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 */ 21ad135b5dSChristopher Siden 22fa9e4066Sahrens /* 2394d1a210STim Haley * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. 24713d6c20SMatthew Ahrens * Copyright (c) 2013 by Delphix. All rights reserved. 25ec5cf9d5SAlexander Stetsenko * Copyright 2011 Nexenta Systems, Inc. All rights reserved. 264e3c9f44SBill Pijewski * Copyright (c) 2012, Joyent, Inc. All rights reserved. 27ec5cf9d5SAlexander Stetsenko */ 28fa9e4066Sahrens 2955da60b9SMark J Musante /* Portions Copyright 2010 Robert Milkowski */ 3055da60b9SMark J Musante 31fa9e4066Sahrens #ifndef _SYS_DMU_H 32fa9e4066Sahrens #define _SYS_DMU_H 33fa9e4066Sahrens 34fa9e4066Sahrens /* 35fa9e4066Sahrens * This file describes the interface that the DMU provides for its 36fa9e4066Sahrens * consumers. 37fa9e4066Sahrens * 38fa9e4066Sahrens * The DMU also interacts with the SPA. That interface is described in 39fa9e4066Sahrens * dmu_spa.h. 40fa9e4066Sahrens */ 41fa9e4066Sahrens 42fa9e4066Sahrens #include <sys/inttypes.h> 43fa9e4066Sahrens #include <sys/types.h> 44fa9e4066Sahrens #include <sys/param.h> 45ecd6cf80Smarks #include <sys/cred.h> 4671eb0538SChris Kirby #include <sys/time.h> 474445fffbSMatthew Ahrens #include <sys/fs/zfs.h> 48fa9e4066Sahrens 49fa9e4066Sahrens #ifdef __cplusplus 50fa9e4066Sahrens extern "C" { 51fa9e4066Sahrens #endif 52fa9e4066Sahrens 53fa9e4066Sahrens struct uio; 54c242f9a0Schunli zhang - Sun Microsystems - Irvine United States struct xuio; 5544eda4d7Smaybee struct page; 56fa9e4066Sahrens struct vnode; 57fa9e4066Sahrens struct spa; 58fa9e4066Sahrens struct zilog; 59fa9e4066Sahrens struct zio; 60fa9e4066Sahrens struct blkptr; 61fa9e4066Sahrens struct zap_cursor; 62fa9e4066Sahrens struct dsl_dataset; 63fa9e4066Sahrens struct dsl_pool; 64fa9e4066Sahrens struct dnode; 65fa9e4066Sahrens struct drr_begin; 66fa9e4066Sahrens struct drr_end; 6744cd46caSbillm struct zbookmark; 6844cd46caSbillm struct spa; 69a2eea2e1Sahrens struct nvlist; 702fdbea25SAleksandr Guzovskiy struct arc_buf; 71b24ab676SJeff Bonwick struct zio_prop; 720a586ceaSMark Shellenbaum struct sa_handle; 73fa9e4066Sahrens 74fa9e4066Sahrens typedef struct objset objset_t; 75fa9e4066Sahrens typedef struct dmu_tx dmu_tx_t; 76fa9e4066Sahrens typedef struct dsl_dir dsl_dir_t; 77fa9e4066Sahrens 78ad135b5dSChristopher Siden typedef enum dmu_object_byteswap { 79ad135b5dSChristopher Siden DMU_BSWAP_UINT8, 80ad135b5dSChristopher Siden DMU_BSWAP_UINT16, 81ad135b5dSChristopher Siden DMU_BSWAP_UINT32, 82ad135b5dSChristopher Siden DMU_BSWAP_UINT64, 83ad135b5dSChristopher Siden DMU_BSWAP_ZAP, 84ad135b5dSChristopher Siden DMU_BSWAP_DNODE, 85ad135b5dSChristopher Siden DMU_BSWAP_OBJSET, 86ad135b5dSChristopher Siden DMU_BSWAP_ZNODE, 87ad135b5dSChristopher Siden DMU_BSWAP_OLDACL, 88ad135b5dSChristopher Siden DMU_BSWAP_ACL, 89ad135b5dSChristopher Siden /* 90ad135b5dSChristopher Siden * Allocating a new byteswap type number makes the on-disk format 91ad135b5dSChristopher Siden * incompatible with any other format that uses the same number. 92ad135b5dSChristopher Siden * 93ad135b5dSChristopher Siden * Data can usually be structured to work with one of the 94ad135b5dSChristopher Siden * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. 95ad135b5dSChristopher Siden */ 96ad135b5dSChristopher Siden DMU_BSWAP_NUMFUNCS 97ad135b5dSChristopher Siden } dmu_object_byteswap_t; 98ad135b5dSChristopher Siden 99ad135b5dSChristopher Siden #define DMU_OT_NEWTYPE 0x80 100ad135b5dSChristopher Siden #define DMU_OT_METADATA 0x40 101ad135b5dSChristopher Siden #define DMU_OT_BYTESWAP_MASK 0x3f 102ad135b5dSChristopher Siden 103ad135b5dSChristopher Siden /* 104ad135b5dSChristopher Siden * Defines a uint8_t object type. Object types specify if the data 105ad135b5dSChristopher Siden * in the object is metadata (boolean) and how to byteswap the data 106ad135b5dSChristopher Siden * (dmu_object_byteswap_t). 107ad135b5dSChristopher Siden */ 108ad135b5dSChristopher Siden #define DMU_OT(byteswap, metadata) \ 109ad135b5dSChristopher Siden (DMU_OT_NEWTYPE | \ 110ad135b5dSChristopher Siden ((metadata) ? DMU_OT_METADATA : 0) | \ 111ad135b5dSChristopher Siden ((byteswap) & DMU_OT_BYTESWAP_MASK)) 112ad135b5dSChristopher Siden 113ad135b5dSChristopher Siden #define DMU_OT_IS_VALID(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 114ad135b5dSChristopher Siden ((ot) & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS : \ 115ad135b5dSChristopher Siden (ot) < DMU_OT_NUMTYPES) 116ad135b5dSChristopher Siden 117ad135b5dSChristopher Siden #define DMU_OT_IS_METADATA(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 118ad135b5dSChristopher Siden ((ot) & DMU_OT_METADATA) : \ 119ad135b5dSChristopher Siden dmu_ot[(ot)].ot_metadata) 120ad135b5dSChristopher Siden 121ad135b5dSChristopher Siden #define DMU_OT_BYTESWAP(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 122ad135b5dSChristopher Siden ((ot) & DMU_OT_BYTESWAP_MASK) : \ 123ad135b5dSChristopher Siden dmu_ot[(ot)].ot_byteswap) 124ad135b5dSChristopher Siden 125fa9e4066Sahrens typedef enum dmu_object_type { 126fa9e4066Sahrens DMU_OT_NONE, 127fa9e4066Sahrens /* general: */ 128fa9e4066Sahrens DMU_OT_OBJECT_DIRECTORY, /* ZAP */ 129fa9e4066Sahrens DMU_OT_OBJECT_ARRAY, /* UINT64 */ 130fa9e4066Sahrens DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ 131fa9e4066Sahrens DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ 132cde58dbcSMatthew Ahrens DMU_OT_BPOBJ, /* UINT64 */ 133cde58dbcSMatthew Ahrens DMU_OT_BPOBJ_HDR, /* UINT64 */ 134fa9e4066Sahrens /* spa: */ 135fa9e4066Sahrens DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ 136fa9e4066Sahrens DMU_OT_SPACE_MAP, /* UINT64 */ 137fa9e4066Sahrens /* zil: */ 138fa9e4066Sahrens DMU_OT_INTENT_LOG, /* UINT64 */ 139fa9e4066Sahrens /* dmu: */ 140fa9e4066Sahrens DMU_OT_DNODE, /* DNODE */ 141fa9e4066Sahrens DMU_OT_OBJSET, /* OBJSET */ 142fa9e4066Sahrens /* dsl: */ 1431649cd4bStabriz DMU_OT_DSL_DIR, /* UINT64 */ 14487e5029aSahrens DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ 14587e5029aSahrens DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ 146fa9e4066Sahrens DMU_OT_DSL_PROPS, /* ZAP */ 1471649cd4bStabriz DMU_OT_DSL_DATASET, /* UINT64 */ 148fa9e4066Sahrens /* zpl: */ 149fa9e4066Sahrens DMU_OT_ZNODE, /* ZNODE */ 150da6c28aaSamw DMU_OT_OLDACL, /* Old ACL */ 151fa9e4066Sahrens DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ 152fa9e4066Sahrens DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ 153fa9e4066Sahrens DMU_OT_MASTER_NODE, /* ZAP */ 154893a6d32Sahrens DMU_OT_UNLINKED_SET, /* ZAP */ 155fa9e4066Sahrens /* zvol: */ 156fa9e4066Sahrens DMU_OT_ZVOL, /* UINT8 */ 157fa9e4066Sahrens DMU_OT_ZVOL_PROP, /* ZAP */ 158fa9e4066Sahrens /* other; for testing only! */ 159fa9e4066Sahrens DMU_OT_PLAIN_OTHER, /* UINT8 */ 160fa9e4066Sahrens DMU_OT_UINT64_OTHER, /* UINT64 */ 161fa9e4066Sahrens DMU_OT_ZAP_OTHER, /* ZAP */ 162ea8dc4b6Seschrock /* new object types: */ 163ea8dc4b6Seschrock DMU_OT_ERROR_LOG, /* ZAP */ 16406eeb2adSek DMU_OT_SPA_HISTORY, /* UINT8 */ 16506eeb2adSek DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ 166b1b8ab34Slling DMU_OT_POOL_PROPS, /* ZAP */ 167ecd6cf80Smarks DMU_OT_DSL_PERMS, /* ZAP */ 168da6c28aaSamw DMU_OT_ACL, /* ACL */ 169da6c28aaSamw DMU_OT_SYSACL, /* SYSACL */ 170da6c28aaSamw DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ 171da6c28aaSamw DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ 172088f3894Sahrens DMU_OT_NEXT_CLONES, /* ZAP */ 1733f9d6ad7SLin Ling DMU_OT_SCAN_QUEUE, /* ZAP */ 17414843421SMatthew Ahrens DMU_OT_USERGROUP_USED, /* ZAP */ 17514843421SMatthew Ahrens DMU_OT_USERGROUP_QUOTA, /* ZAP */ 176842727c2SChris Kirby DMU_OT_USERREFS, /* ZAP */ 177b24ab676SJeff Bonwick DMU_OT_DDT_ZAP, /* ZAP */ 178b24ab676SJeff Bonwick DMU_OT_DDT_STATS, /* ZAP */ 1790a586ceaSMark Shellenbaum DMU_OT_SA, /* System attr */ 1800a586ceaSMark Shellenbaum DMU_OT_SA_MASTER_NODE, /* ZAP */ 1810a586ceaSMark Shellenbaum DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ 1820a586ceaSMark Shellenbaum DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ 1833f9d6ad7SLin Ling DMU_OT_SCAN_XLATE, /* ZAP */ 1843f9d6ad7SLin Ling DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ 185cde58dbcSMatthew Ahrens DMU_OT_DEADLIST, /* ZAP */ 186cde58dbcSMatthew Ahrens DMU_OT_DEADLIST_HDR, /* UINT64 */ 187cde58dbcSMatthew Ahrens DMU_OT_DSL_CLONES, /* ZAP */ 188cde58dbcSMatthew Ahrens DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ 189ad135b5dSChristopher Siden /* 190ad135b5dSChristopher Siden * Do not allocate new object types here. Doing so makes the on-disk 191ad135b5dSChristopher Siden * format incompatible with any other format that uses the same object 192ad135b5dSChristopher Siden * type number. 193ad135b5dSChristopher Siden * 194ad135b5dSChristopher Siden * When creating an object which does not have one of the above types 195ad135b5dSChristopher Siden * use the DMU_OTN_* type with the correct byteswap and metadata 196ad135b5dSChristopher Siden * values. 197ad135b5dSChristopher Siden * 198ad135b5dSChristopher Siden * The DMU_OTN_* types do not have entries in the dmu_ot table, 199ad135b5dSChristopher Siden * use the DMU_OT_IS_METDATA() and DMU_OT_BYTESWAP() macros instead 200ad135b5dSChristopher Siden * of indexing into dmu_ot directly (this works for both DMU_OT_* types 201ad135b5dSChristopher Siden * and DMU_OTN_* types). 202ad135b5dSChristopher Siden */ 203ad135b5dSChristopher Siden DMU_OT_NUMTYPES, 204ad135b5dSChristopher Siden 205ad135b5dSChristopher Siden /* 206ad135b5dSChristopher Siden * Names for valid types declared with DMU_OT(). 207ad135b5dSChristopher Siden */ 208ad135b5dSChristopher Siden DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE), 209ad135b5dSChristopher Siden DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE), 210ad135b5dSChristopher Siden DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE), 211ad135b5dSChristopher Siden DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE), 212ad135b5dSChristopher Siden DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE), 213ad135b5dSChristopher Siden DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE), 214ad135b5dSChristopher Siden DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE), 215ad135b5dSChristopher Siden DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE), 216ad135b5dSChristopher Siden DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE), 217ad135b5dSChristopher Siden DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE), 218fa9e4066Sahrens } dmu_object_type_t; 219fa9e4066Sahrens 2203b2aab18SMatthew Ahrens typedef enum txg_how { 2213b2aab18SMatthew Ahrens TXG_WAIT = 1, 2223b2aab18SMatthew Ahrens TXG_NOWAIT, 223*69962b56SMatthew Ahrens TXG_WAITED, 2243b2aab18SMatthew Ahrens } txg_how_t; 2253b2aab18SMatthew Ahrens 226fa9e4066Sahrens void byteswap_uint64_array(void *buf, size_t size); 227fa9e4066Sahrens void byteswap_uint32_array(void *buf, size_t size); 228fa9e4066Sahrens void byteswap_uint16_array(void *buf, size_t size); 229fa9e4066Sahrens void byteswap_uint8_array(void *buf, size_t size); 230fa9e4066Sahrens void zap_byteswap(void *buf, size_t size); 231da6c28aaSamw void zfs_oldacl_byteswap(void *buf, size_t size); 232fa9e4066Sahrens void zfs_acl_byteswap(void *buf, size_t size); 233fa9e4066Sahrens void zfs_znode_byteswap(void *buf, size_t size); 234fa9e4066Sahrens 2350b69c2f0Sahrens #define DS_FIND_SNAPSHOTS (1<<0) 2360b69c2f0Sahrens #define DS_FIND_CHILDREN (1<<1) 237fa9e4066Sahrens 238fa9e4066Sahrens /* 239fa9e4066Sahrens * The maximum number of bytes that can be accessed as part of one 240fa9e4066Sahrens * operation, including metadata. 241fa9e4066Sahrens */ 242fa9e4066Sahrens #define DMU_MAX_ACCESS (10<<20) /* 10MB */ 243cdb0ab79Smaybee #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ 244fa9e4066Sahrens 24514843421SMatthew Ahrens #define DMU_USERUSED_OBJECT (-1ULL) 24614843421SMatthew Ahrens #define DMU_GROUPUSED_OBJECT (-2ULL) 247b24ab676SJeff Bonwick #define DMU_DEADLIST_OBJECT (-3ULL) 24814843421SMatthew Ahrens 2490a586ceaSMark Shellenbaum /* 2500a586ceaSMark Shellenbaum * artificial blkids for bonus buffer and spill blocks 2510a586ceaSMark Shellenbaum */ 2520a586ceaSMark Shellenbaum #define DMU_BONUS_BLKID (-1ULL) 2530a586ceaSMark Shellenbaum #define DMU_SPILL_BLKID (-2ULL) 254fa9e4066Sahrens /* 255fa9e4066Sahrens * Public routines to create, destroy, open, and close objsets. 256fa9e4066Sahrens */ 257503ad85cSMatthew Ahrens int dmu_objset_hold(const char *name, void *tag, objset_t **osp); 258503ad85cSMatthew Ahrens int dmu_objset_own(const char *name, dmu_objset_type_t type, 259503ad85cSMatthew Ahrens boolean_t readonly, void *tag, objset_t **osp); 260503ad85cSMatthew Ahrens void dmu_objset_rele(objset_t *os, void *tag); 261503ad85cSMatthew Ahrens void dmu_objset_disown(objset_t *os, void *tag); 262503ad85cSMatthew Ahrens int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); 263503ad85cSMatthew Ahrens 2643b2aab18SMatthew Ahrens void dmu_objset_evict_dbufs(objset_t *os); 265ae46e4c7SMatthew Ahrens int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, 266ecd6cf80Smarks void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); 2673b2aab18SMatthew Ahrens int dmu_objset_clone(const char *name, const char *origin); 2683b2aab18SMatthew Ahrens int dsl_destroy_snapshots_nvl(struct nvlist *snaps, boolean_t defer, 2694445fffbSMatthew Ahrens struct nvlist *errlist); 2704445fffbSMatthew Ahrens int dmu_objset_snapshot_one(const char *fsname, const char *snapname); 2714445fffbSMatthew Ahrens int dmu_objset_snapshot_tmp(const char *, const char *, int); 272fd136879SMatthew Ahrens int dmu_objset_find(char *name, int func(const char *, void *), void *arg, 273fa9e4066Sahrens int flags); 274fa9e4066Sahrens void dmu_objset_byteswap(void *buf, size_t size); 2753b2aab18SMatthew Ahrens int dsl_dataset_rename_snapshot(const char *fsname, 2763b2aab18SMatthew Ahrens const char *oldsnapname, const char *newsnapname, boolean_t recursive); 277fa9e4066Sahrens 278fa9e4066Sahrens typedef struct dmu_buf { 279fa9e4066Sahrens uint64_t db_object; /* object that this buffer is part of */ 280fa9e4066Sahrens uint64_t db_offset; /* byte offset in this object */ 281fa9e4066Sahrens uint64_t db_size; /* size of buffer in bytes */ 282fa9e4066Sahrens void *db_data; /* data in buffer */ 283fa9e4066Sahrens } dmu_buf_t; 284fa9e4066Sahrens 285fa9e4066Sahrens typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); 286fa9e4066Sahrens 28799653d4eSeschrock /* 28899653d4eSeschrock * The names of zap entries in the DIRECTORY_OBJECT of the MOS. 28999653d4eSeschrock */ 290fa9e4066Sahrens #define DMU_POOL_DIRECTORY_OBJECT 1 291fa9e4066Sahrens #define DMU_POOL_CONFIG "config" 292ad135b5dSChristopher Siden #define DMU_POOL_FEATURES_FOR_WRITE "features_for_write" 293ad135b5dSChristopher Siden #define DMU_POOL_FEATURES_FOR_READ "features_for_read" 294ad135b5dSChristopher Siden #define DMU_POOL_FEATURE_DESCRIPTIONS "feature_descriptions" 295fa9e4066Sahrens #define DMU_POOL_ROOT_DATASET "root_dataset" 296cde58dbcSMatthew Ahrens #define DMU_POOL_SYNC_BPOBJ "sync_bplist" 297ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" 298ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_LAST "errlog_last" 29999653d4eSeschrock #define DMU_POOL_SPARES "spares" 30099653d4eSeschrock #define DMU_POOL_DEFLATE "deflate" 30106eeb2adSek #define DMU_POOL_HISTORY "history" 302b1b8ab34Slling #define DMU_POOL_PROPS "pool_props" 303fa94a07fSbrendan #define DMU_POOL_L2CACHE "l2cache" 304ca45db41SChris Kirby #define DMU_POOL_TMP_USERREFS "tmp_userrefs" 305b24ab676SJeff Bonwick #define DMU_POOL_DDT "DDT-%s-%s-%s" 306b24ab676SJeff Bonwick #define DMU_POOL_DDT_STATS "DDT-statistics" 3073f9d6ad7SLin Ling #define DMU_POOL_CREATION_VERSION "creation_version" 3083f9d6ad7SLin Ling #define DMU_POOL_SCAN "scan" 309cde58dbcSMatthew Ahrens #define DMU_POOL_FREE_BPOBJ "free_bpobj" 310ad135b5dSChristopher Siden #define DMU_POOL_BPTREE_OBJ "bptree_obj" 311f1745736SMatthew Ahrens #define DMU_POOL_EMPTY_BPOBJ "empty_bpobj" 312088f3894Sahrens 313fa9e4066Sahrens /* 314fa9e4066Sahrens * Allocate an object from this objset. The range of object numbers 315fa9e4066Sahrens * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. 316fa9e4066Sahrens * 317fa9e4066Sahrens * The transaction must be assigned to a txg. The newly allocated 318fa9e4066Sahrens * object will be "held" in the transaction (ie. you can modify the 319fa9e4066Sahrens * newly allocated object in this transaction). 320fa9e4066Sahrens * 321fa9e4066Sahrens * dmu_object_alloc() chooses an object and returns it in *objectp. 322fa9e4066Sahrens * 323fa9e4066Sahrens * dmu_object_claim() allocates a specific object number. If that 324fa9e4066Sahrens * number is already allocated, it fails and returns EEXIST. 325fa9e4066Sahrens * 326fa9e4066Sahrens * Return 0 on success, or ENOSPC or EEXIST as specified above. 327fa9e4066Sahrens */ 328fa9e4066Sahrens uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, 329fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 330fa9e4066Sahrens int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, 331fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 332fa9e4066Sahrens int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, 3332bf405a2SMark Maybee int blocksize, dmu_object_type_t bonustype, int bonuslen); 334fa9e4066Sahrens 335fa9e4066Sahrens /* 336fa9e4066Sahrens * Free an object from this objset. 337fa9e4066Sahrens * 338fa9e4066Sahrens * The object's data will be freed as well (ie. you don't need to call 339fa9e4066Sahrens * dmu_free(object, 0, -1, tx)). 340fa9e4066Sahrens * 341fa9e4066Sahrens * The object need not be held in the transaction. 342fa9e4066Sahrens * 343fa9e4066Sahrens * If there are any holds on this object's buffers (via dmu_buf_hold()), 344fa9e4066Sahrens * or tx holds on the object (via dmu_tx_hold_object()), you can not 345fa9e4066Sahrens * free it; it fails and returns EBUSY. 346fa9e4066Sahrens * 347fa9e4066Sahrens * If the object is not allocated, it fails and returns ENOENT. 348fa9e4066Sahrens * 349fa9e4066Sahrens * Return 0 on success, or EBUSY or ENOENT as specified above. 350fa9e4066Sahrens */ 351fa9e4066Sahrens int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); 352fa9e4066Sahrens 353fa9e4066Sahrens /* 354fa9e4066Sahrens * Find the next allocated or free object. 355fa9e4066Sahrens * 356fa9e4066Sahrens * The objectp parameter is in-out. It will be updated to be the next 3576754306eSahrens * object which is allocated. Ignore objects which have not been 3586754306eSahrens * modified since txg. 359fa9e4066Sahrens * 360fa9e4066Sahrens * XXX Can only be called on a objset with no dirty data. 361fa9e4066Sahrens * 362fa9e4066Sahrens * Returns 0 on success, or ENOENT if there are no more objects. 363fa9e4066Sahrens */ 3646754306eSahrens int dmu_object_next(objset_t *os, uint64_t *objectp, 3656754306eSahrens boolean_t hole, uint64_t txg); 366fa9e4066Sahrens 367fa9e4066Sahrens /* 368fa9e4066Sahrens * Set the data blocksize for an object. 369fa9e4066Sahrens * 370fa9e4066Sahrens * The object cannot have any blocks allcated beyond the first. If 371fa9e4066Sahrens * the first block is allocated already, the new size must be greater 372fa9e4066Sahrens * than the current block size. If these conditions are not met, 373fa9e4066Sahrens * ENOTSUP will be returned. 374fa9e4066Sahrens * 375fa9e4066Sahrens * Returns 0 on success, or EBUSY if there are any holds on the object 376fa9e4066Sahrens * contents, or ENOTSUP as described above. 377fa9e4066Sahrens */ 378fa9e4066Sahrens int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, 379fa9e4066Sahrens int ibs, dmu_tx_t *tx); 380fa9e4066Sahrens 381fa9e4066Sahrens /* 382fa9e4066Sahrens * Set the checksum property on a dnode. The new checksum algorithm will 383fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 384fa9e4066Sahrens */ 385fa9e4066Sahrens void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, 386fa9e4066Sahrens dmu_tx_t *tx); 387fa9e4066Sahrens 388fa9e4066Sahrens /* 389fa9e4066Sahrens * Set the compress property on a dnode. The new compression algorithm will 390fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 391fa9e4066Sahrens */ 392fa9e4066Sahrens void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, 393fa9e4066Sahrens dmu_tx_t *tx); 394fa9e4066Sahrens 39544cd46caSbillm /* 396b24ab676SJeff Bonwick * Decide how to write a block: checksum, compression, number of copies, etc. 39744cd46caSbillm */ 398b24ab676SJeff Bonwick #define WP_NOFILL 0x1 399b24ab676SJeff Bonwick #define WP_DMU_SYNC 0x2 4000a586ceaSMark Shellenbaum #define WP_SPILL 0x4 401b24ab676SJeff Bonwick 402b24ab676SJeff Bonwick void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, 403b24ab676SJeff Bonwick struct zio_prop *zp); 404fa9e4066Sahrens /* 405fa9e4066Sahrens * The bonus data is accessed more or less like a regular buffer. 406fa9e4066Sahrens * You must dmu_bonus_hold() to get the buffer, which will give you a 407fa9e4066Sahrens * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus 408fa9e4066Sahrens * data. As with any normal buffer, you must call dmu_buf_read() to 409fa9e4066Sahrens * read db_data, dmu_buf_will_dirty() before modifying it, and the 410fa9e4066Sahrens * object must be held in an assigned transaction before calling 411fa9e4066Sahrens * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus 412fa9e4066Sahrens * buffer as well. You must release your hold with dmu_buf_rele(). 4133e30c24aSWill Andrews * 4143e30c24aSWill Andrews * Returns ENOENT, EIO, or 0. 415fa9e4066Sahrens */ 416ea8dc4b6Seschrock int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 417fa9e4066Sahrens int dmu_bonus_max(void); 4181934e92fSmaybee int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); 4190a586ceaSMark Shellenbaum int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); 420744947dcSTom Erickson dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); 4210a586ceaSMark Shellenbaum int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); 4220a586ceaSMark Shellenbaum 4230a586ceaSMark Shellenbaum /* 4240a586ceaSMark Shellenbaum * Special spill buffer support used by "SA" framework 4250a586ceaSMark Shellenbaum */ 4260a586ceaSMark Shellenbaum 4270a586ceaSMark Shellenbaum int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); 4280a586ceaSMark Shellenbaum int dmu_spill_hold_by_dnode(struct dnode *dn, uint32_t flags, 4290a586ceaSMark Shellenbaum void *tag, dmu_buf_t **dbp); 4300a586ceaSMark Shellenbaum int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); 431fa9e4066Sahrens 432fa9e4066Sahrens /* 433fa9e4066Sahrens * Obtain the DMU buffer from the specified object which contains the 434fa9e4066Sahrens * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so 435fa9e4066Sahrens * that it will remain in memory. You must release the hold with 436fa9e4066Sahrens * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your 437fa9e4066Sahrens * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. 438fa9e4066Sahrens * 439fa9e4066Sahrens * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill 440fa9e4066Sahrens * on the returned buffer before reading or writing the buffer's 441fa9e4066Sahrens * db_data. The comments for those routines describe what particular 442fa9e4066Sahrens * operations are valid after calling them. 443fa9e4066Sahrens * 444fa9e4066Sahrens * The object number must be a valid, allocated object number. 445fa9e4066Sahrens */ 446ea8dc4b6Seschrock int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, 44747cb52daSJeff Bonwick void *tag, dmu_buf_t **, int flags); 448fa9e4066Sahrens void dmu_buf_add_ref(dmu_buf_t *db, void* tag); 449ea8dc4b6Seschrock void dmu_buf_rele(dmu_buf_t *db, void *tag); 450fa9e4066Sahrens uint64_t dmu_buf_refcount(dmu_buf_t *db); 451fa9e4066Sahrens 452fa9e4066Sahrens /* 453fa9e4066Sahrens * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a 454fa9e4066Sahrens * range of an object. A pointer to an array of dmu_buf_t*'s is 455fa9e4066Sahrens * returned (in *dbpp). 456fa9e4066Sahrens * 457fa9e4066Sahrens * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and 458fa9e4066Sahrens * frees the array. The hold on the array of buffers MUST be released 459fa9e4066Sahrens * with dmu_buf_rele_array. You can NOT release the hold on each buffer 460fa9e4066Sahrens * individually with dmu_buf_rele. 461fa9e4066Sahrens */ 46213506d1eSmaybee int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, 46313506d1eSmaybee uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 464ea8dc4b6Seschrock void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); 465fa9e4066Sahrens 466fa9e4066Sahrens /* 467fa9e4066Sahrens * Returns NULL on success, or the existing user ptr if it's already 468fa9e4066Sahrens * been set. 469fa9e4066Sahrens * 470fa9e4066Sahrens * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). 471fa9e4066Sahrens * 472fa9e4066Sahrens * user_data_ptr_ptr should be NULL, or a pointer to a pointer which 473fa9e4066Sahrens * will be set to db->db_data when you are allowed to access it. Note 474fa9e4066Sahrens * that db->db_data (the pointer) can change when you do dmu_buf_read(), 475fa9e4066Sahrens * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). 476fa9e4066Sahrens * *user_data_ptr_ptr will be set to the new value when it changes. 477fa9e4066Sahrens * 478fa9e4066Sahrens * If non-NULL, pageout func will be called when this buffer is being 479fa9e4066Sahrens * excised from the cache, so that you can clean up the data structure 480fa9e4066Sahrens * pointed to by user_ptr. 481fa9e4066Sahrens * 482fa9e4066Sahrens * dmu_evict_user() will call the pageout func for all buffers in a 483fa9e4066Sahrens * objset with a given pageout func. 484fa9e4066Sahrens */ 485fa9e4066Sahrens void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, 486fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 487fa9e4066Sahrens /* 488fa9e4066Sahrens * set_user_ie is the same as set_user, but request immediate eviction 489fa9e4066Sahrens * when hold count goes to zero. 490fa9e4066Sahrens */ 491fa9e4066Sahrens void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, 492fa9e4066Sahrens void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); 493fa9e4066Sahrens void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, 494fa9e4066Sahrens void *user_ptr, void *user_data_ptr_ptr, 495fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 496fa9e4066Sahrens void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); 497fa9e4066Sahrens 498fa9e4066Sahrens /* 499fa9e4066Sahrens * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. 500fa9e4066Sahrens */ 501fa9e4066Sahrens void *dmu_buf_get_user(dmu_buf_t *db); 502fa9e4066Sahrens 50380901aeaSGeorge Wilson /* 50480901aeaSGeorge Wilson * Returns the blkptr associated with this dbuf, or NULL if not set. 50580901aeaSGeorge Wilson */ 50680901aeaSGeorge Wilson struct blkptr *dmu_buf_get_blkptr(dmu_buf_t *db); 50780901aeaSGeorge Wilson 508fa9e4066Sahrens /* 509fa9e4066Sahrens * Indicate that you are going to modify the buffer's data (db_data). 510fa9e4066Sahrens * 511fa9e4066Sahrens * The transaction (tx) must be assigned to a txg (ie. you've called 512fa9e4066Sahrens * dmu_tx_assign()). The buffer's object must be held in the tx 513fa9e4066Sahrens * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). 514fa9e4066Sahrens */ 515fa9e4066Sahrens void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); 516fa9e4066Sahrens 5173d692628SSanjeev Bagewadi /* 5183d692628SSanjeev Bagewadi * Tells if the given dbuf is freeable. 5193d692628SSanjeev Bagewadi */ 5203d692628SSanjeev Bagewadi boolean_t dmu_buf_freeable(dmu_buf_t *); 5213d692628SSanjeev Bagewadi 522fa9e4066Sahrens /* 523fa9e4066Sahrens * You must create a transaction, then hold the objects which you will 524fa9e4066Sahrens * (or might) modify as part of this transaction. Then you must assign 525fa9e4066Sahrens * the transaction to a transaction group. Once the transaction has 526fa9e4066Sahrens * been assigned, you can modify buffers which belong to held objects as 527fa9e4066Sahrens * part of this transaction. You can't modify buffers before the 528fa9e4066Sahrens * transaction has been assigned; you can't modify buffers which don't 529fa9e4066Sahrens * belong to objects which this transaction holds; you can't hold 530fa9e4066Sahrens * objects once the transaction has been assigned. You may hold an 531fa9e4066Sahrens * object which you are going to free (with dmu_object_free()), but you 532fa9e4066Sahrens * don't have to. 533fa9e4066Sahrens * 534fa9e4066Sahrens * You can abort the transaction before it has been assigned. 535fa9e4066Sahrens * 536fa9e4066Sahrens * Note that you may hold buffers (with dmu_buf_hold) at any time, 537fa9e4066Sahrens * regardless of transaction state. 538fa9e4066Sahrens */ 539fa9e4066Sahrens 540fa9e4066Sahrens #define DMU_NEW_OBJECT (-1ULL) 541fa9e4066Sahrens #define DMU_OBJECT_END (-1ULL) 542fa9e4066Sahrens 543fa9e4066Sahrens dmu_tx_t *dmu_tx_create(objset_t *os); 544fa9e4066Sahrens void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); 545fa9e4066Sahrens void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, 546fa9e4066Sahrens uint64_t len); 54714843421SMatthew Ahrens void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); 548fa9e4066Sahrens void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); 5490a586ceaSMark Shellenbaum void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); 5500a586ceaSMark Shellenbaum void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); 5510a586ceaSMark Shellenbaum void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); 552fa9e4066Sahrens void dmu_tx_abort(dmu_tx_t *tx); 5533b2aab18SMatthew Ahrens int dmu_tx_assign(dmu_tx_t *tx, enum txg_how txg_how); 5548a2f1b91Sahrens void dmu_tx_wait(dmu_tx_t *tx); 555fa9e4066Sahrens void dmu_tx_commit(dmu_tx_t *tx); 556fa9e4066Sahrens 557d20e665cSRicardo M. Correia /* 558d20e665cSRicardo M. Correia * To register a commit callback, dmu_tx_callback_register() must be called. 559d20e665cSRicardo M. Correia * 560d20e665cSRicardo M. Correia * dcb_data is a pointer to caller private data that is passed on as a 561d20e665cSRicardo M. Correia * callback parameter. The caller is responsible for properly allocating and 562d20e665cSRicardo M. Correia * freeing it. 563d20e665cSRicardo M. Correia * 564d20e665cSRicardo M. Correia * When registering a callback, the transaction must be already created, but 565d20e665cSRicardo M. Correia * it cannot be committed or aborted. It can be assigned to a txg or not. 566d20e665cSRicardo M. Correia * 567d20e665cSRicardo M. Correia * The callback will be called after the transaction has been safely written 568d20e665cSRicardo M. Correia * to stable storage and will also be called if the dmu_tx is aborted. 569d20e665cSRicardo M. Correia * If there is any error which prevents the transaction from being committed to 570d20e665cSRicardo M. Correia * disk, the callback will be called with a value of error != 0. 571d20e665cSRicardo M. Correia */ 572d20e665cSRicardo M. Correia typedef void dmu_tx_callback_func_t(void *dcb_data, int error); 573d20e665cSRicardo M. Correia 574d20e665cSRicardo M. Correia void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, 575d20e665cSRicardo M. Correia void *dcb_data); 576d20e665cSRicardo M. Correia 577fa9e4066Sahrens /* 578fa9e4066Sahrens * Free up the data blocks for a defined range of a file. If size is 579ad135b5dSChristopher Siden * -1, the range from offset to end-of-file is freed. 580fa9e4066Sahrens */ 581ea8dc4b6Seschrock int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, 582fa9e4066Sahrens uint64_t size, dmu_tx_t *tx); 583cdb0ab79Smaybee int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, 584cdb0ab79Smaybee uint64_t size); 585713d6c20SMatthew Ahrens int dmu_free_long_object(objset_t *os, uint64_t object); 586fa9e4066Sahrens 587fa9e4066Sahrens /* 588fa9e4066Sahrens * Convenience functions. 589fa9e4066Sahrens * 590fa9e4066Sahrens * Canfail routines will return 0 on success, or an errno if there is a 591fa9e4066Sahrens * nonrecoverable I/O error. 592fa9e4066Sahrens */ 5937bfdf011SNeil Perrin #define DMU_READ_PREFETCH 0 /* prefetch */ 5947bfdf011SNeil Perrin #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ 595ea8dc4b6Seschrock int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 5967bfdf011SNeil Perrin void *buf, uint32_t flags); 597fa9e4066Sahrens void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 598fa9e4066Sahrens const void *buf, dmu_tx_t *tx); 59982c9918fSTim Haley void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 60082c9918fSTim Haley dmu_tx_t *tx); 601feb08c6bSbillm int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); 602feb08c6bSbillm int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, 603feb08c6bSbillm dmu_tx_t *tx); 60494d1a210STim Haley int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, 60594d1a210STim Haley dmu_tx_t *tx); 60644eda4d7Smaybee int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, 60744eda4d7Smaybee uint64_t size, struct page *pp, dmu_tx_t *tx); 6082fdbea25SAleksandr Guzovskiy struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); 6092fdbea25SAleksandr Guzovskiy void dmu_return_arcbuf(struct arc_buf *buf); 6102fdbea25SAleksandr Guzovskiy void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, 6112fdbea25SAleksandr Guzovskiy dmu_tx_t *tx); 612c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_init(struct xuio *uio, int niov); 613c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void dmu_xuio_fini(struct xuio *uio); 614c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, 615c242f9a0Schunli zhang - Sun Microsystems - Irvine United States size_t n); 616c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_cnt(struct xuio *uio); 617c242f9a0Schunli zhang - Sun Microsystems - Irvine United States struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); 618c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void dmu_xuio_clear(struct xuio *uio, int i); 619c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void xuio_stat_wbuf_copied(); 620c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void xuio_stat_wbuf_nocopy(); 621fa9e4066Sahrens 622416e0cd8Sek extern int zfs_prefetch_disable; 623416e0cd8Sek 624fa9e4066Sahrens /* 625fa9e4066Sahrens * Asynchronously try to read in the data. 626fa9e4066Sahrens */ 627fa9e4066Sahrens void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, 628fa9e4066Sahrens uint64_t len); 629fa9e4066Sahrens 630fa9e4066Sahrens typedef struct dmu_object_info { 631b24ab676SJeff Bonwick /* All sizes are in bytes unless otherwise indicated. */ 632fa9e4066Sahrens uint32_t doi_data_block_size; 633fa9e4066Sahrens uint32_t doi_metadata_block_size; 634fa9e4066Sahrens dmu_object_type_t doi_type; 635fa9e4066Sahrens dmu_object_type_t doi_bonus_type; 636b24ab676SJeff Bonwick uint64_t doi_bonus_size; 637fa9e4066Sahrens uint8_t doi_indirection; /* 2 = dnode->indirect->data */ 638fa9e4066Sahrens uint8_t doi_checksum; 639fa9e4066Sahrens uint8_t doi_compress; 640fa9e4066Sahrens uint8_t doi_pad[5]; 641b24ab676SJeff Bonwick uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ 642b24ab676SJeff Bonwick uint64_t doi_max_offset; 643b24ab676SJeff Bonwick uint64_t doi_fill_count; /* number of non-empty blocks */ 644fa9e4066Sahrens } dmu_object_info_t; 645fa9e4066Sahrens 646e45ce728Sahrens typedef void arc_byteswap_func_t(void *buf, size_t size); 647e45ce728Sahrens 648fa9e4066Sahrens typedef struct dmu_object_type_info { 649ad135b5dSChristopher Siden dmu_object_byteswap_t ot_byteswap; 650fa9e4066Sahrens boolean_t ot_metadata; 651fa9e4066Sahrens char *ot_name; 652fa9e4066Sahrens } dmu_object_type_info_t; 653fa9e4066Sahrens 654ad135b5dSChristopher Siden typedef struct dmu_object_byteswap_info { 655ad135b5dSChristopher Siden arc_byteswap_func_t *ob_func; 656ad135b5dSChristopher Siden char *ob_name; 657ad135b5dSChristopher Siden } dmu_object_byteswap_info_t; 658ad135b5dSChristopher Siden 659fa9e4066Sahrens extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; 660ad135b5dSChristopher Siden extern const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS]; 661fa9e4066Sahrens 662fa9e4066Sahrens /* 663fa9e4066Sahrens * Get information on a DMU object. 664fa9e4066Sahrens * 665fa9e4066Sahrens * Return 0 on success or ENOENT if object is not allocated. 666fa9e4066Sahrens * 667fa9e4066Sahrens * If doi is NULL, just indicates whether the object exists. 668fa9e4066Sahrens */ 669fa9e4066Sahrens int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); 6703e30c24aSWill Andrews /* Like dmu_object_info, but faster if you have a held dnode in hand. */ 671fa9e4066Sahrens void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); 6723e30c24aSWill Andrews /* Like dmu_object_info, but faster if you have a held dbuf in hand. */ 673fa9e4066Sahrens void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); 6743e30c24aSWill Andrews /* 6753e30c24aSWill Andrews * Like dmu_object_info_from_db, but faster still when you only care about 6763e30c24aSWill Andrews * the size. This is specifically optimized for zfs_getattr(). 6773e30c24aSWill Andrews */ 678fa9e4066Sahrens void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, 679fa9e4066Sahrens u_longlong_t *nblk512); 680fa9e4066Sahrens 681fa9e4066Sahrens typedef struct dmu_objset_stats { 682a2eea2e1Sahrens uint64_t dds_num_clones; /* number of clones of this */ 683a2eea2e1Sahrens uint64_t dds_creation_txg; 6843cb34c60Sahrens uint64_t dds_guid; 685fa9e4066Sahrens dmu_objset_type_t dds_type; 686fa9e4066Sahrens uint8_t dds_is_snapshot; 68731fd60d3Sahrens uint8_t dds_inconsistent; 6883cb34c60Sahrens char dds_origin[MAXNAMELEN]; 689a2eea2e1Sahrens } dmu_objset_stats_t; 690fa9e4066Sahrens 691a2eea2e1Sahrens /* 692a2eea2e1Sahrens * Get stats on a dataset. 693a2eea2e1Sahrens */ 694a2eea2e1Sahrens void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); 695fa9e4066Sahrens 696a2eea2e1Sahrens /* 697a2eea2e1Sahrens * Add entries to the nvlist for all the objset's properties. See 698a2eea2e1Sahrens * zfs_prop_table[] and zfs(1m) for details on the properties. 699a2eea2e1Sahrens */ 700a2eea2e1Sahrens void dmu_objset_stats(objset_t *os, struct nvlist *nv); 701fa9e4066Sahrens 702a2eea2e1Sahrens /* 703a2eea2e1Sahrens * Get the space usage statistics for statvfs(). 704a2eea2e1Sahrens * 705a2eea2e1Sahrens * refdbytes is the amount of space "referenced" by this objset. 706a2eea2e1Sahrens * availbytes is the amount of space available to this objset, taking 707a2eea2e1Sahrens * into account quotas & reservations, assuming that no other objsets 708a2eea2e1Sahrens * use the space first. These values correspond to the 'referenced' and 709a2eea2e1Sahrens * 'available' properties, described in the zfs(1m) manpage. 710a2eea2e1Sahrens * 711a2eea2e1Sahrens * usedobjs and availobjs are the number of objects currently allocated, 712a2eea2e1Sahrens * and available. 713a2eea2e1Sahrens */ 714a2eea2e1Sahrens void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, 715a2eea2e1Sahrens uint64_t *usedobjsp, uint64_t *availobjsp); 716fa9e4066Sahrens 717fa9e4066Sahrens /* 718a2eea2e1Sahrens * The fsid_guid is a 56-bit ID that can change to avoid collisions. 719a2eea2e1Sahrens * (Contrast with the ds_guid which is a 64-bit ID that will never 720a2eea2e1Sahrens * change, so there is a small probability that it will collide.) 721fa9e4066Sahrens */ 722a2eea2e1Sahrens uint64_t dmu_objset_fsid_guid(objset_t *os); 723fa9e4066Sahrens 72471eb0538SChris Kirby /* 72571eb0538SChris Kirby * Get the [cm]time for an objset's snapshot dir 72671eb0538SChris Kirby */ 72771eb0538SChris Kirby timestruc_t dmu_objset_snap_cmtime(objset_t *os); 72871eb0538SChris Kirby 729fa9e4066Sahrens int dmu_objset_is_snapshot(objset_t *os); 730fa9e4066Sahrens 731fa9e4066Sahrens extern struct spa *dmu_objset_spa(objset_t *os); 732fa9e4066Sahrens extern struct zilog *dmu_objset_zil(objset_t *os); 733fa9e4066Sahrens extern struct dsl_pool *dmu_objset_pool(objset_t *os); 734fa9e4066Sahrens extern struct dsl_dataset *dmu_objset_ds(objset_t *os); 735fa9e4066Sahrens extern void dmu_objset_name(objset_t *os, char *buf); 736fa9e4066Sahrens extern dmu_objset_type_t dmu_objset_type(objset_t *os); 737fa9e4066Sahrens extern uint64_t dmu_objset_id(objset_t *os); 73855da60b9SMark J Musante extern uint64_t dmu_objset_syncprop(objset_t *os); 739e09fa4daSNeil Perrin extern uint64_t dmu_objset_logbias(objset_t *os); 740fa9e4066Sahrens extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, 741b38f0970Sck uint64_t *id, uint64_t *offp, boolean_t *case_conflict); 742ab04eb8eStimh extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, 743ab04eb8eStimh int maxlen, boolean_t *conflict); 74487e5029aSahrens extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, 74587e5029aSahrens uint64_t *idp, uint64_t *offp); 74614843421SMatthew Ahrens 7479966ca11SMatthew Ahrens typedef int objset_used_cb_t(dmu_object_type_t bonustype, 7489966ca11SMatthew Ahrens void *bonus, uint64_t *userp, uint64_t *groupp); 74914843421SMatthew Ahrens extern void dmu_objset_register_type(dmu_objset_type_t ost, 75014843421SMatthew Ahrens objset_used_cb_t *cb); 751f18faf3fSek extern void dmu_objset_set_user(objset_t *os, void *user_ptr); 752f18faf3fSek extern void *dmu_objset_get_user(objset_t *os); 753fa9e4066Sahrens 754fa9e4066Sahrens /* 755fa9e4066Sahrens * Return the txg number for the given assigned transaction. 756fa9e4066Sahrens */ 75787e5029aSahrens uint64_t dmu_tx_get_txg(dmu_tx_t *tx); 758fa9e4066Sahrens 759fa9e4066Sahrens /* 760fa9e4066Sahrens * Synchronous write. 761c5c6ffa0Smaybee * If a parent zio is provided this function initiates a write on the 762c5c6ffa0Smaybee * provided buffer as a child of the parent zio. 763da6c28aaSamw * In the absence of a parent zio, the write is completed synchronously. 764c5c6ffa0Smaybee * At write completion, blk is filled with the bp of the written block. 765fa9e4066Sahrens * Note that while the data covered by this function will be on stable 766c5c6ffa0Smaybee * storage when the write completes this new data does not become a 767fa9e4066Sahrens * permanent part of the file until the associated transaction commits. 768fa9e4066Sahrens */ 769b24ab676SJeff Bonwick 770b24ab676SJeff Bonwick /* 771b24ab676SJeff Bonwick * {zfs,zvol,ztest}_get_done() args 772b24ab676SJeff Bonwick */ 773b24ab676SJeff Bonwick typedef struct zgd { 774b24ab676SJeff Bonwick struct zilog *zgd_zilog; 775b24ab676SJeff Bonwick struct blkptr *zgd_bp; 776b24ab676SJeff Bonwick dmu_buf_t *zgd_db; 777b24ab676SJeff Bonwick struct rl *zgd_rl; 778b24ab676SJeff Bonwick void *zgd_private; 779b24ab676SJeff Bonwick } zgd_t; 780b24ab676SJeff Bonwick 781b24ab676SJeff Bonwick typedef void dmu_sync_cb_t(zgd_t *arg, int error); 782b24ab676SJeff Bonwick int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); 783fa9e4066Sahrens 784fa9e4066Sahrens /* 785fa9e4066Sahrens * Find the next hole or data block in file starting at *off 786fa9e4066Sahrens * Return found offset in *off. Return ESRCH for end of file. 787fa9e4066Sahrens */ 788fa9e4066Sahrens int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, 789fa9e4066Sahrens uint64_t *off); 790fa9e4066Sahrens 791fa9e4066Sahrens /* 792fa9e4066Sahrens * Initial setup and final teardown. 793fa9e4066Sahrens */ 794fa9e4066Sahrens extern void dmu_init(void); 795fa9e4066Sahrens extern void dmu_fini(void); 796fa9e4066Sahrens 797fa9e4066Sahrens typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, 798fa9e4066Sahrens uint64_t object, uint64_t offset, int len); 799fa9e4066Sahrens void dmu_traverse_objset(objset_t *os, uint64_t txg_start, 800fa9e4066Sahrens dmu_traverse_cb_t cb, void *arg); 801fa9e4066Sahrens 8023b2aab18SMatthew Ahrens int dmu_diff(const char *tosnap_name, const char *fromsnap_name, 8033b2aab18SMatthew Ahrens struct vnode *vp, offset_t *offp); 80499d5e173STim Haley 805fa9e4066Sahrens /* CRC64 table */ 806fa9e4066Sahrens #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ 807fa9e4066Sahrens extern uint64_t zfs_crc64_table[256]; 808fa9e4066Sahrens 809fa9e4066Sahrens #ifdef __cplusplus 810fa9e4066Sahrens } 811fa9e4066Sahrens #endif 812fa9e4066Sahrens 813fa9e4066Sahrens #endif /* _SYS_DMU_H */ 814