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 2009 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 page; 49 struct vnode; 50 struct spa; 51 struct zilog; 52 struct zio; 53 struct blkptr; 54 struct zap_cursor; 55 struct dsl_dataset; 56 struct dsl_pool; 57 struct dnode; 58 struct drr_begin; 59 struct drr_end; 60 struct zbookmark; 61 struct spa; 62 struct nvlist; 63 struct arc_buf; 64 struct zio_prop; 65 66 typedef struct objset objset_t; 67 typedef struct dmu_tx dmu_tx_t; 68 typedef struct dsl_dir dsl_dir_t; 69 70 typedef enum dmu_object_type { 71 DMU_OT_NONE, 72 /* general: */ 73 DMU_OT_OBJECT_DIRECTORY, /* ZAP */ 74 DMU_OT_OBJECT_ARRAY, /* UINT64 */ 75 DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ 76 DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ 77 DMU_OT_BPLIST, /* UINT64 */ 78 DMU_OT_BPLIST_HDR, /* UINT64 */ 79 /* spa: */ 80 DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ 81 DMU_OT_SPACE_MAP, /* UINT64 */ 82 /* zil: */ 83 DMU_OT_INTENT_LOG, /* UINT64 */ 84 /* dmu: */ 85 DMU_OT_DNODE, /* DNODE */ 86 DMU_OT_OBJSET, /* OBJSET */ 87 /* dsl: */ 88 DMU_OT_DSL_DIR, /* UINT64 */ 89 DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ 90 DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ 91 DMU_OT_DSL_PROPS, /* ZAP */ 92 DMU_OT_DSL_DATASET, /* UINT64 */ 93 /* zpl: */ 94 DMU_OT_ZNODE, /* ZNODE */ 95 DMU_OT_OLDACL, /* Old ACL */ 96 DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ 97 DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ 98 DMU_OT_MASTER_NODE, /* ZAP */ 99 DMU_OT_UNLINKED_SET, /* ZAP */ 100 /* zvol: */ 101 DMU_OT_ZVOL, /* UINT8 */ 102 DMU_OT_ZVOL_PROP, /* ZAP */ 103 /* other; for testing only! */ 104 DMU_OT_PLAIN_OTHER, /* UINT8 */ 105 DMU_OT_UINT64_OTHER, /* UINT64 */ 106 DMU_OT_ZAP_OTHER, /* ZAP */ 107 /* new object types: */ 108 DMU_OT_ERROR_LOG, /* ZAP */ 109 DMU_OT_SPA_HISTORY, /* UINT8 */ 110 DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ 111 DMU_OT_POOL_PROPS, /* ZAP */ 112 DMU_OT_DSL_PERMS, /* ZAP */ 113 DMU_OT_ACL, /* ACL */ 114 DMU_OT_SYSACL, /* SYSACL */ 115 DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ 116 DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ 117 DMU_OT_NEXT_CLONES, /* ZAP */ 118 DMU_OT_SCRUB_QUEUE, /* ZAP */ 119 DMU_OT_USERGROUP_USED, /* ZAP */ 120 DMU_OT_USERGROUP_QUOTA, /* ZAP */ 121 DMU_OT_USERREFS, /* ZAP */ 122 DMU_OT_DDT_ZAP, /* ZAP */ 123 DMU_OT_DDT_STATS, /* ZAP */ 124 DMU_OT_NUMTYPES 125 } dmu_object_type_t; 126 127 typedef enum dmu_objset_type { 128 DMU_OST_NONE, 129 DMU_OST_META, 130 DMU_OST_ZFS, 131 DMU_OST_ZVOL, 132 DMU_OST_OTHER, /* For testing only! */ 133 DMU_OST_ANY, /* Be careful! */ 134 DMU_OST_NUMTYPES 135 } dmu_objset_type_t; 136 137 void byteswap_uint64_array(void *buf, size_t size); 138 void byteswap_uint32_array(void *buf, size_t size); 139 void byteswap_uint16_array(void *buf, size_t size); 140 void byteswap_uint8_array(void *buf, size_t size); 141 void zap_byteswap(void *buf, size_t size); 142 void zfs_oldacl_byteswap(void *buf, size_t size); 143 void zfs_acl_byteswap(void *buf, size_t size); 144 void zfs_znode_byteswap(void *buf, size_t size); 145 146 #define DS_FIND_SNAPSHOTS (1<<0) 147 #define DS_FIND_CHILDREN (1<<1) 148 149 /* 150 * The maximum number of bytes that can be accessed as part of one 151 * operation, including metadata. 152 */ 153 #define DMU_MAX_ACCESS (10<<20) /* 10MB */ 154 #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ 155 156 #define DMU_USERUSED_OBJECT (-1ULL) 157 #define DMU_GROUPUSED_OBJECT (-2ULL) 158 #define DMU_DEADLIST_OBJECT (-3ULL) 159 160 /* 161 * Public routines to create, destroy, open, and close objsets. 162 */ 163 int dmu_objset_hold(const char *name, void *tag, objset_t **osp); 164 int dmu_objset_own(const char *name, dmu_objset_type_t type, 165 boolean_t readonly, void *tag, objset_t **osp); 166 void dmu_objset_rele(objset_t *os, void *tag); 167 void dmu_objset_disown(objset_t *os, void *tag); 168 int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); 169 170 int dmu_objset_evict_dbufs(objset_t *os); 171 int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, 172 void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); 173 int dmu_objset_clone(const char *name, struct dsl_dataset *clone_origin, 174 uint64_t flags); 175 int dmu_objset_destroy(const char *name, boolean_t defer); 176 int dmu_snapshots_destroy(char *fsname, char *snapname, boolean_t defer); 177 int dmu_objset_snapshot(char *fsname, char *snapname, struct nvlist *props, 178 boolean_t recursive); 179 int dmu_objset_rename(const char *name, const char *newname, 180 boolean_t recursive); 181 int dmu_objset_find(char *name, int func(const char *, void *), void *arg, 182 int flags); 183 void dmu_objset_byteswap(void *buf, size_t size); 184 185 typedef struct dmu_buf { 186 uint64_t db_object; /* object that this buffer is part of */ 187 uint64_t db_offset; /* byte offset in this object */ 188 uint64_t db_size; /* size of buffer in bytes */ 189 void *db_data; /* data in buffer */ 190 } dmu_buf_t; 191 192 typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); 193 194 /* 195 * The names of zap entries in the DIRECTORY_OBJECT of the MOS. 196 */ 197 #define DMU_POOL_DIRECTORY_OBJECT 1 198 #define DMU_POOL_CONFIG "config" 199 #define DMU_POOL_ROOT_DATASET "root_dataset" 200 #define DMU_POOL_SYNC_BPLIST "sync_bplist" 201 #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" 202 #define DMU_POOL_ERRLOG_LAST "errlog_last" 203 #define DMU_POOL_SPARES "spares" 204 #define DMU_POOL_DEFLATE "deflate" 205 #define DMU_POOL_HISTORY "history" 206 #define DMU_POOL_PROPS "pool_props" 207 #define DMU_POOL_L2CACHE "l2cache" 208 #define DMU_POOL_TMP_USERREFS "tmp_userrefs" 209 #define DMU_POOL_DDT "DDT-%s-%s-%s" 210 #define DMU_POOL_DDT_STATS "DDT-statistics" 211 212 /* 4x8 zbookmark_t */ 213 #define DMU_POOL_SCRUB_BOOKMARK "scrub_bookmark" 214 /* 4x8 ddt_bookmark_t */ 215 #define DMU_POOL_SCRUB_DDT_BOOKMARK "scrub_ddt_bookmark" 216 /* 1x8 max_class */ 217 #define DMU_POOL_SCRUB_DDT_CLASS_MAX "scrub_ddt_class_max" 218 /* 1x8 zap obj DMU_OT_SCRUB_QUEUE */ 219 #define DMU_POOL_SCRUB_QUEUE "scrub_queue" 220 /* 1x8 txg */ 221 #define DMU_POOL_SCRUB_MIN_TXG "scrub_min_txg" 222 /* 1x8 txg */ 223 #define DMU_POOL_SCRUB_MAX_TXG "scrub_max_txg" 224 /* 1x4 enum scrub_func */ 225 #define DMU_POOL_SCRUB_FUNC "scrub_func" 226 /* 1x8 count */ 227 #define DMU_POOL_SCRUB_ERRORS "scrub_errors" 228 229 /* 230 * Allocate an object from this objset. The range of object numbers 231 * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. 232 * 233 * The transaction must be assigned to a txg. The newly allocated 234 * object will be "held" in the transaction (ie. you can modify the 235 * newly allocated object in this transaction). 236 * 237 * dmu_object_alloc() chooses an object and returns it in *objectp. 238 * 239 * dmu_object_claim() allocates a specific object number. If that 240 * number is already allocated, it fails and returns EEXIST. 241 * 242 * Return 0 on success, or ENOSPC or EEXIST as specified above. 243 */ 244 uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, 245 int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 246 int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, 247 int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 248 int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, 249 int blocksize, dmu_object_type_t bonustype, int bonuslen); 250 251 /* 252 * Free an object from this objset. 253 * 254 * The object's data will be freed as well (ie. you don't need to call 255 * dmu_free(object, 0, -1, tx)). 256 * 257 * The object need not be held in the transaction. 258 * 259 * If there are any holds on this object's buffers (via dmu_buf_hold()), 260 * or tx holds on the object (via dmu_tx_hold_object()), you can not 261 * free it; it fails and returns EBUSY. 262 * 263 * If the object is not allocated, it fails and returns ENOENT. 264 * 265 * Return 0 on success, or EBUSY or ENOENT as specified above. 266 */ 267 int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); 268 269 /* 270 * Find the next allocated or free object. 271 * 272 * The objectp parameter is in-out. It will be updated to be the next 273 * object which is allocated. Ignore objects which have not been 274 * modified since txg. 275 * 276 * XXX Can only be called on a objset with no dirty data. 277 * 278 * Returns 0 on success, or ENOENT if there are no more objects. 279 */ 280 int dmu_object_next(objset_t *os, uint64_t *objectp, 281 boolean_t hole, uint64_t txg); 282 283 /* 284 * Set the data blocksize for an object. 285 * 286 * The object cannot have any blocks allcated beyond the first. If 287 * the first block is allocated already, the new size must be greater 288 * than the current block size. If these conditions are not met, 289 * ENOTSUP will be returned. 290 * 291 * Returns 0 on success, or EBUSY if there are any holds on the object 292 * contents, or ENOTSUP as described above. 293 */ 294 int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, 295 int ibs, dmu_tx_t *tx); 296 297 /* 298 * Set the checksum property on a dnode. The new checksum algorithm will 299 * apply to all newly written blocks; existing blocks will not be affected. 300 */ 301 void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, 302 dmu_tx_t *tx); 303 304 /* 305 * Set the compress property on a dnode. The new compression algorithm will 306 * apply to all newly written blocks; existing blocks will not be affected. 307 */ 308 void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, 309 dmu_tx_t *tx); 310 311 /* 312 * Decide how to write a block: checksum, compression, number of copies, etc. 313 */ 314 #define WP_NOFILL 0x1 315 #define WP_DMU_SYNC 0x2 316 317 void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, 318 struct zio_prop *zp); 319 /* 320 * The bonus data is accessed more or less like a regular buffer. 321 * You must dmu_bonus_hold() to get the buffer, which will give you a 322 * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus 323 * data. As with any normal buffer, you must call dmu_buf_read() to 324 * read db_data, dmu_buf_will_dirty() before modifying it, and the 325 * object must be held in an assigned transaction before calling 326 * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus 327 * buffer as well. You must release your hold with dmu_buf_rele(). 328 */ 329 int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 330 int dmu_bonus_max(void); 331 int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); 332 333 /* 334 * Obtain the DMU buffer from the specified object which contains the 335 * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so 336 * that it will remain in memory. You must release the hold with 337 * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your 338 * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. 339 * 340 * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill 341 * on the returned buffer before reading or writing the buffer's 342 * db_data. The comments for those routines describe what particular 343 * operations are valid after calling them. 344 * 345 * The object number must be a valid, allocated object number. 346 */ 347 int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, 348 void *tag, dmu_buf_t **); 349 void dmu_buf_add_ref(dmu_buf_t *db, void* tag); 350 void dmu_buf_rele(dmu_buf_t *db, void *tag); 351 uint64_t dmu_buf_refcount(dmu_buf_t *db); 352 353 /* 354 * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a 355 * range of an object. A pointer to an array of dmu_buf_t*'s is 356 * returned (in *dbpp). 357 * 358 * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and 359 * frees the array. The hold on the array of buffers MUST be released 360 * with dmu_buf_rele_array. You can NOT release the hold on each buffer 361 * individually with dmu_buf_rele. 362 */ 363 int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, 364 uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 365 void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); 366 367 /* 368 * Returns NULL on success, or the existing user ptr if it's already 369 * been set. 370 * 371 * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). 372 * 373 * user_data_ptr_ptr should be NULL, or a pointer to a pointer which 374 * will be set to db->db_data when you are allowed to access it. Note 375 * that db->db_data (the pointer) can change when you do dmu_buf_read(), 376 * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). 377 * *user_data_ptr_ptr will be set to the new value when it changes. 378 * 379 * If non-NULL, pageout func will be called when this buffer is being 380 * excised from the cache, so that you can clean up the data structure 381 * pointed to by user_ptr. 382 * 383 * dmu_evict_user() will call the pageout func for all buffers in a 384 * objset with a given pageout func. 385 */ 386 void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, 387 dmu_buf_evict_func_t *pageout_func); 388 /* 389 * set_user_ie is the same as set_user, but request immediate eviction 390 * when hold count goes to zero. 391 */ 392 void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, 393 void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); 394 void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, 395 void *user_ptr, void *user_data_ptr_ptr, 396 dmu_buf_evict_func_t *pageout_func); 397 void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); 398 399 /* 400 * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. 401 */ 402 void *dmu_buf_get_user(dmu_buf_t *db); 403 404 /* 405 * Indicate that you are going to modify the buffer's data (db_data). 406 * 407 * The transaction (tx) must be assigned to a txg (ie. you've called 408 * dmu_tx_assign()). The buffer's object must be held in the tx 409 * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). 410 */ 411 void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); 412 413 /* 414 * Tells if the given dbuf is freeable. 415 */ 416 boolean_t dmu_buf_freeable(dmu_buf_t *); 417 418 /* 419 * You must create a transaction, then hold the objects which you will 420 * (or might) modify as part of this transaction. Then you must assign 421 * the transaction to a transaction group. Once the transaction has 422 * been assigned, you can modify buffers which belong to held objects as 423 * part of this transaction. You can't modify buffers before the 424 * transaction has been assigned; you can't modify buffers which don't 425 * belong to objects which this transaction holds; you can't hold 426 * objects once the transaction has been assigned. You may hold an 427 * object which you are going to free (with dmu_object_free()), but you 428 * don't have to. 429 * 430 * You can abort the transaction before it has been assigned. 431 * 432 * Note that you may hold buffers (with dmu_buf_hold) at any time, 433 * regardless of transaction state. 434 */ 435 436 #define DMU_NEW_OBJECT (-1ULL) 437 #define DMU_OBJECT_END (-1ULL) 438 439 dmu_tx_t *dmu_tx_create(objset_t *os); 440 void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); 441 void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, 442 uint64_t len); 443 void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); 444 void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); 445 void dmu_tx_abort(dmu_tx_t *tx); 446 int dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how); 447 void dmu_tx_wait(dmu_tx_t *tx); 448 void dmu_tx_commit(dmu_tx_t *tx); 449 450 /* 451 * To register a commit callback, dmu_tx_callback_register() must be called. 452 * 453 * dcb_data is a pointer to caller private data that is passed on as a 454 * callback parameter. The caller is responsible for properly allocating and 455 * freeing it. 456 * 457 * When registering a callback, the transaction must be already created, but 458 * it cannot be committed or aborted. It can be assigned to a txg or not. 459 * 460 * The callback will be called after the transaction has been safely written 461 * to stable storage and will also be called if the dmu_tx is aborted. 462 * If there is any error which prevents the transaction from being committed to 463 * disk, the callback will be called with a value of error != 0. 464 */ 465 typedef void dmu_tx_callback_func_t(void *dcb_data, int error); 466 467 void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, 468 void *dcb_data); 469 470 /* 471 * Free up the data blocks for a defined range of a file. If size is 472 * zero, the range from offset to end-of-file is freed. 473 */ 474 int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, 475 uint64_t size, dmu_tx_t *tx); 476 int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, 477 uint64_t size); 478 int dmu_free_object(objset_t *os, uint64_t object); 479 480 /* 481 * Convenience functions. 482 * 483 * Canfail routines will return 0 on success, or an errno if there is a 484 * nonrecoverable I/O error. 485 */ 486 #define DMU_READ_PREFETCH 0 /* prefetch */ 487 #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ 488 int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 489 void *buf, uint32_t flags); 490 void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 491 const void *buf, dmu_tx_t *tx); 492 void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 493 dmu_tx_t *tx); 494 int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); 495 int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, 496 dmu_tx_t *tx); 497 int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, 498 uint64_t size, struct page *pp, dmu_tx_t *tx); 499 struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); 500 void dmu_return_arcbuf(struct arc_buf *buf); 501 void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, 502 dmu_tx_t *tx); 503 504 extern int zfs_prefetch_disable; 505 506 /* 507 * Asynchronously try to read in the data. 508 */ 509 void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, 510 uint64_t len); 511 512 typedef struct dmu_object_info { 513 /* All sizes are in bytes unless otherwise indicated. */ 514 uint32_t doi_data_block_size; 515 uint32_t doi_metadata_block_size; 516 dmu_object_type_t doi_type; 517 dmu_object_type_t doi_bonus_type; 518 uint64_t doi_bonus_size; 519 uint8_t doi_indirection; /* 2 = dnode->indirect->data */ 520 uint8_t doi_checksum; 521 uint8_t doi_compress; 522 uint8_t doi_pad[5]; 523 uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ 524 uint64_t doi_max_offset; 525 uint64_t doi_fill_count; /* number of non-empty blocks */ 526 } dmu_object_info_t; 527 528 typedef void arc_byteswap_func_t(void *buf, size_t size); 529 530 typedef struct dmu_object_type_info { 531 arc_byteswap_func_t *ot_byteswap; 532 boolean_t ot_metadata; 533 char *ot_name; 534 } dmu_object_type_info_t; 535 536 extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; 537 538 /* 539 * Get information on a DMU object. 540 * 541 * Return 0 on success or ENOENT if object is not allocated. 542 * 543 * If doi is NULL, just indicates whether the object exists. 544 */ 545 int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); 546 void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); 547 void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); 548 void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, 549 u_longlong_t *nblk512); 550 551 typedef struct dmu_objset_stats { 552 uint64_t dds_num_clones; /* number of clones of this */ 553 uint64_t dds_creation_txg; 554 uint64_t dds_guid; 555 dmu_objset_type_t dds_type; 556 uint8_t dds_is_snapshot; 557 uint8_t dds_inconsistent; 558 char dds_origin[MAXNAMELEN]; 559 } dmu_objset_stats_t; 560 561 /* 562 * Get stats on a dataset. 563 */ 564 void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); 565 566 /* 567 * Add entries to the nvlist for all the objset's properties. See 568 * zfs_prop_table[] and zfs(1m) for details on the properties. 569 */ 570 void dmu_objset_stats(objset_t *os, struct nvlist *nv); 571 572 /* 573 * Get the space usage statistics for statvfs(). 574 * 575 * refdbytes is the amount of space "referenced" by this objset. 576 * availbytes is the amount of space available to this objset, taking 577 * into account quotas & reservations, assuming that no other objsets 578 * use the space first. These values correspond to the 'referenced' and 579 * 'available' properties, described in the zfs(1m) manpage. 580 * 581 * usedobjs and availobjs are the number of objects currently allocated, 582 * and available. 583 */ 584 void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, 585 uint64_t *usedobjsp, uint64_t *availobjsp); 586 587 /* 588 * The fsid_guid is a 56-bit ID that can change to avoid collisions. 589 * (Contrast with the ds_guid which is a 64-bit ID that will never 590 * change, so there is a small probability that it will collide.) 591 */ 592 uint64_t dmu_objset_fsid_guid(objset_t *os); 593 594 /* 595 * Get the [cm]time for an objset's snapshot dir 596 */ 597 timestruc_t dmu_objset_snap_cmtime(objset_t *os); 598 599 int dmu_objset_is_snapshot(objset_t *os); 600 601 extern struct spa *dmu_objset_spa(objset_t *os); 602 extern struct zilog *dmu_objset_zil(objset_t *os); 603 extern struct dsl_pool *dmu_objset_pool(objset_t *os); 604 extern struct dsl_dataset *dmu_objset_ds(objset_t *os); 605 extern void dmu_objset_name(objset_t *os, char *buf); 606 extern dmu_objset_type_t dmu_objset_type(objset_t *os); 607 extern uint64_t dmu_objset_id(objset_t *os); 608 extern uint64_t dmu_objset_logbias(objset_t *os); 609 extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, 610 uint64_t *id, uint64_t *offp, boolean_t *case_conflict); 611 extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, 612 int maxlen, boolean_t *conflict); 613 extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, 614 uint64_t *idp, uint64_t *offp); 615 616 typedef int objset_used_cb_t(dmu_object_type_t bonustype, 617 void *bonus, uint64_t *userp, uint64_t *groupp); 618 extern void dmu_objset_register_type(dmu_objset_type_t ost, 619 objset_used_cb_t *cb); 620 extern void dmu_objset_set_user(objset_t *os, void *user_ptr); 621 extern void *dmu_objset_get_user(objset_t *os); 622 623 /* 624 * Return the txg number for the given assigned transaction. 625 */ 626 uint64_t dmu_tx_get_txg(dmu_tx_t *tx); 627 628 /* 629 * Synchronous write. 630 * If a parent zio is provided this function initiates a write on the 631 * provided buffer as a child of the parent zio. 632 * In the absence of a parent zio, the write is completed synchronously. 633 * At write completion, blk is filled with the bp of the written block. 634 * Note that while the data covered by this function will be on stable 635 * storage when the write completes this new data does not become a 636 * permanent part of the file until the associated transaction commits. 637 */ 638 639 /* 640 * {zfs,zvol,ztest}_get_done() args 641 */ 642 typedef struct zgd { 643 struct zilog *zgd_zilog; 644 struct blkptr *zgd_bp; 645 dmu_buf_t *zgd_db; 646 struct rl *zgd_rl; 647 void *zgd_private; 648 } zgd_t; 649 650 typedef void dmu_sync_cb_t(zgd_t *arg, int error); 651 int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); 652 653 /* 654 * Find the next hole or data block in file starting at *off 655 * Return found offset in *off. Return ESRCH for end of file. 656 */ 657 int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, 658 uint64_t *off); 659 660 /* 661 * Initial setup and final teardown. 662 */ 663 extern void dmu_init(void); 664 extern void dmu_fini(void); 665 666 typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, 667 uint64_t object, uint64_t offset, int len); 668 void dmu_traverse_objset(objset_t *os, uint64_t txg_start, 669 dmu_traverse_cb_t cb, void *arg); 670 671 int dmu_sendbackup(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, 672 struct vnode *vp, offset_t *off); 673 674 typedef struct dmu_recv_cookie { 675 /* 676 * This structure is opaque! 677 * 678 * If logical and real are different, we are recving the stream 679 * into the "real" temporary clone, and then switching it with 680 * the "logical" target. 681 */ 682 struct dsl_dataset *drc_logical_ds; 683 struct dsl_dataset *drc_real_ds; 684 struct drr_begin *drc_drrb; 685 char *drc_tosnap; 686 char *drc_top_ds; 687 boolean_t drc_newfs; 688 boolean_t drc_force; 689 } dmu_recv_cookie_t; 690 691 int dmu_recv_begin(char *tofs, char *tosnap, char *topds, struct drr_begin *, 692 boolean_t force, objset_t *origin, dmu_recv_cookie_t *); 693 int dmu_recv_stream(dmu_recv_cookie_t *drc, struct vnode *vp, offset_t *voffp); 694 int dmu_recv_end(dmu_recv_cookie_t *drc); 695 696 /* CRC64 table */ 697 #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ 698 extern uint64_t zfs_crc64_table[256]; 699 700 #ifdef __cplusplus 701 } 702 #endif 703 704 #endif /* _SYS_DMU_H */ 705