_asize_func_t *vdev_op_rebuild_asize; 122 vdev_metaslab_init_func_t *vdev_op_metaslab_init; 123 vdev_config_generate_func_t *vdev_op_config_generate; 124 vdev_nparity_func_t *vdev_op_nparity; 125 vdev_ndisks_func_t *vdev_op_ndisks; 126 char vdev_op_type[16]; 127 boolean_t vdev_op_leaf; 128 } vdev_ops_t; 129 130 /* 131 * Virtual device properties 132 */ 133 struct vdev_cache_entry { 134 struct abd *ve_abd; 135 uint64_t ve_offset; 136 clock_t ve_lastused; 137 avl_node_t ve_offset_node; 138 avl_node_t ve_lastused_node; 139 uint32_t ve_hits; 140 uint16_t ve_missed_update; 141 zio_t *ve_fill_io; 142 }; 143 144 struct vdev_cache { 145 avl_tree_t vc_offset_tree; 146 avl_tree_t vc_lastused_tree; 147 kmutex_t vc_lock; 148 }; 149 150 typedef struct vdev_queue_class { 151 uint32_t vqc_active; 152 153 /* 154 * Sorted by offset or timestamp, depending on if the queue is 155 * LBA-ordered vs FIFO. 156 */ 157 avl_tree_t vqc_queued_tree; 158 } vdev_queue_class_t; 159 160 struct vdev_queue { 161 vdev_t *vq_vdev; 162 vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE]; 163 avl_tree_t vq_active_tree; 164 avl_tree_t vq_read_offset_tree; 165 avl_tree_t vq_write_offset_tree; 166 avl_tree_t vq_trim_offset_tree; 167 uint64_t vq_last_offset; 168 zio_priority_t vq_last_prio; /* Last sent I/O priority. */ 169 uint32_t vq_ia_active; /* Active interactive I/Os. */ 170 uint32_t vq_nia_credit; /* Non-interactive I/Os credit. */ 171 hrtime_t vq_io_complete_ts; /* time last i/o completed */ 172 hrtime_t vq_io_delta_ts; 173 zio_t vq_io_search; /* used as local for stack reduction */ 174 kmutex_t vq_lock; 175 }; 176 177 typedef enum vdev_alloc_bias { 178 VDEV_BIAS_NONE, 179 VDEV_BIAS_LOG, /* dedicated to ZIL data (SLOG) */ 180 VDEV_BIAS_SPECIAL, /* dedicated to ddt, metadata, and small blks */ 181 VDEV_BIAS_DEDUP /* dedicated to dedup metadata */ 182 } vdev_alloc_bias_t; 183 184 185 /* 186 * On-disk indirect vdev state. 187 * 188 * An indirect vdev is described exclusively in the MOS config of a pool. 189 * The config for an indirect vdev includes several fields, which are 190 * accessed in memory by a vdev_indirect_config_t. 191 */ 192 typedef struct vdev_indirect_config { 193 /* 194 * Object (in MOS) which contains the indirect mapping. This object 195 * contains an array of vdev_indirect_mapping_entry_phys_t ordered by 196 * vimep_src. The bonus buffer for this object is a 197 * vdev_indirect_mapping_phys_t. This object is allocated when a vdev 198 * removal is initiated. 199 * 200 * Note that this object can be empty if none of the data on the vdev 201 * has been copied yet. 202 */ 203 uint64_t vic_mapping_object; 204 205 /* 206 * Object (in MOS) which contains the birth times for the mapping 207 * entries. This object contains an array of 208 * vdev_indirect_birth_entry_phys_t sorted by vibe_offset. The bonus 209 * buffer for this object is a vdev_indirect_birth_phys_t. This object 210 * is allocated when a vdev removal is initiated. 211 * 212 * Note that this object can be empty if none of the vdev has yet been 213 * copied. 214 */ 215 uint64_t vic_births_object; 216 217 /* 218 * This is the vdev ID which was removed previous to this vdev, or 219 * UINT64_MAX if there are no previously removed vdevs. 220 */ 221 uint64_t vic_prev_indirect_vdev; 222 } vdev_indirect_config_t; 223 224 /* 225 * Virtual device descriptor 226 */ 227 struct vdev { 228 /* 229 * Common to all vdev types. 230 */ 231 uint64_t vdev_id; /* child number in vdev parent */ 232 uint64_t vdev_guid; /* unique ID for this vdev */ 233 uint64_t vdev_guid_sum; /* self guid + all child guids */ 234 uint64_t vdev_orig_guid; /* orig. guid prior to remove */ 235 uint64_t vdev_asize; /* allocatable device capacity */ 236 uint64_t vdev_min_asize; /* min acceptable asize */ 237 uint64_t vdev_max_asize; /* max acceptable asize */ 238 uint64_t vdev_ashift; /* block alignment shift */ 239 240 /* 241 * Logical block alignment shift 242 * 243 * The smallest sized/aligned I/O supported by the device. 244 */ 245 uint64_t vdev_logical_ashift; 246 /* 247 * Physical block alignment shift 248 * 249 * The device supports logical I/Os with vdev_logical_ashift 250 * size/alignment, but optimum performance will be achieved by 251 * aligning/sizing requests to vdev_physical_ashift. Smaller 252 * requests may be inflated or incur device level read-modify-write 253 * operations. 254 * 255 * May be 0 to indicate no preference (i.e. use vdev_logical_ashift). 256 */ 257 uint64_t vdev_physical_ashift; 258 uint64_t vdev_state; /* see VDEV_STATE_* #defines */ 259 uint64_t vdev_prevstate; /* used when reopening a vdev */ 260 vdev_ops_t *vdev_ops; /* vdev operations */ 261 spa_t *vdev_spa; /* spa for this vdev */ 262 void *vdev_tsd; /* type-specific data */ 263 vdev_t *vdev_top; /* top-level vdev */ 264 vdev_t *vdev_parent; /* parent vdev */ 265 vdev_t **vdev_child; /* array of children */ 266 uint64_t vdev_children; /* number of children */ 267 vdev_stat_t vdev_stat; /* virtual device statistics */ 268 vdev_stat_ex_t vdev_stat_ex; /* extended statistics */ 269 boolean_t vdev_expanding; /* expand the vdev? */ 270 boolean_t vdev_reopening; /* reopen in progress? */ 271 boolean_t vdev_nonrot; /* true if solid state */ 272 int vdev_load_error; /* error on last load */ 273 int vdev_open_error; /* error on last open */ 274 int vdev_validate_error; /* error on last validate */ 275 kthread_t *vdev_open_thread; /* thread opening children */ 276 kthread_t *vdev_validate_thread; /* thread validating children */ 277 uint64_t vdev_crtxg; /* txg when top-level was added */ 278 279 /* 280 * Top-level vdev state. 281 */ 282 uint64_t vdev_ms_array; /* metaslab array object */ 283 uint64_t vdev_ms_shift; /* metaslab size shift */ 284 uint64_t vdev_ms_count; /* number of metaslabs */ 285 metaslab_group_t *vdev_mg; /* metaslab group */ 286 metaslab_group_t *vdev_log_mg; /* embedded slog metaslab group */ 287 metaslab_t **vdev_ms; /* metaslab array */ 288 uint64_t vdev_pending_fastwrite; /* allocated fastwrites */ 289 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */ 290 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */ 291 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */ 292 boolean_t vdev_remove_wanted; /* async remove wanted? */ 293 boolean_t vdev_probe_wanted; /* async probe wanted? */ 294 list_node_t vdev_config_dirty_node; /* config dirty list */ 295 list_node_t vdev_state_dirty_node; /* state dirty list */ 296 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */ 297 uint64_t vdev_islog; /* is an intent log device */ 298 uint64_t vdev_noalloc; /* device is passivated? */ 299 uint64_t vdev_removing; /* device is being removed? */ 300 boolean_t vdev_ishole; /* is a hole in the namespace */ 301 uint64_t vdev_top_zap; 302 vdev_alloc_bias_t vdev_alloc_bias; /* metaslab allocation bias */ 303 304 /* pool checkpoint related */ 305 space_map_t *vdev_checkpoint_sm; /* contains reserved blocks */ 306 307 /* Initialize related */ 308 boolean_t vdev_initialize_exit_wanted; 309 vdev_initializing_state_t vdev_initialize_state; 310 list_node_t vdev_initialize_node; 311 kthread_t *vdev_initialize_thread; 312 /* Protects vdev_initialize_thread and vdev_initialize_state. */ 313 kmutex_t vdev_initialize_lock; 314 kcondvar_t vdev_initialize_cv; 315 uint64_t vdev_initialize_offset[TXG_SIZE]; 316 uint64_t vdev_initialize_last_offset; 317 range_tree_t *vdev_initialize_tree; /* valid while initializing */ 318 uint64_t vdev_initialize_bytes_est; 319 uint64_t vdev_initialize_bytes_done; 320 uint64_t vdev_initialize_action_time; /* start and end time */ 321 322 /* TRIM related */ 323 boolean_t vdev_trim_exit_wanted; 324 boolean_t vdev_autotrim_exit_wanted; 325 vdev_trim_state_t vdev_trim_state; 326 list_node_t vdev_trim_node; 327 kmutex_t vdev_autotrim_lock; 328 kcondvar_t vdev_autotrim_cv; 329 kthread_t *vdev_autotrim_thread; 330 /* Protects vdev_trim_thread and vdev_trim_state. */ 331 kmutex_t vdev_trim_lock; 332 kcondvar_t vdev_trim_cv; 333 kthread_t *vdev_trim_thread; 334 uint64_t vdev_trim_offset[TXG_SIZE]; 335 uint64_t vdev_trim_last_offset; 336 uint64_t vdev_trim_bytes_est; 337 uint64_t vdev_trim_bytes_done; 338 uint64_t vdev_trim_rate; /* requested rate (bytes/sec) */ 339 uint64_t vdev_trim_partial; /* requested partial TRIM */ 340 uint64_t vdev_trim_secure; /* requested secure TRIM */ 341 uint64_t vdev_trim_action_time; /* start and end time */ 342 343 /* Rebuild related */ 344 boolean_t vdev_rebuilding; 345 boolean_t vdev_rebuild_exit_wanted; 346 boolean_t vdev_rebuild_cancel_wanted; 347 boolean_t vdev_rebuild_reset_wanted; 348 kmutex_t vdev_rebuild_lock; 349 kcondvar_t vdev_rebuild_cv; 350 kthread_t *vdev_rebuild_thread; 351 vdev_rebuild_t vdev_rebuild_config; 352 353 /* For limiting outstanding I/Os (initialize, TRIM) */ 354 kmutex_t vdev_initialize_io_lock; 355 kcondvar_t vdev_initialize_io_cv; 356 uint64_t vdev_initialize_inflight; 357 kmutex_t vdev_trim_io_lock; 358 kcondvar_t vdev_trim_io_cv; 359 uint64_t vdev_trim_inflight[3]; 360 361 /* 362 * Values stored in the config for an indirect or removing vdev. 363 */ 364 vdev_indirect_config_t vdev_indirect_config; 365 366 /* 367 * The vdev_indirect_rwlock protects the vdev_indirect_mapping 368 * pointer from changing on indirect vdevs (when it is condensed). 369 * Note that removing (not yet indirect) vdevs have different 370 * access patterns (the mapping is not accessed from open context, 371 * e.g. from zio_read) and locking strategy (e.g. svr_lock). 372 */ 373 krwlock_t vdev_indirect_rwlock; 374 vdev_indirect_mapping_t *vdev_indirect_mapping; 375 vdev_indirect_births_t *vdev_indirect_births; 376 377 /* 378 * In memory data structures used to manage the obsolete sm, for 379 * indirect or removing vdevs. 380 * 381 * The vdev_obsolete_segments is the in-core record of the segments 382 * that are no longer referenced anywhere in the pool (due to 383 * being freed or remapped and not referenced by any snapshots). 384 * During a sync, segments are added to vdev_obsolete_segments 385 * via vdev_indirect_mark_obsolete(); at the end of each sync 386 * pass, this is appended to vdev_obsolete_sm via 387 * vdev_indirect_sync_obsolete(). The vdev_obsolete_lock 388 * protects against concurrent modifications of vdev_obsolete_segments 389 * from multiple zio threads. 390 */ 391 kmutex_t vdev_obsolete_lock; 392 range_tree_t *vdev_obsolete_segments; 393 space_map_t *vdev_obsolete_sm; 394 395 /* 396 * Protects the vdev_scan_io_queue field itself as well as the 397 * structure's contents (when present). 398 */ 399 kmutex_t vdev_scan_io_queue_lock; 400 struct dsl_scan_io_queue *vdev_scan_io_queue; 401 402 /* 403 * Leaf vdev state. 404 */ 405 range_tree_t *vdev_dtl[DTL_TYPES]; /* dirty time logs */ 406 space_map_t *vdev_dtl_sm; /* dirty time log space map */ 407 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */ 408 uint64_t vdev_dtl_object; /* DTL object */ 409 uint64_t vdev_psize; /* physical device capacity */ 410 uint64_t vdev_wholedisk; /* true if this is a whole disk */ 411 uint64_t vdev_offline; /* persistent offline state */ 412 uint64_t vdev_faulted; /* persistent faulted state */ 413 uint64_t vdev_degraded; /* persistent degraded state */ 414 uint64_t vdev_removed; /* persistent removed state */ 415 uint64_t vdev_resilver_txg; /* persistent resilvering state */ 416 uint64_t vdev_rebuild_txg; /* persistent rebuilding state */ 417 char *vdev_path; /* vdev path (if any) */ 418 char *vdev_devid; /* vdev devid (if any) */ 419 char *vdev_physpath; /* vdev device path (if any) */ 420 char *vdev_enc_sysfs_path; /* enclosure sysfs path */ 421 char *vdev_fru; /* physical FRU location */ 422 uint64_t vdev_not_present; /* not present during import */ 423 uint64_t vdev_unspare; /* unspare when resilvering done */ 424 boolean_t vdev_nowritecache; /* true if flushwritecache failed */ 425 boolean_t vdev_has_trim; /* TRIM is supported */ 426 boolean_t vdev_has_securetrim; /* secure TRIM is supported */ 427 boolean_t vdev_checkremove; /* temporary online test */ 428 boolean_t vdev_forcefault; /* force online fault */ 429 boolean_t vdev_splitting; /* split or repair in progress */ 430 boolean_t vdev_delayed_close; /* delayed device close? */ 431 boolean_t vdev_tmpoffline; /* device taken offline temporarily? */ 432 boolean_t vdev_detached; /* device detached? */ 433 boolean_t vdev_cant_read; /* vdev is failing all reads */ 434 boolean_t vdev_cant_write; /* vdev is failing all writes */ 435 boolean_t vdev_isspare; /* was a hot spare */ 436 boolean_t vdev_isl2cache; /* was a l2cache device */ 437 boolean_t vdev_copy_uberblocks; /* post expand copy uberblocks */ 438 boolean_t vdev_resilver_deferred; /* resilver deferred */ 439 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */ 440 vdev_cache_t vdev_cache; /* physical block cache */ 441 spa_aux_vdev_t *vdev_aux; /* for l2cache and spares vdevs */ 442 zio_t *vdev_probe_zio; /* root of current probe */ 443 vdev_aux_t vdev_label_aux; /* on-disk aux state */ 444 uint64_t vdev_leaf_zap; 445 hrtime_t vdev_mmp_pending; /* 0 if write finished */ 446 uint64_t vdev_mmp_kstat_id; /* to find kstat entry */ 447 uint64_t vdev_expansion_time; /* vdev's last expansion time */ 448 list_node_t vdev_leaf_node; /* leaf vdev list */ 449 450 /* 451 * For DTrace to work in userland (libzpool) context, these fields must 452 * remain at the end of the structure. DTrace will use the kernel's 453 * CTF definition for 'struct vdev', and since the size of a kmutex_t is 454 * larger in userland, the offsets for the rest of the fields would be 455 * incorrect. 456 */ 457 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */ 458 kmutex_t vdev_stat_lock; /* vdev_stat */ 459 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */ 460 461 /* 462 * We rate limit ZIO delay, deadman, and checksum events, since they 463 * can flood ZED with tons of events when a drive is acting up. 464 */ 465 zfs_ratelimit_t vdev_delay_rl; 466 zfs_ratelimit_t vdev_deadman_rl; 467 zfs_ratelimit_t vdev_checksum_rl; 468 }; 469 470 #define VDEV_PAD_SIZE (8 << 10) 471 /* 2 padding areas (vl_pad1 and vl_be) to skip */ 472 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2 473 #define VDEV_PHYS_SIZE (112 << 10) 474 #define VDEV_UBERBLOCK_RING (128 << 10) 475 476 /* 477 * MMP blocks occupy the last MMP_BLOCKS_PER_LABEL slots in the uberblock 478 * ring when MMP is enabled. 479 */ 480 #define MMP_BLOCKS_PER_LABEL 1 481 482 /* The largest uberblock we support is 8k. */ 483 #define MAX_UBERBLOCK_SHIFT (13) 484 #define VDEV_UBERBLOCK_SHIFT(vd) \ 485 MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \ 486 MAX_UBERBLOCK_SHIFT) 487 #define VDEV_UBERBLOCK_COUNT(vd) \ 488 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd)) 489 #define VDEV_UBERBLOCK_OFFSET(vd, n) \ 490 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)]) 491 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd)) 492 493 typedef struct vdev_phys { 494 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)]; 495 zio_eck_t vp_zbt; 496 } vdev_phys_t; 497 498 typedef enum vbe_vers { 499 /* 500 * The bootenv file is stored as ascii text in the envblock. 501 * It is used by the GRUB bootloader used on Linux to store the 502 * contents of the grubenv file. The file is stored as raw ASCII, 503 * and is protected by an embedded checksum. By default, GRUB will 504 * check if the boot filesystem supports storing the environment data 505 * in a special location, and if so, will invoke filesystem specific 506 * logic to retrieve it. This can be overridden by a variable, should 507 * the user so desire. 508 */ 509 VB_RAW = 0, 510 511 /* 512 * The bootenv file is converted to an nvlist and then packed into the 513 * envblock. 514 */ 515 VB_NVLIST = 1 516 } vbe_vers_t; 517 518 typedef struct vdev_boot_envblock { 519 uint64_t vbe_version; 520 char vbe_bootenv[VDEV_PAD_SIZE - sizeof (uint64_t) - 521 sizeof (zio_eck_t)]; 522 zio_eck_t vbe_zbt; 523 } vdev_boot_envblock_t; 524 _Static_assert(sizeof (vdev_boot_envblock_t) == VDEV_PAD_SIZE, 525 "vdev_boot_envblock_t wrong size"); 526 527 typedef struct vdev_label { 528 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */ 529 vdev_boot_envblock_t vl_be; /* 8K */ 530 vdev_phys_t vl_vdev_phys; /* 112K */ 531 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */ 532 } vdev_label_t; /* 256K total */ 533 534 /* 535 * vdev_dirty() flags 536 */ 537 #define VDD_METASLAB 0x01 538 #define VDD_DTL 0x02 539 540 /* Offset of embedded boot loader region on each label */ 541 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t)) 542 /* 543 * Size of embedded boot loader region on each label. 544 * The total size of the first two labels plus the boot area is 4MB. 545 */ 546 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */ 547 548 /* 549 * Size of label regions at the start and end of each leaf device. 550 */ 551 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE) 552 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t)) 553 #define VDEV_LABELS 4 554 #define VDEV_BEST_LABEL VDEV_LABELS 555 #define VDEV_OFFSET_IS_LABEL(vd, off) \ 556 (((off) < VDEV_LABEL_START_SIZE) || \ 557 ((off) >= ((vd)->vdev_psize - VDEV_LABEL_END_SIZE))) 558 559 #define VDEV_ALLOC_LOAD 0 560 #define VDEV_ALLOC_ADD 1 561 #define VDEV_ALLOC_SPARE 2 562 #define VDEV_ALLOC_L2CACHE 3 563 #define VDEV_ALLOC_ROOTPOOL 4 564 #define VDEV_ALLOC_SPLIT 5 565 #define VDEV_ALLOC_ATTACH 6 566 567 /* 568 * Allocate or free a vdev 569 */ 570 extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, 571 vdev_ops_t *ops); 572 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config, 573 vdev_t *parent, uint_t id, int alloctype); 574 extern void vdev_free(vdev_t *vd); 575 576 /* 577 * Add or remove children and parents 578 */ 579 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd); 580 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd); 581 extern void vdev_compact_children(vdev_t *pvd); 582 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops); 583 extern void vdev_remove_parent(vdev_t *cvd); 584 585 /* 586 * vdev sync load and sync 587 */ 588 extern boolean_t vdev_log_state_valid(vdev_t *vd); 589 extern int vdev_load(vdev_t *vd); 590 extern int vdev_dtl_load(vdev_t *vd); 591 extern void vdev_sync(vdev_t *vd, uint64_t txg); 592 extern void vdev_sync_done(vdev_t *vd, uint64_t txg); 593 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg); 594 extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg); 595 596 /* 597 * Available vdev types. 598 */ 599 extern vdev_ops_t vdev_root_ops; 600 extern vdev_ops_t vdev_mirror_ops; 601 extern vdev_ops_t vdev_replacing_ops; 602 extern vdev_ops_t vdev_raidz_ops; 603 extern vdev_ops_t vdev_draid_ops; 604 extern vdev_ops_t vdev_draid_spare_ops; 605 extern vdev_ops_t vdev_disk_ops; 606 extern vdev_ops_t vdev_file_ops; 607 extern vdev_ops_t vdev_missing_ops; 608 extern vdev_ops_t vdev_hole_ops; 609 extern vdev_ops_t vdev_spare_ops; 610 extern vdev_ops_t vdev_indirect_ops; 611 612 /* 613 * Common size functions 614 */ 615 extern void vdev_default_xlate(vdev_t *vd, const range_seg64_t *logical_rs, 616 range_seg64_t *physical_rs, range_seg64_t *remain_rs); 617 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize); 618 extern uint64_t vdev_default_min_asize(vdev_t *vd); 619 extern uint64_t vdev_get_min_asize(vdev_t *vd); 620 extern void vdev_set_min_asize(vdev_t *vd); 621 extern uint64_t vdev_get_min_alloc(vdev_t *vd); 622 extern uint64_t vdev_get_nparity(vdev_t *vd); 623 extern uint64_t vdev_get_ndisks(vdev_t *vd); 624 625 /* 626 * Global variables 627 */ 628 extern int zfs_vdev_standard_sm_blksz; 629 630 /* 631 * Functions from vdev_indirect.c 632 */ 633 extern void vdev_indirect_sync_obsolete(vdev_t *vd, dmu_tx_t *tx); 634 extern boolean_t vdev_indirect_should_condense(vdev_t *vd); 635 extern void spa_condense_indirect_start_sync(vdev_t *vd, dmu_tx_t *tx); 636 extern int vdev_obsolete_sm_object(vdev_t *vd, uint64_t *sm_obj); 637 extern int vdev_obsolete_counts_are_precise(vdev_t *vd, boolean_t *are_precise); 638 639 /* 640 * Other miscellaneous functions 641 */ 642 int vdev_checkpoint_sm_object(vdev_t *vd, uint64_t *sm_obj); 643 void vdev_metaslab_group_create(vdev_t *vd); 644 645 /* 646 * Vdev ashift optimization tunables 647 */ 648 extern uint64_t zfs_vdev_min_auto_ashift; 649 extern uint64_t zfs_vdev_max_auto_ashift; 650 int param_set_min_auto_ashift(ZFS_MODULE_PARAM_ARGS); 651 int param_set_max_auto_ashift(ZFS_MODULE_PARAM_ARGS); 652 653 #ifdef __cplusplus 654 } 655 #endif 656 657 #endif /* _SYS_VDEV_IMPL_H */ 658