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 https://opensource.org/licenses/CDDL-1.0. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. 24 * Copyright 2011 Nexenta Systems, Inc. All rights reserved. 25 * Copyright (c) 2012, 2024 by Delphix. All rights reserved. 26 * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. 27 * Copyright (c) 2013, Joyent, Inc. All rights reserved. 28 * Copyright 2016 Toomas Soome <tsoome@me.com> 29 * Copyright (c) 2019, Allan Jude 30 * Copyright (c) 2019, 2023, 2024, Klara Inc. 31 * Copyright (c) 2019-2020, Michael Niewöhner 32 */ 33 34 #ifndef _ZIO_H 35 #define _ZIO_H 36 37 #include <sys/zio_priority.h> 38 #include <sys/zfs_context.h> 39 #include <sys/spa.h> 40 #include <sys/txg.h> 41 #include <sys/avl.h> 42 #include <sys/fs/zfs.h> 43 #include <sys/zio_impl.h> 44 45 #ifdef __cplusplus 46 extern "C" { 47 #endif 48 49 /* 50 * Embedded checksum 51 */ 52 #define ZEC_MAGIC 0x210da7ab10c7a11ULL 53 54 typedef struct zio_eck { 55 uint64_t zec_magic; /* for validation, endianness */ 56 zio_cksum_t zec_cksum; /* 256-bit checksum */ 57 } zio_eck_t; 58 59 /* 60 * Gang block headers are self-checksumming and contain an array 61 * of block pointers. 62 */ 63 #define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE 64 #define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \ 65 sizeof (zio_eck_t)) / sizeof (blkptr_t)) 66 #define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \ 67 sizeof (zio_eck_t) - \ 68 (SPA_GBH_NBLKPTRS * sizeof (blkptr_t))) /\ 69 sizeof (uint64_t)) 70 71 typedef struct zio_gbh { 72 blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS]; 73 uint64_t zg_filler[SPA_GBH_FILLER]; 74 zio_eck_t zg_tail; 75 } zio_gbh_phys_t; 76 77 enum zio_checksum { 78 ZIO_CHECKSUM_INHERIT = 0, 79 ZIO_CHECKSUM_ON, 80 ZIO_CHECKSUM_OFF, 81 ZIO_CHECKSUM_LABEL, 82 ZIO_CHECKSUM_GANG_HEADER, 83 ZIO_CHECKSUM_ZILOG, 84 ZIO_CHECKSUM_FLETCHER_2, 85 ZIO_CHECKSUM_FLETCHER_4, 86 ZIO_CHECKSUM_SHA256, 87 ZIO_CHECKSUM_ZILOG2, 88 ZIO_CHECKSUM_NOPARITY, 89 ZIO_CHECKSUM_SHA512, 90 ZIO_CHECKSUM_SKEIN, 91 ZIO_CHECKSUM_EDONR, 92 ZIO_CHECKSUM_BLAKE3, 93 ZIO_CHECKSUM_FUNCTIONS 94 }; 95 96 /* 97 * The number of "legacy" compression functions which can be set on individual 98 * objects. 99 */ 100 #define ZIO_CHECKSUM_LEGACY_FUNCTIONS ZIO_CHECKSUM_ZILOG2 101 102 #define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_4 103 #define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON 104 105 #define ZIO_CHECKSUM_MASK 0xffULL 106 #define ZIO_CHECKSUM_VERIFY (1U << 8) 107 108 #define ZIO_DEDUPCHECKSUM ZIO_CHECKSUM_SHA256 109 110 /* macros defining encryption lengths */ 111 #define ZIO_OBJSET_MAC_LEN 32 112 #define ZIO_DATA_IV_LEN 12 113 #define ZIO_DATA_SALT_LEN 8 114 #define ZIO_DATA_MAC_LEN 16 115 116 /* 117 * The number of "legacy" compression functions which can be set on individual 118 * objects. 119 */ 120 #define ZIO_COMPRESS_LEGACY_FUNCTIONS ZIO_COMPRESS_LZ4 121 122 /* 123 * The meaning of "compress = on" selected by the compression features enabled 124 * on a given pool. 125 */ 126 #define ZIO_COMPRESS_LEGACY_ON_VALUE ZIO_COMPRESS_LZJB 127 #define ZIO_COMPRESS_LZ4_ON_VALUE ZIO_COMPRESS_LZ4 128 129 #define ZIO_COMPRESS_DEFAULT ZIO_COMPRESS_ON 130 131 #define BOOTFS_COMPRESS_VALID(compress) \ 132 ((compress) == ZIO_COMPRESS_LZJB || \ 133 (compress) == ZIO_COMPRESS_LZ4 || \ 134 (compress) == ZIO_COMPRESS_GZIP_1 || \ 135 (compress) == ZIO_COMPRESS_GZIP_2 || \ 136 (compress) == ZIO_COMPRESS_GZIP_3 || \ 137 (compress) == ZIO_COMPRESS_GZIP_4 || \ 138 (compress) == ZIO_COMPRESS_GZIP_5 || \ 139 (compress) == ZIO_COMPRESS_GZIP_6 || \ 140 (compress) == ZIO_COMPRESS_GZIP_7 || \ 141 (compress) == ZIO_COMPRESS_GZIP_8 || \ 142 (compress) == ZIO_COMPRESS_GZIP_9 || \ 143 (compress) == ZIO_COMPRESS_ZLE || \ 144 (compress) == ZIO_COMPRESS_ZSTD || \ 145 (compress) == ZIO_COMPRESS_ON || \ 146 (compress) == ZIO_COMPRESS_OFF) 147 148 149 #define ZIO_COMPRESS_ALGO(x) (x & SPA_COMPRESSMASK) 150 #define ZIO_COMPRESS_LEVEL(x) ((x & ~SPA_COMPRESSMASK) >> SPA_COMPRESSBITS) 151 #define ZIO_COMPRESS_RAW(type, level) (type | ((level) << SPA_COMPRESSBITS)) 152 153 #define ZIO_COMPLEVEL_ZSTD(level) \ 154 ZIO_COMPRESS_RAW(ZIO_COMPRESS_ZSTD, level) 155 156 #define ZIO_FAILURE_MODE_WAIT 0 157 #define ZIO_FAILURE_MODE_CONTINUE 1 158 #define ZIO_FAILURE_MODE_PANIC 2 159 160 typedef enum zio_suspend_reason { 161 ZIO_SUSPEND_NONE = 0, 162 ZIO_SUSPEND_IOERR, 163 ZIO_SUSPEND_MMP, 164 } zio_suspend_reason_t; 165 166 /* 167 * This was originally an enum type. However, those are 32-bit and there is no 168 * way to make a 64-bit enum type. Since we ran out of bits for flags, we were 169 * forced to upgrade it to a uint64_t. 170 */ 171 typedef uint64_t zio_flag_t; 172 /* 173 * Flags inherited by gang, ddt, and vdev children, 174 * and that must be equal for two zios to aggregate 175 */ 176 #define ZIO_FLAG_DONT_AGGREGATE (1ULL << 0) 177 #define ZIO_FLAG_IO_REPAIR (1ULL << 1) 178 #define ZIO_FLAG_SELF_HEAL (1ULL << 2) 179 #define ZIO_FLAG_RESILVER (1ULL << 3) 180 #define ZIO_FLAG_SCRUB (1ULL << 4) 181 #define ZIO_FLAG_SCAN_THREAD (1ULL << 5) 182 #define ZIO_FLAG_PHYSICAL (1ULL << 6) 183 184 #define ZIO_FLAG_AGG_INHERIT (ZIO_FLAG_CANFAIL - 1) 185 186 /* 187 * Flags inherited by ddt, gang, and vdev children. 188 */ 189 #define ZIO_FLAG_CANFAIL (1ULL << 7) /* must be first for INHERIT */ 190 #define ZIO_FLAG_SPECULATIVE (1ULL << 8) 191 #define ZIO_FLAG_CONFIG_WRITER (1ULL << 9) 192 #define ZIO_FLAG_DONT_RETRY (1ULL << 10) 193 #define ZIO_FLAG_NODATA (1ULL << 12) 194 #define ZIO_FLAG_INDUCE_DAMAGE (1ULL << 13) 195 #define ZIO_FLAG_IO_ALLOCATING (1ULL << 14) 196 197 #define ZIO_FLAG_DDT_INHERIT (ZIO_FLAG_IO_RETRY - 1) 198 #define ZIO_FLAG_GANG_INHERIT (ZIO_FLAG_IO_RETRY - 1) 199 200 /* 201 * Flags inherited by vdev children. 202 */ 203 #define ZIO_FLAG_IO_RETRY (1ULL << 15) /* must be first for INHERIT */ 204 #define ZIO_FLAG_PROBE (1ULL << 16) 205 #define ZIO_FLAG_TRYHARD (1ULL << 17) 206 #define ZIO_FLAG_OPTIONAL (1ULL << 18) 207 208 #define ZIO_FLAG_VDEV_INHERIT (ZIO_FLAG_DONT_QUEUE - 1) 209 210 /* 211 * Flags not inherited by any children. 212 */ 213 #define ZIO_FLAG_DONT_QUEUE (1ULL << 19) /* must be first for INHERIT */ 214 #define ZIO_FLAG_DONT_PROPAGATE (1ULL << 20) 215 #define ZIO_FLAG_IO_BYPASS (1ULL << 21) 216 #define ZIO_FLAG_IO_REWRITE (1ULL << 22) 217 #define ZIO_FLAG_RAW_COMPRESS (1ULL << 23) 218 #define ZIO_FLAG_RAW_ENCRYPT (1ULL << 24) 219 #define ZIO_FLAG_GANG_CHILD (1ULL << 25) 220 #define ZIO_FLAG_DDT_CHILD (1ULL << 26) 221 #define ZIO_FLAG_GODFATHER (1ULL << 27) 222 #define ZIO_FLAG_NOPWRITE (1ULL << 28) 223 #define ZIO_FLAG_REEXECUTED (1ULL << 29) 224 #define ZIO_FLAG_DELEGATED (1ULL << 30) 225 226 #define ZIO_ALLOCATOR_NONE (-1) 227 #define ZIO_HAS_ALLOCATOR(zio) ((zio)->io_allocator != ZIO_ALLOCATOR_NONE) 228 229 #define ZIO_FLAG_MUSTSUCCEED 0 230 #define ZIO_FLAG_RAW (ZIO_FLAG_RAW_COMPRESS | ZIO_FLAG_RAW_ENCRYPT) 231 232 #define ZIO_DDT_CHILD_FLAGS(zio) \ 233 (((zio)->io_flags & ZIO_FLAG_DDT_INHERIT) | \ 234 ZIO_FLAG_DDT_CHILD | ZIO_FLAG_CANFAIL) 235 236 #define ZIO_GANG_CHILD_FLAGS(zio) \ 237 (((zio)->io_flags & ZIO_FLAG_GANG_INHERIT) | \ 238 ZIO_FLAG_GANG_CHILD | ZIO_FLAG_CANFAIL) 239 240 #define ZIO_VDEV_CHILD_FLAGS(zio) \ 241 (((zio)->io_flags & ZIO_FLAG_VDEV_INHERIT) | \ 242 ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_CANFAIL) 243 244 #define ZIO_CHILD_BIT(x) (1U << (x)) 245 #define ZIO_CHILD_BIT_IS_SET(val, x) ((val) & (1U << (x))) 246 247 enum zio_child { 248 ZIO_CHILD_VDEV = 0, 249 ZIO_CHILD_GANG, 250 ZIO_CHILD_DDT, 251 ZIO_CHILD_LOGICAL, 252 ZIO_CHILD_TYPES 253 }; 254 255 #define ZIO_CHILD_VDEV_BIT ZIO_CHILD_BIT(ZIO_CHILD_VDEV) 256 #define ZIO_CHILD_GANG_BIT ZIO_CHILD_BIT(ZIO_CHILD_GANG) 257 #define ZIO_CHILD_DDT_BIT ZIO_CHILD_BIT(ZIO_CHILD_DDT) 258 #define ZIO_CHILD_LOGICAL_BIT ZIO_CHILD_BIT(ZIO_CHILD_LOGICAL) 259 #define ZIO_CHILD_ALL_BITS \ 260 (ZIO_CHILD_VDEV_BIT | ZIO_CHILD_GANG_BIT | \ 261 ZIO_CHILD_DDT_BIT | ZIO_CHILD_LOGICAL_BIT) 262 263 enum zio_wait_type { 264 ZIO_WAIT_READY = 0, 265 ZIO_WAIT_DONE, 266 ZIO_WAIT_TYPES 267 }; 268 269 typedef void zio_done_func_t(zio_t *zio); 270 271 extern int zio_exclude_metadata; 272 extern int zio_dva_throttle_enabled; 273 extern const char *const zio_type_name[ZIO_TYPES]; 274 275 /* 276 * A bookmark is a four-tuple <objset, object, level, blkid> that uniquely 277 * identifies any block in the pool. By convention, the meta-objset (MOS) 278 * is objset 0, and the meta-dnode is object 0. This covers all blocks 279 * except root blocks and ZIL blocks, which are defined as follows: 280 * 281 * Root blocks (objset_phys_t) are object 0, level -1: <objset, 0, -1, 0>. 282 * ZIL blocks are bookmarked <objset, 0, -2, blkid == ZIL sequence number>. 283 * dmu_sync()ed ZIL data blocks are bookmarked <objset, object, -2, blkid>. 284 * dnode visit bookmarks are <objset, object id of dnode, -3, 0>. 285 * 286 * Note: this structure is called a bookmark because its original purpose 287 * was to remember where to resume a pool-wide traverse. 288 * 289 * Note: this structure is passed between userland and the kernel, and is 290 * stored on disk (by virtue of being incorporated into other on-disk 291 * structures, e.g. dsl_scan_phys_t). 292 * 293 * If the head_errlog feature is enabled a different on-disk format for error 294 * logs is used. This introduces the use of an error bookmark, a four-tuple 295 * <object, level, blkid, birth> that uniquely identifies any error block 296 * in the pool. The birth transaction group is used to track whether the block 297 * has been overwritten by newer data or added to a snapshot since its marking 298 * as an error. 299 */ 300 struct zbookmark_phys { 301 uint64_t zb_objset; 302 uint64_t zb_object; 303 int64_t zb_level; 304 uint64_t zb_blkid; 305 }; 306 307 struct zbookmark_err_phys { 308 uint64_t zb_object; 309 int64_t zb_level; 310 uint64_t zb_blkid; 311 uint64_t zb_birth; 312 }; 313 314 #define SET_BOOKMARK(zb, objset, object, level, blkid) \ 315 { \ 316 (zb)->zb_objset = objset; \ 317 (zb)->zb_object = object; \ 318 (zb)->zb_level = level; \ 319 (zb)->zb_blkid = blkid; \ 320 } 321 322 #define ZB_DESTROYED_OBJSET (-1ULL) 323 324 #define ZB_ROOT_OBJECT (0ULL) 325 #define ZB_ROOT_LEVEL (-1LL) 326 #define ZB_ROOT_BLKID (0ULL) 327 328 #define ZB_ZIL_OBJECT (0ULL) 329 #define ZB_ZIL_LEVEL (-2LL) 330 331 #define ZB_DNODE_LEVEL (-3LL) 332 #define ZB_DNODE_BLKID (0ULL) 333 334 #define ZB_IS_ZERO(zb) \ 335 ((zb)->zb_objset == 0 && (zb)->zb_object == 0 && \ 336 (zb)->zb_level == 0 && (zb)->zb_blkid == 0) 337 #define ZB_IS_ROOT(zb) \ 338 ((zb)->zb_object == ZB_ROOT_OBJECT && \ 339 (zb)->zb_level == ZB_ROOT_LEVEL && \ 340 (zb)->zb_blkid == ZB_ROOT_BLKID) 341 342 typedef struct zio_prop { 343 enum zio_checksum zp_checksum; 344 enum zio_compress zp_compress; 345 uint8_t zp_complevel; 346 uint8_t zp_level; 347 uint8_t zp_copies; 348 dmu_object_type_t zp_type; 349 boolean_t zp_dedup; 350 boolean_t zp_dedup_verify; 351 boolean_t zp_nopwrite; 352 boolean_t zp_brtwrite; 353 boolean_t zp_encrypt; 354 boolean_t zp_byteorder; 355 uint8_t zp_salt[ZIO_DATA_SALT_LEN]; 356 uint8_t zp_iv[ZIO_DATA_IV_LEN]; 357 uint8_t zp_mac[ZIO_DATA_MAC_LEN]; 358 uint32_t zp_zpl_smallblk; 359 } zio_prop_t; 360 361 typedef struct zio_cksum_report zio_cksum_report_t; 362 363 typedef void zio_cksum_finish_f(zio_cksum_report_t *rep, 364 const abd_t *good_data); 365 typedef void zio_cksum_free_f(void *cbdata, size_t size); 366 367 struct zio_bad_cksum; /* defined in zio_checksum.h */ 368 struct dnode_phys; 369 struct abd; 370 371 struct zio_cksum_report { 372 struct zio_cksum_report *zcr_next; 373 nvlist_t *zcr_ereport; 374 nvlist_t *zcr_detector; 375 void *zcr_cbdata; 376 size_t zcr_cbinfo; /* passed to zcr_free() */ 377 uint64_t zcr_sector; 378 uint64_t zcr_align; 379 uint64_t zcr_length; 380 zio_cksum_finish_f *zcr_finish; 381 zio_cksum_free_f *zcr_free; 382 383 /* internal use only */ 384 struct zio_bad_cksum *zcr_ckinfo; /* information from failure */ 385 }; 386 387 typedef struct zio_vsd_ops { 388 zio_done_func_t *vsd_free; 389 } zio_vsd_ops_t; 390 391 typedef struct zio_gang_node { 392 zio_gbh_phys_t *gn_gbh; 393 struct zio_gang_node *gn_child[SPA_GBH_NBLKPTRS]; 394 } zio_gang_node_t; 395 396 typedef zio_t *zio_gang_issue_func_t(zio_t *zio, blkptr_t *bp, 397 zio_gang_node_t *gn, struct abd *data, uint64_t offset); 398 399 typedef void zio_transform_func_t(zio_t *zio, struct abd *data, uint64_t size); 400 401 typedef struct zio_transform { 402 struct abd *zt_orig_abd; 403 uint64_t zt_orig_size; 404 uint64_t zt_bufsize; 405 zio_transform_func_t *zt_transform; 406 struct zio_transform *zt_next; 407 } zio_transform_t; 408 409 typedef zio_t *zio_pipe_stage_t(zio_t *zio); 410 411 /* 412 * The io_reexecute flags are distinct from io_flags because the child must 413 * be able to propagate them to the parent. The normal io_flags are local 414 * to the zio, not protected by any lock, and not modifiable by children; 415 * the reexecute flags are protected by io_lock, modifiable by children, 416 * and always propagated -- even when ZIO_FLAG_DONT_PROPAGATE is set. 417 */ 418 #define ZIO_REEXECUTE_NOW 0x01 419 #define ZIO_REEXECUTE_SUSPEND 0x02 420 421 /* 422 * The io_trim flags are used to specify the type of TRIM to perform. They 423 * only apply to ZIO_TYPE_TRIM zios are distinct from io_flags. 424 */ 425 enum trim_flag { 426 ZIO_TRIM_SECURE = 1U << 0, 427 }; 428 429 typedef struct zio_alloc_list { 430 list_t zal_list; 431 uint64_t zal_size; 432 } zio_alloc_list_t; 433 434 typedef struct zio_link { 435 zio_t *zl_parent; 436 zio_t *zl_child; 437 list_node_t zl_parent_node; 438 list_node_t zl_child_node; 439 } zio_link_t; 440 441 enum zio_qstate { 442 ZIO_QS_NONE = 0, 443 ZIO_QS_QUEUED, 444 ZIO_QS_ACTIVE, 445 }; 446 447 struct zio { 448 /* Core information about this I/O */ 449 zbookmark_phys_t io_bookmark; 450 zio_prop_t io_prop; 451 zio_type_t io_type; 452 enum zio_child io_child_type; 453 enum trim_flag io_trim_flags; 454 zio_priority_t io_priority; 455 uint8_t io_reexecute; 456 uint8_t io_state[ZIO_WAIT_TYPES]; 457 uint64_t io_txg; 458 spa_t *io_spa; 459 blkptr_t *io_bp; 460 blkptr_t *io_bp_override; 461 blkptr_t io_bp_copy; 462 list_t io_parent_list; 463 list_t io_child_list; 464 zio_t *io_logical; 465 zio_transform_t *io_transform_stack; 466 467 /* Callback info */ 468 zio_done_func_t *io_ready; 469 zio_done_func_t *io_children_ready; 470 zio_done_func_t *io_done; 471 void *io_private; 472 int64_t io_prev_space_delta; /* DMU private */ 473 blkptr_t io_bp_orig; 474 /* io_lsize != io_orig_size iff this is a raw write */ 475 uint64_t io_lsize; 476 477 /* Data represented by this I/O */ 478 struct abd *io_abd; 479 struct abd *io_orig_abd; 480 uint64_t io_size; 481 uint64_t io_orig_size; 482 483 /* Stuff for the vdev stack */ 484 vdev_t *io_vd; 485 void *io_vsd; 486 const zio_vsd_ops_t *io_vsd_ops; 487 metaslab_class_t *io_metaslab_class; /* dva throttle class */ 488 489 enum zio_qstate io_queue_state; /* vdev queue state */ 490 union { 491 list_node_t l; 492 avl_node_t a; 493 } io_queue_node ____cacheline_aligned; /* allocator and vdev queues */ 494 avl_node_t io_offset_node; /* vdev offset queues */ 495 uint64_t io_offset; 496 hrtime_t io_timestamp; /* submitted at */ 497 hrtime_t io_queued_timestamp; 498 hrtime_t io_target_timestamp; 499 hrtime_t io_delta; /* vdev queue service delta */ 500 hrtime_t io_delay; /* Device access time (disk or */ 501 /* file). */ 502 zio_alloc_list_t io_alloc_list; 503 504 /* Internal pipeline state */ 505 zio_flag_t io_flags; 506 enum zio_stage io_stage; 507 enum zio_stage io_pipeline; 508 zio_flag_t io_orig_flags; 509 enum zio_stage io_orig_stage; 510 enum zio_stage io_orig_pipeline; 511 enum zio_stage io_pipeline_trace; 512 int io_error; 513 int io_child_error[ZIO_CHILD_TYPES]; 514 uint64_t io_children[ZIO_CHILD_TYPES][ZIO_WAIT_TYPES]; 515 uint64_t *io_stall; 516 zio_t *io_gang_leader; 517 zio_gang_node_t *io_gang_tree; 518 void *io_executor; 519 void *io_waiter; 520 void *io_bio; 521 kmutex_t io_lock; 522 kcondvar_t io_cv; 523 int io_allocator; 524 525 /* FMA state */ 526 zio_cksum_report_t *io_cksum_report; 527 uint64_t io_ena; 528 529 /* Taskq dispatching state */ 530 taskq_ent_t io_tqent; 531 }; 532 533 enum blk_verify_flag { 534 BLK_VERIFY_ONLY, 535 BLK_VERIFY_LOG, 536 BLK_VERIFY_HALT 537 }; 538 539 enum blk_config_flag { 540 BLK_CONFIG_HELD, // SCL_VDEV held for writer 541 BLK_CONFIG_NEEDED, // SCL_VDEV should be obtained for reader 542 BLK_CONFIG_SKIP, // skip checks which require SCL_VDEV 543 }; 544 545 extern int zio_bookmark_compare(const void *, const void *); 546 547 extern zio_t *zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, 548 zio_done_func_t *done, void *priv, zio_flag_t flags); 549 550 extern zio_t *zio_root(spa_t *spa, 551 zio_done_func_t *done, void *priv, zio_flag_t flags); 552 553 extern void zio_destroy(zio_t *zio); 554 555 extern zio_t *zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, 556 struct abd *data, uint64_t lsize, zio_done_func_t *done, void *priv, 557 zio_priority_t priority, zio_flag_t flags, const zbookmark_phys_t *zb); 558 559 extern zio_t *zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 560 struct abd *data, uint64_t size, uint64_t psize, const zio_prop_t *zp, 561 zio_done_func_t *ready, zio_done_func_t *children_ready, 562 zio_done_func_t *done, void *priv, zio_priority_t priority, 563 zio_flag_t flags, const zbookmark_phys_t *zb); 564 565 extern zio_t *zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 566 struct abd *data, uint64_t size, zio_done_func_t *done, void *priv, 567 zio_priority_t priority, zio_flag_t flags, zbookmark_phys_t *zb); 568 569 extern void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, 570 boolean_t nopwrite, boolean_t brtwrite); 571 572 extern void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp); 573 574 extern zio_t *zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, 575 const blkptr_t *bp, 576 zio_done_func_t *done, void *priv, zio_flag_t flags); 577 578 extern zio_t *zio_trim(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, 579 zio_done_func_t *done, void *priv, zio_priority_t priority, 580 zio_flag_t flags, enum trim_flag trim_flags); 581 582 extern zio_t *zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 583 uint64_t size, struct abd *data, int checksum, 584 zio_done_func_t *done, void *priv, zio_priority_t priority, 585 zio_flag_t flags, boolean_t labels); 586 587 extern zio_t *zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 588 uint64_t size, struct abd *data, int checksum, 589 zio_done_func_t *done, void *priv, zio_priority_t priority, 590 zio_flag_t flags, boolean_t labels); 591 592 extern zio_t *zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, 593 const blkptr_t *bp, zio_flag_t flags); 594 595 extern int zio_alloc_zil(spa_t *spa, objset_t *os, uint64_t txg, 596 blkptr_t *new_bp, uint64_t size, boolean_t *slog); 597 extern void zio_flush(zio_t *zio, vdev_t *vd); 598 extern void zio_shrink(zio_t *zio, uint64_t size); 599 600 extern int zio_wait(zio_t *zio); 601 extern void zio_nowait(zio_t *zio); 602 extern void zio_execute(void *zio); 603 extern void zio_interrupt(void *zio); 604 extern void zio_delay_init(zio_t *zio); 605 extern void zio_delay_interrupt(zio_t *zio); 606 extern void zio_deadman(zio_t *zio, const char *tag); 607 608 extern zio_t *zio_walk_parents(zio_t *cio, zio_link_t **); 609 extern zio_t *zio_walk_children(zio_t *pio, zio_link_t **); 610 extern zio_t *zio_unique_parent(zio_t *cio); 611 extern void zio_add_child(zio_t *pio, zio_t *cio); 612 extern void zio_add_child_first(zio_t *pio, zio_t *cio); 613 614 extern void *zio_buf_alloc(size_t size); 615 extern void zio_buf_free(void *buf, size_t size); 616 extern void *zio_data_buf_alloc(size_t size); 617 extern void zio_data_buf_free(void *buf, size_t size); 618 619 extern void zio_push_transform(zio_t *zio, struct abd *abd, uint64_t size, 620 uint64_t bufsize, zio_transform_func_t *transform); 621 extern void zio_pop_transforms(zio_t *zio); 622 623 extern void zio_resubmit_stage_async(void *); 624 625 extern zio_t *zio_vdev_child_io(zio_t *zio, blkptr_t *bp, vdev_t *vd, 626 uint64_t offset, struct abd *data, uint64_t size, int type, 627 zio_priority_t priority, zio_flag_t flags, 628 zio_done_func_t *done, void *priv); 629 630 extern zio_t *zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, 631 struct abd *data, uint64_t size, zio_type_t type, zio_priority_t priority, 632 zio_flag_t flags, zio_done_func_t *done, void *priv); 633 634 extern void zio_vdev_io_bypass(zio_t *zio); 635 extern void zio_vdev_io_reissue(zio_t *zio); 636 extern void zio_vdev_io_redone(zio_t *zio); 637 638 extern void zio_change_priority(zio_t *pio, zio_priority_t priority); 639 640 extern void zio_checksum_verified(zio_t *zio); 641 extern int zio_worst_error(int e1, int e2); 642 643 extern enum zio_checksum zio_checksum_select(enum zio_checksum child, 644 enum zio_checksum parent); 645 extern enum zio_checksum zio_checksum_dedup_select(spa_t *spa, 646 enum zio_checksum child, enum zio_checksum parent); 647 extern enum zio_compress zio_compress_select(spa_t *spa, 648 enum zio_compress child, enum zio_compress parent); 649 extern uint8_t zio_complevel_select(spa_t *spa, enum zio_compress compress, 650 uint8_t child, uint8_t parent); 651 652 extern void zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t); 653 extern int zio_resume(spa_t *spa); 654 extern void zio_resume_wait(spa_t *spa); 655 656 extern boolean_t zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp, 657 enum blk_config_flag blk_config, enum blk_verify_flag blk_verify); 658 659 /* 660 * Initial setup and teardown. 661 */ 662 extern void zio_init(void); 663 extern void zio_fini(void); 664 665 /* 666 * Fault injection 667 */ 668 struct zinject_record; 669 extern uint32_t zio_injection_enabled; 670 extern int zio_inject_fault(char *name, int flags, int *id, 671 struct zinject_record *record); 672 extern int zio_inject_list_next(int *id, char *name, size_t buflen, 673 struct zinject_record *record); 674 extern int zio_clear_fault(int id); 675 extern void zio_handle_panic_injection(spa_t *spa, const char *tag, 676 uint64_t type); 677 extern int zio_handle_decrypt_injection(spa_t *spa, const zbookmark_phys_t *zb, 678 uint64_t type, int error); 679 extern int zio_handle_fault_injection(zio_t *zio, int error); 680 extern int zio_handle_device_injection(vdev_t *vd, zio_t *zio, int error); 681 extern int zio_handle_device_injections(vdev_t *vd, zio_t *zio, int err1, 682 int err2); 683 extern int zio_handle_label_injection(zio_t *zio, int error); 684 extern void zio_handle_ignored_writes(zio_t *zio); 685 extern hrtime_t zio_handle_io_delay(zio_t *zio); 686 extern void zio_handle_import_delay(spa_t *spa, hrtime_t elapsed); 687 extern void zio_handle_export_delay(spa_t *spa, hrtime_t elapsed); 688 689 /* 690 * Checksum ereport functions 691 */ 692 extern int zfs_ereport_start_checksum(spa_t *spa, vdev_t *vd, 693 const zbookmark_phys_t *zb, struct zio *zio, uint64_t offset, 694 uint64_t length, struct zio_bad_cksum *info); 695 extern void zfs_ereport_finish_checksum(zio_cksum_report_t *report, 696 const abd_t *good_data, const abd_t *bad_data, boolean_t drop_if_identical); 697 698 extern void zfs_ereport_free_checksum(zio_cksum_report_t *report); 699 700 /* If we have the good data in hand, this function can be used */ 701 extern int zfs_ereport_post_checksum(spa_t *spa, vdev_t *vd, 702 const zbookmark_phys_t *zb, struct zio *zio, uint64_t offset, 703 uint64_t length, const abd_t *good_data, const abd_t *bad_data, 704 struct zio_bad_cksum *info); 705 706 void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr); 707 extern void zfs_ereport_snapshot_post(const char *subclass, spa_t *spa, 708 const char *name); 709 710 /* Called from spa_sync(), but primarily an injection handler */ 711 extern void spa_handle_ignored_writes(spa_t *spa); 712 713 /* zbookmark_phys functions */ 714 boolean_t zbookmark_subtree_completed(const struct dnode_phys *dnp, 715 const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block); 716 boolean_t zbookmark_subtree_tbd(const struct dnode_phys *dnp, 717 const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block); 718 int zbookmark_compare(uint16_t dbss1, uint8_t ibs1, uint16_t dbss2, 719 uint8_t ibs2, const zbookmark_phys_t *zb1, const zbookmark_phys_t *zb2); 720 721 #ifdef __cplusplus 722 } 723 #endif 724 725 #endif /* _ZIO_H */ 726