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 2010 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _SYS_DMU_H 27 #define _SYS_DMU_H 28 29 /* 30 * This file describes the interface that the DMU provides for its 31 * consumers. 32 * 33 * The DMU also interacts with the SPA. That interface is described in 34 * dmu_spa.h. 35 */ 36 37 #include <sys/inttypes.h> 38 #include <sys/types.h> 39 #include <sys/param.h> 40 #include <sys/cred.h> 41 #include <sys/time.h> 42 43 #ifdef __cplusplus 44 extern "C" { 45 #endif 46 47 struct uio; 48 struct xuio; 49 struct page; 50 struct vnode; 51 struct spa; 52 struct zilog; 53 struct zio; 54 struct blkptr; 55 struct zap_cursor; 56 struct dsl_dataset; 57 struct dsl_pool; 58 struct dnode; 59 struct drr_begin; 60 struct drr_end; 61 struct zbookmark; 62 struct spa; 63 struct nvlist; 64 struct arc_buf; 65 struct zio_prop; 66 67 typedef struct objset objset_t; 68 typedef struct dmu_tx dmu_tx_t; 69 typedef struct dsl_dir dsl_dir_t; 70 71 typedef enum dmu_object_type { 72 DMU_OT_NONE, 73 /* general: */ 74 DMU_OT_OBJECT_DIRECTORY, /* ZAP */ 75 DMU_OT_OBJECT_ARRAY, /* UINT64 */ 76 DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ 77 DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ 78 DMU_OT_BPLIST, /* UINT64 */ 79 DMU_OT_BPLIST_HDR, /* UINT64 */ 80 /* spa: */ 81 DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ 82 DMU_OT_SPACE_MAP, /* UINT64 */ 83 /* zil: */ 84 DMU_OT_INTENT_LOG, /* UINT64 */ 85 /* dmu: */ 86 DMU_OT_DNODE, /* DNODE */ 87 DMU_OT_OBJSET, /* OBJSET */ 88 /* dsl: */ 89 DMU_OT_DSL_DIR, /* UINT64 */ 90 DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ 91 DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ 92 DMU_OT_DSL_PROPS, /* ZAP */ 93 DMU_OT_DSL_DATASET, /* UINT64 */ 94 /* zpl: */ 95 DMU_OT_ZNODE, /* ZNODE */ 96 DMU_OT_OLDACL, /* Old ACL */ 97 DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ 98 DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ 99 DMU_OT_MASTER_NODE, /* ZAP */ 100 DMU_OT_UNLINKED_SET, /* ZAP */ 101 /* zvol: */ 102 DMU_OT_ZVOL, /* UINT8 */ 103 DMU_OT_ZVOL_PROP, /* ZAP */ 104 /* other; for testing only! */ 105 DMU_OT_PLAIN_OTHER, /* UINT8 */ 106 DMU_OT_UINT64_OTHER, /* UINT64 */ 107 DMU_OT_ZAP_OTHER, /* ZAP */ 108 /* new object types: */ 109 DMU_OT_ERROR_LOG, /* ZAP */ 110 DMU_OT_SPA_HISTORY, /* UINT8 */ 111 DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ 112 DMU_OT_POOL_PROPS, /* ZAP */ 113 DMU_OT_DSL_PERMS, /* ZAP */ 114 DMU_OT_ACL, /* ACL */ 115 DMU_OT_SYSACL, /* SYSACL */ 116 DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ 117 DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ 118 DMU_OT_NEXT_CLONES, /* ZAP */ 119 DMU_OT_SCRUB_QUEUE, /* ZAP */ 120 DMU_OT_USERGROUP_USED, /* ZAP */ 121 DMU_OT_USERGROUP_QUOTA, /* ZAP */ 122 DMU_OT_USERREFS, /* ZAP */ 123 DMU_OT_DDT_ZAP, /* ZAP */ 124 DMU_OT_DDT_STATS, /* ZAP */ 125 DMU_OT_NUMTYPES 126 } dmu_object_type_t; 127 128 typedef enum dmu_objset_type { 129 DMU_OST_NONE, 130 DMU_OST_META, 131 DMU_OST_ZFS, 132 DMU_OST_ZVOL, 133 DMU_OST_OTHER, /* For testing only! */ 134 DMU_OST_ANY, /* Be careful! */ 135 DMU_OST_NUMTYPES 136 } dmu_objset_type_t; 137 138 void byteswap_uint64_array(void *buf, size_t size); 139 void byteswap_uint32_array(void *buf, size_t size); 140 void byteswap_uint16_array(void *buf, size_t size); 141 void byteswap_uint8_array(void *buf, size_t size); 142 void zap_byteswap(void *buf, size_t size); 143 void zfs_oldacl_byteswap(void *buf, size_t size); 144 void zfs_acl_byteswap(void *buf, size_t size); 145 void zfs_znode_byteswap(void *buf, size_t size); 146 147 #define DS_FIND_SNAPSHOTS (1<<0) 148 #define DS_FIND_CHILDREN (1<<1) 149 150 /* 151 * The maximum number of bytes that can be accessed as part of one 152 * operation, including metadata. 153 */ 154 #define DMU_MAX_ACCESS (10<<20) /* 10MB */ 155 #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ 156 157 #define DMU_USERUSED_OBJECT (-1ULL) 158 #define DMU_GROUPUSED_OBJECT (-2ULL) 159 #define DMU_DEADLIST_OBJECT (-3ULL) 160 161 /* 162 * Public routines to create, destroy, open, and close objsets. 163 */ 164 int dmu_objset_hold(const char *name, void *tag, objset_t **osp); 165 int dmu_objset_own(const char *name, dmu_objset_type_t type, 166 boolean_t readonly, void *tag, objset_t **osp); 167 void dmu_objset_rele(objset_t *os, void *tag); 168 void dmu_objset_disown(objset_t *os, void *tag); 169 int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); 170 171 int dmu_objset_evict_dbufs(objset_t *os); 172 int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, 173 void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); 174 int dmu_objset_clone(const char *name, struct dsl_dataset *clone_origin, 175 uint64_t flags); 176 int dmu_objset_destroy(const char *name, boolean_t defer); 177 int dmu_snapshots_destroy(char *fsname, char *snapname, boolean_t defer); 178 int dmu_objset_snapshot(char *fsname, char *snapname, struct nvlist *props, 179 boolean_t recursive); 180 int dmu_objset_rename(const char *name, const char *newname, 181 boolean_t recursive); 182 int dmu_objset_find(char *name, int func(const char *, void *), void *arg, 183 int flags); 184 void dmu_objset_byteswap(void *buf, size_t size); 185 186 typedef struct dmu_buf { 187 uint64_t db_object; /* object that this buffer is part of */ 188 uint64_t db_offset; /* byte offset in this object */ 189 uint64_t db_size; /* size of buffer in bytes */ 190 void *db_data; /* data in buffer */ 191 } dmu_buf_t; 192 193 typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); 194 195 /* 196 * The names of zap entries in the DIRECTORY_OBJECT of the MOS. 197 */ 198 #define DMU_POOL_DIRECTORY_OBJECT 1 199 #define DMU_POOL_CONFIG "config" 200 #define DMU_POOL_ROOT_DATASET "root_dataset" 201 #define DMU_POOL_SYNC_BPLIST "sync_bplist" 202 #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" 203 #define DMU_POOL_ERRLOG_LAST "errlog_last" 204 #define DMU_POOL_SPARES "spares" 205 #define DMU_POOL_DEFLATE "deflate" 206 #define DMU_POOL_HISTORY "history" 207 #define DMU_POOL_PROPS "pool_props" 208 #define DMU_POOL_L2CACHE "l2cache" 209 #define DMU_POOL_TMP_USERREFS "tmp_userrefs" 210 #define DMU_POOL_DDT "DDT-%s-%s-%s" 211 #define DMU_POOL_DDT_STATS "DDT-statistics" 212 213 /* 4x8 zbookmark_t */ 214 #define DMU_POOL_SCRUB_BOOKMARK "scrub_bookmark" 215 /* 4x8 ddt_bookmark_t */ 216 #define DMU_POOL_SCRUB_DDT_BOOKMARK "scrub_ddt_bookmark" 217 /* 1x8 max_class */ 218 #define DMU_POOL_SCRUB_DDT_CLASS_MAX "scrub_ddt_class_max" 219 /* 1x8 zap obj DMU_OT_SCRUB_QUEUE */ 220 #define DMU_POOL_SCRUB_QUEUE "scrub_queue" 221 /* 1x8 txg */ 222 #define DMU_POOL_SCRUB_MIN_TXG "scrub_min_txg" 223 /* 1x8 txg */ 224 #define DMU_POOL_SCRUB_MAX_TXG "scrub_max_txg" 225 /* 1x4 enum scrub_func */ 226 #define DMU_POOL_SCRUB_FUNC "scrub_func" 227 /* 1x8 count */ 228 #define DMU_POOL_SCRUB_ERRORS "scrub_errors" 229 230 /* 231 * Allocate an object from this objset. The range of object numbers 232 * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. 233 * 234 * The transaction must be assigned to a txg. The newly allocated 235 * object will be "held" in the transaction (ie. you can modify the 236 * newly allocated object in this transaction). 237 * 238 * dmu_object_alloc() chooses an object and returns it in *objectp. 239 * 240 * dmu_object_claim() allocates a specific object number. If that 241 * number is already allocated, it fails and returns EEXIST. 242 * 243 * Return 0 on success, or ENOSPC or EEXIST as specified above. 244 */ 245 uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, 246 int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 247 int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, 248 int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 249 int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, 250 int blocksize, dmu_object_type_t bonustype, int bonuslen); 251 252 /* 253 * Free an object from this objset. 254 * 255 * The object's data will be freed as well (ie. you don't need to call 256 * dmu_free(object, 0, -1, tx)). 257 * 258 * The object need not be held in the transaction. 259 * 260 * If there are any holds on this object's buffers (via dmu_buf_hold()), 261 * or tx holds on the object (via dmu_tx_hold_object()), you can not 262 * free it; it fails and returns EBUSY. 263 * 264 * If the object is not allocated, it fails and returns ENOENT. 265 * 266 * Return 0 on success, or EBUSY or ENOENT as specified above. 267 */ 268 int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); 269 270 /* 271 * Find the next allocated or free object. 272 * 273 * The objectp parameter is in-out. It will be updated to be the next 274 * object which is allocated. Ignore objects which have not been 275 * modified since txg. 276 * 277 * XXX Can only be called on a objset with no dirty data. 278 * 279 * Returns 0 on success, or ENOENT if there are no more objects. 280 */ 281 int dmu_object_next(objset_t *os, uint64_t *objectp, 282 boolean_t hole, uint64_t txg); 283 284 /* 285 * Set the data blocksize for an object. 286 * 287 * The object cannot have any blocks allcated beyond the first. If 288 * the first block is allocated already, the new size must be greater 289 * than the current block size. If these conditions are not met, 290 * ENOTSUP will be returned. 291 * 292 * Returns 0 on success, or EBUSY if there are any holds on the object 293 * contents, or ENOTSUP as described above. 294 */ 295 int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, 296 int ibs, dmu_tx_t *tx); 297 298 /* 299 * Set the checksum property on a dnode. The new checksum algorithm will 300 * apply to all newly written blocks; existing blocks will not be affected. 301 */ 302 void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, 303 dmu_tx_t *tx); 304 305 /* 306 * Set the compress property on a dnode. The new compression algorithm will 307 * apply to all newly written blocks; existing blocks will not be affected. 308 */ 309 void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, 310 dmu_tx_t *tx); 311 312 /* 313 * Decide how to write a block: checksum, compression, number of copies, etc. 314 */ 315 #define WP_NOFILL 0x1 316 #define WP_DMU_SYNC 0x2 317 318 void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, 319 struct zio_prop *zp); 320 /* 321 * The bonus data is accessed more or less like a regular buffer. 322 * You must dmu_bonus_hold() to get the buffer, which will give you a 323 * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus 324 * data. As with any normal buffer, you must call dmu_buf_read() to 325 * read db_data, dmu_buf_will_dirty() before modifying it, and the 326 * object must be held in an assigned transaction before calling 327 * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus 328 * buffer as well. You must release your hold with dmu_buf_rele(). 329 */ 330 int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 331 int dmu_bonus_max(void); 332 int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); 333 334 /* 335 * Obtain the DMU buffer from the specified object which contains the 336 * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so 337 * that it will remain in memory. You must release the hold with 338 * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your 339 * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. 340 * 341 * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill 342 * on the returned buffer before reading or writing the buffer's 343 * db_data. The comments for those routines describe what particular 344 * operations are valid after calling them. 345 * 346 * The object number must be a valid, allocated object number. 347 */ 348 int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, 349 void *tag, dmu_buf_t **); 350 void dmu_buf_add_ref(dmu_buf_t *db, void* tag); 351 void dmu_buf_rele(dmu_buf_t *db, void *tag); 352 uint64_t dmu_buf_refcount(dmu_buf_t *db); 353 354 /* 355 * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a 356 * range of an object. A pointer to an array of dmu_buf_t*'s is 357 * returned (in *dbpp). 358 * 359 * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and 360 * frees the array. The hold on the array of buffers MUST be released 361 * with dmu_buf_rele_array. You can NOT release the hold on each buffer 362 * individually with dmu_buf_rele. 363 */ 364 int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, 365 uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 366 void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); 367 368 /* 369 * Returns NULL on success, or the existing user ptr if it's already 370 * been set. 371 * 372 * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). 373 * 374 * user_data_ptr_ptr should be NULL, or a pointer to a pointer which 375 * will be set to db->db_data when you are allowed to access it. Note 376 * that db->db_data (the pointer) can change when you do dmu_buf_read(), 377 * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). 378 * *user_data_ptr_ptr will be set to the new value when it changes. 379 * 380 * If non-NULL, pageout func will be called when this buffer is being 381 * excised from the cache, so that you can clean up the data structure 382 * pointed to by user_ptr. 383 * 384 * dmu_evict_user() will call the pageout func for all buffers in a 385 * objset with a given pageout func. 386 */ 387 void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, 388 dmu_buf_evict_func_t *pageout_func); 389 /* 390 * set_user_ie is the same as set_user, but request immediate eviction 391 * when hold count goes to zero. 392 */ 393 void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, 394 void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); 395 void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, 396 void *user_ptr, void *user_data_ptr_ptr, 397 dmu_buf_evict_func_t *pageout_func); 398 void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); 399 400 /* 401 * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. 402 */ 403 void *dmu_buf_get_user(dmu_buf_t *db); 404 405 /* 406 * Indicate that you are going to modify the buffer's data (db_data). 407 * 408 * The transaction (tx) must be assigned to a txg (ie. you've called 409 * dmu_tx_assign()). The buffer's object must be held in the tx 410 * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). 411 */ 412 void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); 413 414 /* 415 * Tells if the given dbuf is freeable. 416 */ 417 boolean_t dmu_buf_freeable(dmu_buf_t *); 418 419 /* 420 * You must create a transaction, then hold the objects which you will 421 * (or might) modify as part of this transaction. Then you must assign 422 * the transaction to a transaction group. Once the transaction has 423 * been assigned, you can modify buffers which belong to held objects as 424 * part of this transaction. You can't modify buffers before the 425 * transaction has been assigned; you can't modify buffers which don't 426 * belong to objects which this transaction holds; you can't hold 427 * objects once the transaction has been assigned. You may hold an 428 * object which you are going to free (with dmu_object_free()), but you 429 * don't have to. 430 * 431 * You can abort the transaction before it has been assigned. 432 * 433 * Note that you may hold buffers (with dmu_buf_hold) at any time, 434 * regardless of transaction state. 435 */ 436 437 #define DMU_NEW_OBJECT (-1ULL) 438 #define DMU_OBJECT_END (-1ULL) 439 440 dmu_tx_t *dmu_tx_create(objset_t *os); 441 void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); 442 void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, 443 uint64_t len); 444 void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); 445 void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); 446 void dmu_tx_abort(dmu_tx_t *tx); 447 int dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how); 448 void dmu_tx_wait(dmu_tx_t *tx); 449 void dmu_tx_commit(dmu_tx_t *tx); 450 451 /* 452 * To register a commit callback, dmu_tx_callback_register() must be called. 453 * 454 * dcb_data is a pointer to caller private data that is passed on as a 455 * callback parameter. The caller is responsible for properly allocating and 456 * freeing it. 457 * 458 * When registering a callback, the transaction must be already created, but 459 * it cannot be committed or aborted. It can be assigned to a txg or not. 460 * 461 * The callback will be called after the transaction has been safely written 462 * to stable storage and will also be called if the dmu_tx is aborted. 463 * If there is any error which prevents the transaction from being committed to 464 * disk, the callback will be called with a value of error != 0. 465 */ 466 typedef void dmu_tx_callback_func_t(void *dcb_data, int error); 467 468 void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, 469 void *dcb_data); 470 471 /* 472 * Free up the data blocks for a defined range of a file. If size is 473 * zero, the range from offset to end-of-file is freed. 474 */ 475 int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, 476 uint64_t size, dmu_tx_t *tx); 477 int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, 478 uint64_t size); 479 int dmu_free_object(objset_t *os, uint64_t object); 480 481 /* 482 * Convenience functions. 483 * 484 * Canfail routines will return 0 on success, or an errno if there is a 485 * nonrecoverable I/O error. 486 */ 487 #define DMU_READ_PREFETCH 0 /* prefetch */ 488 #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ 489 int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 490 void *buf, uint32_t flags); 491 void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 492 const void *buf, dmu_tx_t *tx); 493 void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 494 dmu_tx_t *tx); 495 int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); 496 int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, 497 dmu_tx_t *tx); 498 int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, 499 uint64_t size, struct page *pp, dmu_tx_t *tx); 500 struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); 501 void dmu_return_arcbuf(struct arc_buf *buf); 502 void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, 503 dmu_tx_t *tx); 504 int dmu_xuio_init(struct xuio *uio, int niov); 505 void dmu_xuio_fini(struct xuio *uio); 506 int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, 507 size_t n); 508 int dmu_xuio_cnt(struct xuio *uio); 509 struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); 510 void dmu_xuio_clear(struct xuio *uio, int i); 511 void xuio_stat_wbuf_copied(); 512 void xuio_stat_wbuf_nocopy(); 513 514 extern int zfs_prefetch_disable; 515 516 /* 517 * Asynchronously try to read in the data. 518 */ 519 void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, 520 uint64_t len); 521 522 typedef struct dmu_object_info { 523 /* All sizes are in bytes unless otherwise indicated. */ 524 uint32_t doi_data_block_size; 525 uint32_t doi_metadata_block_size; 526 dmu_object_type_t doi_type; 527 dmu_object_type_t doi_bonus_type; 528 uint64_t doi_bonus_size; 529 uint8_t doi_indirection; /* 2 = dnode->indirect->data */ 530 uint8_t doi_checksum; 531 uint8_t doi_compress; 532 uint8_t doi_pad[5]; 533 uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ 534 uint64_t doi_max_offset; 535 uint64_t doi_fill_count; /* number of non-empty blocks */ 536 } dmu_object_info_t; 537 538 typedef void arc_byteswap_func_t(void *buf, size_t size); 539 540 typedef struct dmu_object_type_info { 541 arc_byteswap_func_t *ot_byteswap; 542 boolean_t ot_metadata; 543 char *ot_name; 544 } dmu_object_type_info_t; 545 546 extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; 547 548 /* 549 * Get information on a DMU object. 550 * 551 * Return 0 on success or ENOENT if object is not allocated. 552 * 553 * If doi is NULL, just indicates whether the object exists. 554 */ 555 int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); 556 void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); 557 void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); 558 void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, 559 u_longlong_t *nblk512); 560 561 typedef struct dmu_objset_stats { 562 uint64_t dds_num_clones; /* number of clones of this */ 563 uint64_t dds_creation_txg; 564 uint64_t dds_guid; 565 dmu_objset_type_t dds_type; 566 uint8_t dds_is_snapshot; 567 uint8_t dds_inconsistent; 568 char dds_origin[MAXNAMELEN]; 569 } dmu_objset_stats_t; 570 571 /* 572 * Get stats on a dataset. 573 */ 574 void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); 575 576 /* 577 * Add entries to the nvlist for all the objset's properties. See 578 * zfs_prop_table[] and zfs(1m) for details on the properties. 579 */ 580 void dmu_objset_stats(objset_t *os, struct nvlist *nv); 581 582 /* 583 * Get the space usage statistics for statvfs(). 584 * 585 * refdbytes is the amount of space "referenced" by this objset. 586 * availbytes is the amount of space available to this objset, taking 587 * into account quotas & reservations, assuming that no other objsets 588 * use the space first. These values correspond to the 'referenced' and 589 * 'available' properties, described in the zfs(1m) manpage. 590 * 591 * usedobjs and availobjs are the number of objects currently allocated, 592 * and available. 593 */ 594 void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, 595 uint64_t *usedobjsp, uint64_t *availobjsp); 596 597 /* 598 * The fsid_guid is a 56-bit ID that can change to avoid collisions. 599 * (Contrast with the ds_guid which is a 64-bit ID that will never 600 * change, so there is a small probability that it will collide.) 601 */ 602 uint64_t dmu_objset_fsid_guid(objset_t *os); 603 604 /* 605 * Get the [cm]time for an objset's snapshot dir 606 */ 607 timestruc_t dmu_objset_snap_cmtime(objset_t *os); 608 609 int dmu_objset_is_snapshot(objset_t *os); 610 611 extern struct spa *dmu_objset_spa(objset_t *os); 612 extern struct zilog *dmu_objset_zil(objset_t *os); 613 extern struct dsl_pool *dmu_objset_pool(objset_t *os); 614 extern struct dsl_dataset *dmu_objset_ds(objset_t *os); 615 extern void dmu_objset_name(objset_t *os, char *buf); 616 extern dmu_objset_type_t dmu_objset_type(objset_t *os); 617 extern uint64_t dmu_objset_id(objset_t *os); 618 extern uint64_t dmu_objset_logbias(objset_t *os); 619 extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, 620 uint64_t *id, uint64_t *offp, boolean_t *case_conflict); 621 extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, 622 int maxlen, boolean_t *conflict); 623 extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, 624 uint64_t *idp, uint64_t *offp); 625 626 typedef int objset_used_cb_t(dmu_object_type_t bonustype, 627 void *bonus, uint64_t *userp, uint64_t *groupp); 628 extern void dmu_objset_register_type(dmu_objset_type_t ost, 629 objset_used_cb_t *cb); 630 extern void dmu_objset_set_user(objset_t *os, void *user_ptr); 631 extern void *dmu_objset_get_user(objset_t *os); 632 633 /* 634 * Return the txg number for the given assigned transaction. 635 */ 636 uint64_t dmu_tx_get_txg(dmu_tx_t *tx); 637 638 /* 639 * Synchronous write. 640 * If a parent zio is provided this function initiates a write on the 641 * provided buffer as a child of the parent zio. 642 * In the absence of a parent zio, the write is completed synchronously. 643 * At write completion, blk is filled with the bp of the written block. 644 * Note that while the data covered by this function will be on stable 645 * storage when the write completes this new data does not become a 646 * permanent part of the file until the associated transaction commits. 647 */ 648 649 /* 650 * {zfs,zvol,ztest}_get_done() args 651 */ 652 typedef struct zgd { 653 struct zilog *zgd_zilog; 654 struct blkptr *zgd_bp; 655 dmu_buf_t *zgd_db; 656 struct rl *zgd_rl; 657 void *zgd_private; 658 } zgd_t; 659 660 typedef void dmu_sync_cb_t(zgd_t *arg, int error); 661 int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); 662 663 /* 664 * Find the next hole or data block in file starting at *off 665 * Return found offset in *off. Return ESRCH for end of file. 666 */ 667 int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, 668 uint64_t *off); 669 670 /* 671 * Initial setup and final teardown. 672 */ 673 extern void dmu_init(void); 674 extern void dmu_fini(void); 675 676 typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, 677 uint64_t object, uint64_t offset, int len); 678 void dmu_traverse_objset(objset_t *os, uint64_t txg_start, 679 dmu_traverse_cb_t cb, void *arg); 680 681 int dmu_sendbackup(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, 682 struct vnode *vp, offset_t *off); 683 684 typedef struct dmu_recv_cookie { 685 /* 686 * This structure is opaque! 687 * 688 * If logical and real are different, we are recving the stream 689 * into the "real" temporary clone, and then switching it with 690 * the "logical" target. 691 */ 692 struct dsl_dataset *drc_logical_ds; 693 struct dsl_dataset *drc_real_ds; 694 struct drr_begin *drc_drrb; 695 char *drc_tosnap; 696 char *drc_top_ds; 697 boolean_t drc_newfs; 698 boolean_t drc_force; 699 } dmu_recv_cookie_t; 700 701 int dmu_recv_begin(char *tofs, char *tosnap, char *topds, struct drr_begin *, 702 boolean_t force, objset_t *origin, dmu_recv_cookie_t *); 703 int dmu_recv_stream(dmu_recv_cookie_t *drc, struct vnode *vp, offset_t *voffp); 704 int dmu_recv_end(dmu_recv_cookie_t *drc); 705 706 /* CRC64 table */ 707 #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ 708 extern uint64_t zfs_crc64_table[256]; 709 710 #ifdef __cplusplus 711 } 712 #endif 713 714 #endif /* _SYS_DMU_H */ 715