1fa9e4066Sahrens /* 2fa9e4066Sahrens * CDDL HEADER START 3fa9e4066Sahrens * 4fa9e4066Sahrens * The contents of this file are subject to the terms of the 5ea8dc4b6Seschrock * Common Development and Distribution License (the "License"). 6ea8dc4b6Seschrock * You may not use this file except in compliance with the License. 7fa9e4066Sahrens * 8fa9e4066Sahrens * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9fa9e4066Sahrens * or http://www.opensolaris.org/os/licensing. 10fa9e4066Sahrens * See the License for the specific language governing permissions 11fa9e4066Sahrens * and limitations under the License. 12fa9e4066Sahrens * 13fa9e4066Sahrens * When distributing Covered Code, include this CDDL HEADER in each 14fa9e4066Sahrens * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15fa9e4066Sahrens * If applicable, add the following below this CDDL HEADER, with the 16fa9e4066Sahrens * fields enclosed by brackets "[]" replaced with your own identifying 17fa9e4066Sahrens * information: Portions Copyright [yyyy] [name of copyright owner] 18fa9e4066Sahrens * 19fa9e4066Sahrens * CDDL HEADER END 20fa9e4066Sahrens */ 21fa9e4066Sahrens /* 227f7322feSeschrock * Copyright 2006 Sun Microsystems, Inc. All rights reserved. 23fa9e4066Sahrens * Use is subject to license terms. 24fa9e4066Sahrens */ 25fa9e4066Sahrens 26fa9e4066Sahrens #ifndef _SYS_DMU_H 27fa9e4066Sahrens #define _SYS_DMU_H 28fa9e4066Sahrens 29fa9e4066Sahrens #pragma ident "%Z%%M% %I% %E% SMI" 30fa9e4066Sahrens 31fa9e4066Sahrens /* 32fa9e4066Sahrens * This file describes the interface that the DMU provides for its 33fa9e4066Sahrens * consumers. 34fa9e4066Sahrens * 35fa9e4066Sahrens * The DMU also interacts with the SPA. That interface is described in 36fa9e4066Sahrens * dmu_spa.h. 37fa9e4066Sahrens */ 38fa9e4066Sahrens 39fa9e4066Sahrens #include <sys/inttypes.h> 40fa9e4066Sahrens #include <sys/types.h> 41fa9e4066Sahrens #include <sys/param.h> 42fa9e4066Sahrens 43fa9e4066Sahrens #ifdef __cplusplus 44fa9e4066Sahrens extern "C" { 45fa9e4066Sahrens #endif 46fa9e4066Sahrens 47fa9e4066Sahrens struct uio; 48fa9e4066Sahrens struct vnode; 49fa9e4066Sahrens struct spa; 50fa9e4066Sahrens struct zilog; 51fa9e4066Sahrens struct zio; 52fa9e4066Sahrens struct blkptr; 53fa9e4066Sahrens struct zap_cursor; 54fa9e4066Sahrens struct dsl_dataset; 55fa9e4066Sahrens struct dsl_pool; 56fa9e4066Sahrens struct dnode; 57fa9e4066Sahrens struct drr_begin; 58fa9e4066Sahrens struct drr_end; 5944cd46caSbillm struct zbookmark; 6044cd46caSbillm struct spa; 61fa9e4066Sahrens 62fa9e4066Sahrens typedef struct objset objset_t; 63fa9e4066Sahrens typedef struct dmu_tx dmu_tx_t; 64fa9e4066Sahrens typedef struct dsl_dir dsl_dir_t; 65fa9e4066Sahrens 66fa9e4066Sahrens typedef enum dmu_object_type { 67fa9e4066Sahrens DMU_OT_NONE, 68fa9e4066Sahrens /* general: */ 69fa9e4066Sahrens DMU_OT_OBJECT_DIRECTORY, /* ZAP */ 70fa9e4066Sahrens DMU_OT_OBJECT_ARRAY, /* UINT64 */ 71fa9e4066Sahrens DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ 72fa9e4066Sahrens DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ 73fa9e4066Sahrens DMU_OT_BPLIST, /* UINT64 */ 74fa9e4066Sahrens DMU_OT_BPLIST_HDR, /* UINT64 */ 75fa9e4066Sahrens /* spa: */ 76fa9e4066Sahrens DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ 77fa9e4066Sahrens DMU_OT_SPACE_MAP, /* UINT64 */ 78fa9e4066Sahrens /* zil: */ 79fa9e4066Sahrens DMU_OT_INTENT_LOG, /* UINT64 */ 80fa9e4066Sahrens /* dmu: */ 81fa9e4066Sahrens DMU_OT_DNODE, /* DNODE */ 82fa9e4066Sahrens DMU_OT_OBJSET, /* OBJSET */ 83fa9e4066Sahrens /* dsl: */ 841649cd4bStabriz DMU_OT_DSL_DIR, /* UINT64 */ 8587e5029aSahrens DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ 8687e5029aSahrens DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ 87fa9e4066Sahrens DMU_OT_DSL_PROPS, /* ZAP */ 881649cd4bStabriz DMU_OT_DSL_DATASET, /* UINT64 */ 89fa9e4066Sahrens /* zpl: */ 90fa9e4066Sahrens DMU_OT_ZNODE, /* ZNODE */ 91fa9e4066Sahrens DMU_OT_ACL, /* ACL */ 92fa9e4066Sahrens DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ 93fa9e4066Sahrens DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ 94fa9e4066Sahrens DMU_OT_MASTER_NODE, /* ZAP */ 95fa9e4066Sahrens DMU_OT_DELETE_QUEUE, /* ZAP */ 96fa9e4066Sahrens /* zvol: */ 97fa9e4066Sahrens DMU_OT_ZVOL, /* UINT8 */ 98fa9e4066Sahrens DMU_OT_ZVOL_PROP, /* ZAP */ 99fa9e4066Sahrens /* other; for testing only! */ 100fa9e4066Sahrens DMU_OT_PLAIN_OTHER, /* UINT8 */ 101fa9e4066Sahrens DMU_OT_UINT64_OTHER, /* UINT64 */ 102fa9e4066Sahrens DMU_OT_ZAP_OTHER, /* ZAP */ 103ea8dc4b6Seschrock /* new object types: */ 104ea8dc4b6Seschrock DMU_OT_ERROR_LOG, /* ZAP */ 105fa9e4066Sahrens 106fa9e4066Sahrens DMU_OT_NUMTYPES 107fa9e4066Sahrens } dmu_object_type_t; 108fa9e4066Sahrens 109fa9e4066Sahrens typedef enum dmu_objset_type { 110fa9e4066Sahrens DMU_OST_NONE, 111fa9e4066Sahrens DMU_OST_META, 112fa9e4066Sahrens DMU_OST_ZFS, 113fa9e4066Sahrens DMU_OST_ZVOL, 114fa9e4066Sahrens DMU_OST_OTHER, /* For testing only! */ 115fa9e4066Sahrens DMU_OST_ANY, /* Be careful! */ 116fa9e4066Sahrens DMU_OST_NUMTYPES 117fa9e4066Sahrens } dmu_objset_type_t; 118fa9e4066Sahrens 119fa9e4066Sahrens void byteswap_uint64_array(void *buf, size_t size); 120fa9e4066Sahrens void byteswap_uint32_array(void *buf, size_t size); 121fa9e4066Sahrens void byteswap_uint16_array(void *buf, size_t size); 122fa9e4066Sahrens void byteswap_uint8_array(void *buf, size_t size); 123fa9e4066Sahrens void zap_byteswap(void *buf, size_t size); 124fa9e4066Sahrens void zfs_acl_byteswap(void *buf, size_t size); 125fa9e4066Sahrens void zfs_znode_byteswap(void *buf, size_t size); 126fa9e4066Sahrens 127fa9e4066Sahrens #define DS_MODE_NONE 0 /* invalid, to aid debugging */ 128fa9e4066Sahrens #define DS_MODE_STANDARD 1 /* normal access, no special needs */ 129fa9e4066Sahrens #define DS_MODE_PRIMARY 2 /* the "main" access, e.g. a mount */ 130fa9e4066Sahrens #define DS_MODE_EXCLUSIVE 3 /* exclusive access, e.g. to destroy */ 131fa9e4066Sahrens #define DS_MODE_LEVELS 4 132fa9e4066Sahrens #define DS_MODE_LEVEL(x) ((x) & (DS_MODE_LEVELS - 1)) 133fa9e4066Sahrens #define DS_MODE_READONLY 0x8 134fa9e4066Sahrens #define DS_MODE_IS_READONLY(x) ((x) & DS_MODE_READONLY) 135e1930233Sbonwick #define DS_MODE_INCONSISTENT 0x10 136e1930233Sbonwick #define DS_MODE_IS_INCONSISTENT(x) ((x) & DS_MODE_INCONSISTENT) 137fa9e4066Sahrens 138*0b69c2f0Sahrens #define DS_FIND_SNAPSHOTS (1<<0) 139*0b69c2f0Sahrens #define DS_FIND_CHILDREN (1<<1) 140fa9e4066Sahrens 141fa9e4066Sahrens /* 142fa9e4066Sahrens * The maximum number of bytes that can be accessed as part of one 143fa9e4066Sahrens * operation, including metadata. 144fa9e4066Sahrens */ 145fa9e4066Sahrens #define DMU_MAX_ACCESS (10<<20) /* 10MB */ 146fa9e4066Sahrens 147fa9e4066Sahrens /* 148fa9e4066Sahrens * Public routines to create, destroy, open, and close objsets. 149fa9e4066Sahrens */ 150fa9e4066Sahrens int dmu_objset_open(const char *name, dmu_objset_type_t type, int mode, 151fa9e4066Sahrens objset_t **osp); 152fa9e4066Sahrens void dmu_objset_close(objset_t *os); 153436b2950Sperrin int dmu_objset_evict_dbufs(objset_t *os, int try); 154fa9e4066Sahrens int dmu_objset_create(const char *name, dmu_objset_type_t type, 155fa9e4066Sahrens objset_t *clone_parent, 156fa9e4066Sahrens void (*func)(objset_t *os, void *arg, dmu_tx_t *tx), void *arg); 157fa9e4066Sahrens int dmu_objset_destroy(const char *name); 1581d452cf5Sahrens int dmu_snapshots_destroy(char *fsname, char *snapname); 159fa9e4066Sahrens int dmu_objset_rollback(const char *name); 1601d452cf5Sahrens int dmu_objset_snapshot(char *fsname, char *snapname, boolean_t recursive); 161fa9e4066Sahrens int dmu_objset_rename(const char *name, const char *newname); 162fa9e4066Sahrens void dmu_objset_set_quota(objset_t *os, uint64_t quota); 163fa9e4066Sahrens uint64_t dmu_objset_get_quota(objset_t *os); 164fa9e4066Sahrens int dmu_objset_request_reservation(objset_t *os, uint64_t reservation); 1651d452cf5Sahrens int dmu_objset_find(char *name, int func(char *, void *), void *arg, 166fa9e4066Sahrens int flags); 167fa9e4066Sahrens void dmu_objset_byteswap(void *buf, size_t size); 168fa9e4066Sahrens 169fa9e4066Sahrens typedef struct dmu_buf { 170fa9e4066Sahrens uint64_t db_object; /* object that this buffer is part of */ 171fa9e4066Sahrens uint64_t db_offset; /* byte offset in this object */ 172fa9e4066Sahrens uint64_t db_size; /* size of buffer in bytes */ 173fa9e4066Sahrens void *db_data; /* data in buffer */ 174fa9e4066Sahrens } dmu_buf_t; 175fa9e4066Sahrens 176fa9e4066Sahrens typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); 177fa9e4066Sahrens 178fa9e4066Sahrens /* 179fa9e4066Sahrens * Callback function to perform byte swapping on a block. 180fa9e4066Sahrens */ 181fa9e4066Sahrens typedef void dmu_byteswap_func_t(void *buf, size_t size); 182fa9e4066Sahrens 18399653d4eSeschrock /* 18499653d4eSeschrock * The names of zap entries in the DIRECTORY_OBJECT of the MOS. 18599653d4eSeschrock */ 186fa9e4066Sahrens #define DMU_POOL_DIRECTORY_OBJECT 1 187fa9e4066Sahrens #define DMU_POOL_CONFIG "config" 188fa9e4066Sahrens #define DMU_POOL_ROOT_DATASET "root_dataset" 189fa9e4066Sahrens #define DMU_POOL_SYNC_BPLIST "sync_bplist" 190ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" 191ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_LAST "errlog_last" 19299653d4eSeschrock #define DMU_POOL_SPARES "spares" 19399653d4eSeschrock #define DMU_POOL_DEFLATE "deflate" 194fa9e4066Sahrens 195fa9e4066Sahrens /* 196fa9e4066Sahrens * Allocate an object from this objset. The range of object numbers 197fa9e4066Sahrens * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. 198fa9e4066Sahrens * 199fa9e4066Sahrens * The transaction must be assigned to a txg. The newly allocated 200fa9e4066Sahrens * object will be "held" in the transaction (ie. you can modify the 201fa9e4066Sahrens * newly allocated object in this transaction). 202fa9e4066Sahrens * 203fa9e4066Sahrens * dmu_object_alloc() chooses an object and returns it in *objectp. 204fa9e4066Sahrens * 205fa9e4066Sahrens * dmu_object_claim() allocates a specific object number. If that 206fa9e4066Sahrens * number is already allocated, it fails and returns EEXIST. 207fa9e4066Sahrens * 208fa9e4066Sahrens * Return 0 on success, or ENOSPC or EEXIST as specified above. 209fa9e4066Sahrens */ 210fa9e4066Sahrens uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, 211fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 212fa9e4066Sahrens int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, 213fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 214fa9e4066Sahrens int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, 215fa9e4066Sahrens int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx); 216fa9e4066Sahrens 217fa9e4066Sahrens /* 218fa9e4066Sahrens * Free an object from this objset. 219fa9e4066Sahrens * 220fa9e4066Sahrens * The object's data will be freed as well (ie. you don't need to call 221fa9e4066Sahrens * dmu_free(object, 0, -1, tx)). 222fa9e4066Sahrens * 223fa9e4066Sahrens * The object need not be held in the transaction. 224fa9e4066Sahrens * 225fa9e4066Sahrens * If there are any holds on this object's buffers (via dmu_buf_hold()), 226fa9e4066Sahrens * or tx holds on the object (via dmu_tx_hold_object()), you can not 227fa9e4066Sahrens * free it; it fails and returns EBUSY. 228fa9e4066Sahrens * 229fa9e4066Sahrens * If the object is not allocated, it fails and returns ENOENT. 230fa9e4066Sahrens * 231fa9e4066Sahrens * Return 0 on success, or EBUSY or ENOENT as specified above. 232fa9e4066Sahrens */ 233fa9e4066Sahrens int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); 234fa9e4066Sahrens 235fa9e4066Sahrens /* 236fa9e4066Sahrens * Find the next allocated or free object. 237fa9e4066Sahrens * 238fa9e4066Sahrens * The objectp parameter is in-out. It will be updated to be the next 239fa9e4066Sahrens * object which is allocated. 240fa9e4066Sahrens * 241fa9e4066Sahrens * XXX Can only be called on a objset with no dirty data. 242fa9e4066Sahrens * 243fa9e4066Sahrens * Returns 0 on success, or ENOENT if there are no more objects. 244fa9e4066Sahrens */ 245fa9e4066Sahrens int dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole); 246fa9e4066Sahrens 247fa9e4066Sahrens /* 248fa9e4066Sahrens * Set the data blocksize for an object. 249fa9e4066Sahrens * 250fa9e4066Sahrens * The object cannot have any blocks allcated beyond the first. If 251fa9e4066Sahrens * the first block is allocated already, the new size must be greater 252fa9e4066Sahrens * than the current block size. If these conditions are not met, 253fa9e4066Sahrens * ENOTSUP will be returned. 254fa9e4066Sahrens * 255fa9e4066Sahrens * Returns 0 on success, or EBUSY if there are any holds on the object 256fa9e4066Sahrens * contents, or ENOTSUP as described above. 257fa9e4066Sahrens */ 258fa9e4066Sahrens int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, 259fa9e4066Sahrens int ibs, dmu_tx_t *tx); 260fa9e4066Sahrens 261fa9e4066Sahrens /* 262fa9e4066Sahrens * Set the checksum property on a dnode. The new checksum algorithm will 263fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 264fa9e4066Sahrens */ 265fa9e4066Sahrens void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, 266fa9e4066Sahrens dmu_tx_t *tx); 267fa9e4066Sahrens 268fa9e4066Sahrens /* 269fa9e4066Sahrens * Set the compress property on a dnode. The new compression algorithm will 270fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 271fa9e4066Sahrens */ 272fa9e4066Sahrens void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, 273fa9e4066Sahrens dmu_tx_t *tx); 274fa9e4066Sahrens 27544cd46caSbillm /* 27644cd46caSbillm * Decide how many copies of a given block we should make. Can be from 27744cd46caSbillm * 1 to SPA_DVAS_PER_BP. 27844cd46caSbillm */ 27944cd46caSbillm int dmu_get_replication_level(struct spa *spa, struct zbookmark *zb, 28044cd46caSbillm dmu_object_type_t ot); 281fa9e4066Sahrens /* 282fa9e4066Sahrens * The bonus data is accessed more or less like a regular buffer. 283fa9e4066Sahrens * You must dmu_bonus_hold() to get the buffer, which will give you a 284fa9e4066Sahrens * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus 285fa9e4066Sahrens * data. As with any normal buffer, you must call dmu_buf_read() to 286fa9e4066Sahrens * read db_data, dmu_buf_will_dirty() before modifying it, and the 287fa9e4066Sahrens * object must be held in an assigned transaction before calling 288fa9e4066Sahrens * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus 289fa9e4066Sahrens * buffer as well. You must release your hold with dmu_buf_rele(). 290fa9e4066Sahrens */ 291ea8dc4b6Seschrock int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 292fa9e4066Sahrens int dmu_bonus_max(void); 293fa9e4066Sahrens 294fa9e4066Sahrens /* 295fa9e4066Sahrens * Obtain the DMU buffer from the specified object which contains the 296fa9e4066Sahrens * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so 297fa9e4066Sahrens * that it will remain in memory. You must release the hold with 298fa9e4066Sahrens * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your 299fa9e4066Sahrens * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. 300fa9e4066Sahrens * 301fa9e4066Sahrens * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill 302fa9e4066Sahrens * on the returned buffer before reading or writing the buffer's 303fa9e4066Sahrens * db_data. The comments for those routines describe what particular 304fa9e4066Sahrens * operations are valid after calling them. 305fa9e4066Sahrens * 306fa9e4066Sahrens * The object number must be a valid, allocated object number. 307fa9e4066Sahrens */ 308ea8dc4b6Seschrock int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, 309ea8dc4b6Seschrock void *tag, dmu_buf_t **); 310fa9e4066Sahrens void dmu_buf_add_ref(dmu_buf_t *db, void* tag); 311ea8dc4b6Seschrock void dmu_buf_rele(dmu_buf_t *db, void *tag); 312fa9e4066Sahrens uint64_t dmu_buf_refcount(dmu_buf_t *db); 313fa9e4066Sahrens 314fa9e4066Sahrens /* 315fa9e4066Sahrens * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a 316fa9e4066Sahrens * range of an object. A pointer to an array of dmu_buf_t*'s is 317fa9e4066Sahrens * returned (in *dbpp). 318fa9e4066Sahrens * 319fa9e4066Sahrens * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and 320fa9e4066Sahrens * frees the array. The hold on the array of buffers MUST be released 321fa9e4066Sahrens * with dmu_buf_rele_array. You can NOT release the hold on each buffer 322fa9e4066Sahrens * individually with dmu_buf_rele. 323fa9e4066Sahrens */ 324ea8dc4b6Seschrock int dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset, 325ea8dc4b6Seschrock uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 32613506d1eSmaybee int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, 32713506d1eSmaybee uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 328ea8dc4b6Seschrock void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); 329fa9e4066Sahrens 330fa9e4066Sahrens /* 331fa9e4066Sahrens * Returns NULL on success, or the existing user ptr if it's already 332fa9e4066Sahrens * been set. 333fa9e4066Sahrens * 334fa9e4066Sahrens * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). 335fa9e4066Sahrens * 336fa9e4066Sahrens * user_data_ptr_ptr should be NULL, or a pointer to a pointer which 337fa9e4066Sahrens * will be set to db->db_data when you are allowed to access it. Note 338fa9e4066Sahrens * that db->db_data (the pointer) can change when you do dmu_buf_read(), 339fa9e4066Sahrens * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). 340fa9e4066Sahrens * *user_data_ptr_ptr will be set to the new value when it changes. 341fa9e4066Sahrens * 342fa9e4066Sahrens * If non-NULL, pageout func will be called when this buffer is being 343fa9e4066Sahrens * excised from the cache, so that you can clean up the data structure 344fa9e4066Sahrens * pointed to by user_ptr. 345fa9e4066Sahrens * 346fa9e4066Sahrens * dmu_evict_user() will call the pageout func for all buffers in a 347fa9e4066Sahrens * objset with a given pageout func. 348fa9e4066Sahrens */ 349fa9e4066Sahrens void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, 350fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 351fa9e4066Sahrens /* 352fa9e4066Sahrens * set_user_ie is the same as set_user, but request immediate eviction 353fa9e4066Sahrens * when hold count goes to zero. 354fa9e4066Sahrens */ 355fa9e4066Sahrens void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, 356fa9e4066Sahrens void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); 357fa9e4066Sahrens void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, 358fa9e4066Sahrens void *user_ptr, void *user_data_ptr_ptr, 359fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 360fa9e4066Sahrens void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); 361fa9e4066Sahrens 362fa9e4066Sahrens void dmu_buf_hold_data(dmu_buf_t *db); 363fa9e4066Sahrens void dmu_buf_rele_data(dmu_buf_t *db); 364fa9e4066Sahrens 365fa9e4066Sahrens /* 366fa9e4066Sahrens * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. 367fa9e4066Sahrens */ 368fa9e4066Sahrens void *dmu_buf_get_user(dmu_buf_t *db); 369fa9e4066Sahrens 370fa9e4066Sahrens /* 371fa9e4066Sahrens * Indicate that you are going to modify the buffer's data (db_data). 372fa9e4066Sahrens * 373fa9e4066Sahrens * The transaction (tx) must be assigned to a txg (ie. you've called 374fa9e4066Sahrens * dmu_tx_assign()). The buffer's object must be held in the tx 375fa9e4066Sahrens * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). 376fa9e4066Sahrens */ 377fa9e4066Sahrens void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); 378fa9e4066Sahrens 379fa9e4066Sahrens /* 380fa9e4066Sahrens * You must create a transaction, then hold the objects which you will 381fa9e4066Sahrens * (or might) modify as part of this transaction. Then you must assign 382fa9e4066Sahrens * the transaction to a transaction group. Once the transaction has 383fa9e4066Sahrens * been assigned, you can modify buffers which belong to held objects as 384fa9e4066Sahrens * part of this transaction. You can't modify buffers before the 385fa9e4066Sahrens * transaction has been assigned; you can't modify buffers which don't 386fa9e4066Sahrens * belong to objects which this transaction holds; you can't hold 387fa9e4066Sahrens * objects once the transaction has been assigned. You may hold an 388fa9e4066Sahrens * object which you are going to free (with dmu_object_free()), but you 389fa9e4066Sahrens * don't have to. 390fa9e4066Sahrens * 391fa9e4066Sahrens * You can abort the transaction before it has been assigned. 392fa9e4066Sahrens * 393fa9e4066Sahrens * Note that you may hold buffers (with dmu_buf_hold) at any time, 394fa9e4066Sahrens * regardless of transaction state. 395fa9e4066Sahrens */ 396fa9e4066Sahrens 397fa9e4066Sahrens #define DMU_NEW_OBJECT (-1ULL) 398fa9e4066Sahrens #define DMU_OBJECT_END (-1ULL) 399fa9e4066Sahrens 400fa9e4066Sahrens dmu_tx_t *dmu_tx_create(objset_t *os); 401fa9e4066Sahrens void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); 402fa9e4066Sahrens void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, 403fa9e4066Sahrens uint64_t len); 404ea8dc4b6Seschrock void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, char *name); 405fa9e4066Sahrens void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); 406fa9e4066Sahrens void dmu_tx_abort(dmu_tx_t *tx); 407fa9e4066Sahrens int dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how); 4088a2f1b91Sahrens void dmu_tx_wait(dmu_tx_t *tx); 409fa9e4066Sahrens void dmu_tx_commit(dmu_tx_t *tx); 410fa9e4066Sahrens 411fa9e4066Sahrens /* 412fa9e4066Sahrens * Free up the data blocks for a defined range of a file. If size is 413fa9e4066Sahrens * zero, the range from offset to end-of-file is freed. 414fa9e4066Sahrens */ 415ea8dc4b6Seschrock int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, 416fa9e4066Sahrens uint64_t size, dmu_tx_t *tx); 417fa9e4066Sahrens 418fa9e4066Sahrens /* 419fa9e4066Sahrens * Convenience functions. 420fa9e4066Sahrens * 421fa9e4066Sahrens * Canfail routines will return 0 on success, or an errno if there is a 422fa9e4066Sahrens * nonrecoverable I/O error. 423fa9e4066Sahrens */ 424ea8dc4b6Seschrock int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 425fa9e4066Sahrens void *buf); 426fa9e4066Sahrens void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 427fa9e4066Sahrens const void *buf, dmu_tx_t *tx); 428fa9e4066Sahrens int dmu_write_uio(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 429fa9e4066Sahrens struct uio *uio, dmu_tx_t *tx); 430fa9e4066Sahrens 431fa9e4066Sahrens /* 432fa9e4066Sahrens * Asynchronously try to read in the data. 433fa9e4066Sahrens */ 434fa9e4066Sahrens void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, 435fa9e4066Sahrens uint64_t len); 436fa9e4066Sahrens 437fa9e4066Sahrens typedef struct dmu_object_info { 438fa9e4066Sahrens /* All sizes are in bytes. */ 439fa9e4066Sahrens uint32_t doi_data_block_size; 440fa9e4066Sahrens uint32_t doi_metadata_block_size; 441fa9e4066Sahrens uint64_t doi_bonus_size; 442fa9e4066Sahrens dmu_object_type_t doi_type; 443fa9e4066Sahrens dmu_object_type_t doi_bonus_type; 444fa9e4066Sahrens uint8_t doi_indirection; /* 2 = dnode->indirect->data */ 445fa9e4066Sahrens uint8_t doi_checksum; 446fa9e4066Sahrens uint8_t doi_compress; 447fa9e4066Sahrens uint8_t doi_pad[5]; 448fa9e4066Sahrens /* Values below are number of 512-byte blocks. */ 449fa9e4066Sahrens uint64_t doi_physical_blks; /* data + metadata */ 450fa9e4066Sahrens uint64_t doi_max_block_offset; 451fa9e4066Sahrens } dmu_object_info_t; 452fa9e4066Sahrens 453fa9e4066Sahrens typedef struct dmu_object_type_info { 454fa9e4066Sahrens dmu_byteswap_func_t *ot_byteswap; 455fa9e4066Sahrens boolean_t ot_metadata; 456fa9e4066Sahrens char *ot_name; 457fa9e4066Sahrens } dmu_object_type_info_t; 458fa9e4066Sahrens 459fa9e4066Sahrens extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; 460fa9e4066Sahrens 461fa9e4066Sahrens /* 462fa9e4066Sahrens * Get information on a DMU object. 463fa9e4066Sahrens * 464fa9e4066Sahrens * Return 0 on success or ENOENT if object is not allocated. 465fa9e4066Sahrens * 466fa9e4066Sahrens * If doi is NULL, just indicates whether the object exists. 467fa9e4066Sahrens */ 468fa9e4066Sahrens int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); 469fa9e4066Sahrens void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); 470fa9e4066Sahrens void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); 471fa9e4066Sahrens void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, 472fa9e4066Sahrens u_longlong_t *nblk512); 473fa9e4066Sahrens 474fa9e4066Sahrens /* 475fa9e4066Sahrens * Get the maximum nonzero offset in the object (ie. this offset and all 476fa9e4066Sahrens * offsets following are zero). 477fa9e4066Sahrens * 478fa9e4066Sahrens * XXX Perhaps integrate this with dmu_object_info(), although that 479fa9e4066Sahrens * would then have to bring in the indirect blocks. 480fa9e4066Sahrens */ 481fa9e4066Sahrens uint64_t dmu_object_max_nonzero_offset(objset_t *os, uint64_t object); 482fa9e4066Sahrens 483fa9e4066Sahrens typedef struct dmu_objset_stats { 484fa9e4066Sahrens dmu_objset_type_t dds_type; 485fa9e4066Sahrens uint8_t dds_is_snapshot; 48631fd60d3Sahrens uint8_t dds_inconsistent; 48731fd60d3Sahrens uint8_t dds_pad[2]; 488fa9e4066Sahrens 489fa9e4066Sahrens uint64_t dds_creation_time; 490fa9e4066Sahrens uint64_t dds_creation_txg; 491fa9e4066Sahrens 492fa9e4066Sahrens char dds_clone_of[MAXNAMELEN]; 493fa9e4066Sahrens 494fa9e4066Sahrens /* How much data is there in this objset? */ 495fa9e4066Sahrens 496fa9e4066Sahrens /* 497fa9e4066Sahrens * Space referenced, taking into account pending writes and 498fa9e4066Sahrens * frees. Only relavent to filesystems and snapshots (not 499fa9e4066Sahrens * collections). 500fa9e4066Sahrens */ 501fa9e4066Sahrens uint64_t dds_space_refd; 502fa9e4066Sahrens 503fa9e4066Sahrens /* 504fa9e4066Sahrens * Space "used", taking into account pending writes and frees, and 505fa9e4066Sahrens * children's reservations (in bytes). This is the amount of 506fa9e4066Sahrens * space that will be freed if this and all dependent items are 507fa9e4066Sahrens * destroyed (eg. child datasets, objsets, and snapshots). So 508fa9e4066Sahrens * for snapshots, this is the amount of space unique to this 509fa9e4066Sahrens * snapshot. 510fa9e4066Sahrens */ 511fa9e4066Sahrens uint64_t dds_space_used; 512fa9e4066Sahrens 513fa9e4066Sahrens /* 514fa9e4066Sahrens * Compressed and uncompressed bytes consumed. Does not take 515fa9e4066Sahrens * into account reservations. Used for computing compression 516fa9e4066Sahrens * ratio. 517fa9e4066Sahrens */ 518fa9e4066Sahrens uint64_t dds_compressed_bytes; 519fa9e4066Sahrens uint64_t dds_uncompressed_bytes; 520fa9e4066Sahrens 521fa9e4066Sahrens /* 522fa9e4066Sahrens * The ds_fsid_guid is a 56-bit ID that can change to avoid 523fa9e4066Sahrens * collisions. The ds_guid is a 64-bit ID that will never 524fa9e4066Sahrens * change, so there is a small probability that it will collide. 525fa9e4066Sahrens */ 526fa9e4066Sahrens uint64_t dds_fsid_guid; 527fa9e4066Sahrens 528fa9e4066Sahrens uint64_t dds_objects_used; /* number of objects used */ 529fa9e4066Sahrens uint64_t dds_objects_avail; /* number of objects available */ 530fa9e4066Sahrens 531fa9e4066Sahrens uint64_t dds_num_clones; /* number of clones of this */ 532fa9e4066Sahrens 533fa9e4066Sahrens /* The dataset's administratively-set quota, in bytes. */ 534fa9e4066Sahrens uint64_t dds_quota; 535fa9e4066Sahrens 536fa9e4066Sahrens /* The dataset's administratively-set reservation, in bytes */ 537fa9e4066Sahrens uint64_t dds_reserved; 538fa9e4066Sahrens 539fa9e4066Sahrens /* 540fa9e4066Sahrens * The amount of additional space that this dataset can consume. 541fa9e4066Sahrens * Takes into account quotas & reservations. 542fa9e4066Sahrens * (Assuming that no other datasets consume it first.) 543fa9e4066Sahrens */ 544fa9e4066Sahrens uint64_t dds_available; 545fa9e4066Sahrens 546fa9e4066Sahrens /* 547ea8dc4b6Seschrock * Used for debugging purposes 548fa9e4066Sahrens */ 549fa9e4066Sahrens uint64_t dds_last_txg; 550fa9e4066Sahrens } dmu_objset_stats_t; 551fa9e4066Sahrens 552fa9e4066Sahrens /* 553fa9e4066Sahrens * Get stats on a dataset. 554fa9e4066Sahrens */ 555fa9e4066Sahrens void dmu_objset_stats(objset_t *os, dmu_objset_stats_t *dds); 556fa9e4066Sahrens 557fa9e4066Sahrens int dmu_objset_is_snapshot(objset_t *os); 558fa9e4066Sahrens 559fa9e4066Sahrens extern struct spa *dmu_objset_spa(objset_t *os); 560fa9e4066Sahrens extern struct zilog *dmu_objset_zil(objset_t *os); 561fa9e4066Sahrens extern struct dsl_pool *dmu_objset_pool(objset_t *os); 562fa9e4066Sahrens extern struct dsl_dataset *dmu_objset_ds(objset_t *os); 563fa9e4066Sahrens extern void dmu_objset_name(objset_t *os, char *buf); 564fa9e4066Sahrens extern dmu_objset_type_t dmu_objset_type(objset_t *os); 565fa9e4066Sahrens extern uint64_t dmu_objset_id(objset_t *os); 566fa9e4066Sahrens extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, 567fa9e4066Sahrens uint64_t *id, uint64_t *offp); 56887e5029aSahrens extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, 56987e5029aSahrens uint64_t *idp, uint64_t *offp); 570fa9e4066Sahrens 571fa9e4066Sahrens /* 572fa9e4066Sahrens * Return the txg number for the given assigned transaction. 573fa9e4066Sahrens */ 57487e5029aSahrens uint64_t dmu_tx_get_txg(dmu_tx_t *tx); 575fa9e4066Sahrens 576fa9e4066Sahrens /* 577fa9e4066Sahrens * Synchronous write. 578c5c6ffa0Smaybee * If a parent zio is provided this function initiates a write on the 579c5c6ffa0Smaybee * provided buffer as a child of the parent zio. 580c5c6ffa0Smaybee * In the absense of a parent zio, the write is completed synchronously. 581c5c6ffa0Smaybee * At write completion, blk is filled with the bp of the written block. 582fa9e4066Sahrens * Note that while the data covered by this function will be on stable 583c5c6ffa0Smaybee * storage when the write completes this new data does not become a 584fa9e4066Sahrens * permanent part of the file until the associated transaction commits. 585fa9e4066Sahrens */ 586c5c6ffa0Smaybee typedef void dmu_sync_cb_t(dmu_buf_t *db, void *arg); 587c5c6ffa0Smaybee int dmu_sync(struct zio *zio, dmu_buf_t *db, 588c5c6ffa0Smaybee struct blkptr *bp, uint64_t txg, dmu_sync_cb_t *done, void *arg); 589fa9e4066Sahrens 590fa9e4066Sahrens /* 591fa9e4066Sahrens * Find the next hole or data block in file starting at *off 592fa9e4066Sahrens * Return found offset in *off. Return ESRCH for end of file. 593fa9e4066Sahrens */ 594fa9e4066Sahrens int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, 595fa9e4066Sahrens uint64_t *off); 596fa9e4066Sahrens 597fa9e4066Sahrens /* 598fa9e4066Sahrens * Initial setup and final teardown. 599fa9e4066Sahrens */ 600fa9e4066Sahrens extern void dmu_init(void); 601fa9e4066Sahrens extern void dmu_fini(void); 602fa9e4066Sahrens 603fa9e4066Sahrens typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, 604fa9e4066Sahrens uint64_t object, uint64_t offset, int len); 605fa9e4066Sahrens void dmu_traverse_objset(objset_t *os, uint64_t txg_start, 606fa9e4066Sahrens dmu_traverse_cb_t cb, void *arg); 607fa9e4066Sahrens 608fa9e4066Sahrens int dmu_sendbackup(objset_t *tosnap, objset_t *fromsnap, struct vnode *vp); 609ea8dc4b6Seschrock int dmu_recvbackup(char *tosnap, struct drr_begin *drrb, uint64_t *sizep, 610fa9e4066Sahrens struct vnode *vp, uint64_t voffset); 611fa9e4066Sahrens 612fa9e4066Sahrens /* CRC64 table */ 613fa9e4066Sahrens #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ 614fa9e4066Sahrens extern uint64_t zfs_crc64_table[256]; 615fa9e4066Sahrens 616fa9e4066Sahrens #ifdef __cplusplus 617fa9e4066Sahrens } 618fa9e4066Sahrens #endif 619fa9e4066Sahrens 620fa9e4066Sahrens #endif /* _SYS_DMU_H */ 621