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. 24edf345e6SMatthew Ahrens * Copyright (c) 2011, 2014 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. 27e4be62a2SSaso Kiselkov * Copyright 2013 DEY Storage Systems, Inc. 28ec5cf9d5SAlexander Stetsenko */ 29fa9e4066Sahrens 3055da60b9SMark J Musante /* Portions Copyright 2010 Robert Milkowski */ 3155da60b9SMark J Musante 32fa9e4066Sahrens #ifndef _SYS_DMU_H 33fa9e4066Sahrens #define _SYS_DMU_H 34fa9e4066Sahrens 35fa9e4066Sahrens /* 36fa9e4066Sahrens * This file describes the interface that the DMU provides for its 37fa9e4066Sahrens * consumers. 38fa9e4066Sahrens * 39fa9e4066Sahrens * The DMU also interacts with the SPA. That interface is described in 40fa9e4066Sahrens * dmu_spa.h. 41fa9e4066Sahrens */ 42fa9e4066Sahrens 43fa9e4066Sahrens #include <sys/inttypes.h> 44fa9e4066Sahrens #include <sys/types.h> 45fa9e4066Sahrens #include <sys/param.h> 46ecd6cf80Smarks #include <sys/cred.h> 4771eb0538SChris Kirby #include <sys/time.h> 484445fffbSMatthew Ahrens #include <sys/fs/zfs.h> 49fa9e4066Sahrens 50fa9e4066Sahrens #ifdef __cplusplus 51fa9e4066Sahrens extern "C" { 52fa9e4066Sahrens #endif 53fa9e4066Sahrens 54fa9e4066Sahrens struct uio; 55c242f9a0Schunli zhang - Sun Microsystems - Irvine United States struct xuio; 5644eda4d7Smaybee struct page; 57fa9e4066Sahrens struct vnode; 58fa9e4066Sahrens struct spa; 59fa9e4066Sahrens struct zilog; 60fa9e4066Sahrens struct zio; 61fa9e4066Sahrens struct blkptr; 62fa9e4066Sahrens struct zap_cursor; 63fa9e4066Sahrens struct dsl_dataset; 64fa9e4066Sahrens struct dsl_pool; 65fa9e4066Sahrens struct dnode; 66fa9e4066Sahrens struct drr_begin; 67fa9e4066Sahrens struct drr_end; 68*7802d7bfSMatthew Ahrens struct zbookmark_phys; 6944cd46caSbillm struct spa; 70a2eea2e1Sahrens struct nvlist; 712fdbea25SAleksandr Guzovskiy struct arc_buf; 72b24ab676SJeff Bonwick struct zio_prop; 730a586ceaSMark Shellenbaum struct sa_handle; 74fa9e4066Sahrens 75fa9e4066Sahrens typedef struct objset objset_t; 76fa9e4066Sahrens typedef struct dmu_tx dmu_tx_t; 77fa9e4066Sahrens typedef struct dsl_dir dsl_dir_t; 78fa9e4066Sahrens 79ad135b5dSChristopher Siden typedef enum dmu_object_byteswap { 80ad135b5dSChristopher Siden DMU_BSWAP_UINT8, 81ad135b5dSChristopher Siden DMU_BSWAP_UINT16, 82ad135b5dSChristopher Siden DMU_BSWAP_UINT32, 83ad135b5dSChristopher Siden DMU_BSWAP_UINT64, 84ad135b5dSChristopher Siden DMU_BSWAP_ZAP, 85ad135b5dSChristopher Siden DMU_BSWAP_DNODE, 86ad135b5dSChristopher Siden DMU_BSWAP_OBJSET, 87ad135b5dSChristopher Siden DMU_BSWAP_ZNODE, 88ad135b5dSChristopher Siden DMU_BSWAP_OLDACL, 89ad135b5dSChristopher Siden DMU_BSWAP_ACL, 90ad135b5dSChristopher Siden /* 91ad135b5dSChristopher Siden * Allocating a new byteswap type number makes the on-disk format 92ad135b5dSChristopher Siden * incompatible with any other format that uses the same number. 93ad135b5dSChristopher Siden * 94ad135b5dSChristopher Siden * Data can usually be structured to work with one of the 95ad135b5dSChristopher Siden * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. 96ad135b5dSChristopher Siden */ 97ad135b5dSChristopher Siden DMU_BSWAP_NUMFUNCS 98ad135b5dSChristopher Siden } dmu_object_byteswap_t; 99ad135b5dSChristopher Siden 100ad135b5dSChristopher Siden #define DMU_OT_NEWTYPE 0x80 101ad135b5dSChristopher Siden #define DMU_OT_METADATA 0x40 102ad135b5dSChristopher Siden #define DMU_OT_BYTESWAP_MASK 0x3f 103ad135b5dSChristopher Siden 104ad135b5dSChristopher Siden /* 105ad135b5dSChristopher Siden * Defines a uint8_t object type. Object types specify if the data 106ad135b5dSChristopher Siden * in the object is metadata (boolean) and how to byteswap the data 107ad135b5dSChristopher Siden * (dmu_object_byteswap_t). 108ad135b5dSChristopher Siden */ 109ad135b5dSChristopher Siden #define DMU_OT(byteswap, metadata) \ 110ad135b5dSChristopher Siden (DMU_OT_NEWTYPE | \ 111ad135b5dSChristopher Siden ((metadata) ? DMU_OT_METADATA : 0) | \ 112ad135b5dSChristopher Siden ((byteswap) & DMU_OT_BYTESWAP_MASK)) 113ad135b5dSChristopher Siden 114ad135b5dSChristopher Siden #define DMU_OT_IS_VALID(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 115ad135b5dSChristopher Siden ((ot) & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS : \ 116ad135b5dSChristopher Siden (ot) < DMU_OT_NUMTYPES) 117ad135b5dSChristopher Siden 118ad135b5dSChristopher Siden #define DMU_OT_IS_METADATA(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 119ad135b5dSChristopher Siden ((ot) & DMU_OT_METADATA) : \ 120ad135b5dSChristopher Siden dmu_ot[(ot)].ot_metadata) 121ad135b5dSChristopher Siden 1225d7b4d43SMatthew Ahrens /* 1235d7b4d43SMatthew Ahrens * These object types use bp_fill != 1 for their L0 bp's. Therefore they can't 1245d7b4d43SMatthew Ahrens * have their data embedded (i.e. use a BP_IS_EMBEDDED() bp), because bp_fill 1255d7b4d43SMatthew Ahrens * is repurposed for embedded BPs. 1265d7b4d43SMatthew Ahrens */ 1275d7b4d43SMatthew Ahrens #define DMU_OT_HAS_FILL(ot) \ 1285d7b4d43SMatthew Ahrens ((ot) == DMU_OT_DNODE || (ot) == DMU_OT_OBJSET) 1295d7b4d43SMatthew Ahrens 130ad135b5dSChristopher Siden #define DMU_OT_BYTESWAP(ot) (((ot) & DMU_OT_NEWTYPE) ? \ 131ad135b5dSChristopher Siden ((ot) & DMU_OT_BYTESWAP_MASK) : \ 132ad135b5dSChristopher Siden dmu_ot[(ot)].ot_byteswap) 133ad135b5dSChristopher Siden 134fa9e4066Sahrens typedef enum dmu_object_type { 135fa9e4066Sahrens DMU_OT_NONE, 136fa9e4066Sahrens /* general: */ 137fa9e4066Sahrens DMU_OT_OBJECT_DIRECTORY, /* ZAP */ 138fa9e4066Sahrens DMU_OT_OBJECT_ARRAY, /* UINT64 */ 139fa9e4066Sahrens DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ 140fa9e4066Sahrens DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ 141cde58dbcSMatthew Ahrens DMU_OT_BPOBJ, /* UINT64 */ 142cde58dbcSMatthew Ahrens DMU_OT_BPOBJ_HDR, /* UINT64 */ 143fa9e4066Sahrens /* spa: */ 144fa9e4066Sahrens DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ 145fa9e4066Sahrens DMU_OT_SPACE_MAP, /* UINT64 */ 146fa9e4066Sahrens /* zil: */ 147fa9e4066Sahrens DMU_OT_INTENT_LOG, /* UINT64 */ 148fa9e4066Sahrens /* dmu: */ 149fa9e4066Sahrens DMU_OT_DNODE, /* DNODE */ 150fa9e4066Sahrens DMU_OT_OBJSET, /* OBJSET */ 151fa9e4066Sahrens /* dsl: */ 1521649cd4bStabriz DMU_OT_DSL_DIR, /* UINT64 */ 15387e5029aSahrens DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ 15487e5029aSahrens DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ 155fa9e4066Sahrens DMU_OT_DSL_PROPS, /* ZAP */ 1561649cd4bStabriz DMU_OT_DSL_DATASET, /* UINT64 */ 157fa9e4066Sahrens /* zpl: */ 158fa9e4066Sahrens DMU_OT_ZNODE, /* ZNODE */ 159da6c28aaSamw DMU_OT_OLDACL, /* Old ACL */ 160fa9e4066Sahrens DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ 161fa9e4066Sahrens DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ 162fa9e4066Sahrens DMU_OT_MASTER_NODE, /* ZAP */ 163893a6d32Sahrens DMU_OT_UNLINKED_SET, /* ZAP */ 164fa9e4066Sahrens /* zvol: */ 165fa9e4066Sahrens DMU_OT_ZVOL, /* UINT8 */ 166fa9e4066Sahrens DMU_OT_ZVOL_PROP, /* ZAP */ 167fa9e4066Sahrens /* other; for testing only! */ 168fa9e4066Sahrens DMU_OT_PLAIN_OTHER, /* UINT8 */ 169fa9e4066Sahrens DMU_OT_UINT64_OTHER, /* UINT64 */ 170fa9e4066Sahrens DMU_OT_ZAP_OTHER, /* ZAP */ 171ea8dc4b6Seschrock /* new object types: */ 172ea8dc4b6Seschrock DMU_OT_ERROR_LOG, /* ZAP */ 17306eeb2adSek DMU_OT_SPA_HISTORY, /* UINT8 */ 17406eeb2adSek DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ 175b1b8ab34Slling DMU_OT_POOL_PROPS, /* ZAP */ 176ecd6cf80Smarks DMU_OT_DSL_PERMS, /* ZAP */ 177da6c28aaSamw DMU_OT_ACL, /* ACL */ 178da6c28aaSamw DMU_OT_SYSACL, /* SYSACL */ 179da6c28aaSamw DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ 180da6c28aaSamw DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ 181088f3894Sahrens DMU_OT_NEXT_CLONES, /* ZAP */ 1823f9d6ad7SLin Ling DMU_OT_SCAN_QUEUE, /* ZAP */ 18314843421SMatthew Ahrens DMU_OT_USERGROUP_USED, /* ZAP */ 18414843421SMatthew Ahrens DMU_OT_USERGROUP_QUOTA, /* ZAP */ 185842727c2SChris Kirby DMU_OT_USERREFS, /* ZAP */ 186b24ab676SJeff Bonwick DMU_OT_DDT_ZAP, /* ZAP */ 187b24ab676SJeff Bonwick DMU_OT_DDT_STATS, /* ZAP */ 1880a586ceaSMark Shellenbaum DMU_OT_SA, /* System attr */ 1890a586ceaSMark Shellenbaum DMU_OT_SA_MASTER_NODE, /* ZAP */ 1900a586ceaSMark Shellenbaum DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ 1910a586ceaSMark Shellenbaum DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ 1923f9d6ad7SLin Ling DMU_OT_SCAN_XLATE, /* ZAP */ 1933f9d6ad7SLin Ling DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ 194cde58dbcSMatthew Ahrens DMU_OT_DEADLIST, /* ZAP */ 195cde58dbcSMatthew Ahrens DMU_OT_DEADLIST_HDR, /* UINT64 */ 196cde58dbcSMatthew Ahrens DMU_OT_DSL_CLONES, /* ZAP */ 197cde58dbcSMatthew Ahrens DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ 198ad135b5dSChristopher Siden /* 199ad135b5dSChristopher Siden * Do not allocate new object types here. Doing so makes the on-disk 200ad135b5dSChristopher Siden * format incompatible with any other format that uses the same object 201ad135b5dSChristopher Siden * type number. 202ad135b5dSChristopher Siden * 203ad135b5dSChristopher Siden * When creating an object which does not have one of the above types 204ad135b5dSChristopher Siden * use the DMU_OTN_* type with the correct byteswap and metadata 205ad135b5dSChristopher Siden * values. 206ad135b5dSChristopher Siden * 207ad135b5dSChristopher Siden * The DMU_OTN_* types do not have entries in the dmu_ot table, 208ad135b5dSChristopher Siden * use the DMU_OT_IS_METDATA() and DMU_OT_BYTESWAP() macros instead 209ad135b5dSChristopher Siden * of indexing into dmu_ot directly (this works for both DMU_OT_* types 210ad135b5dSChristopher Siden * and DMU_OTN_* types). 211ad135b5dSChristopher Siden */ 212ad135b5dSChristopher Siden DMU_OT_NUMTYPES, 213ad135b5dSChristopher Siden 214ad135b5dSChristopher Siden /* 215ad135b5dSChristopher Siden * Names for valid types declared with DMU_OT(). 216ad135b5dSChristopher Siden */ 217ad135b5dSChristopher Siden DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE), 218ad135b5dSChristopher Siden DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE), 219ad135b5dSChristopher Siden DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE), 220ad135b5dSChristopher Siden DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE), 221ad135b5dSChristopher Siden DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE), 222ad135b5dSChristopher Siden DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE), 223ad135b5dSChristopher Siden DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE), 224ad135b5dSChristopher Siden DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE), 225ad135b5dSChristopher Siden DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE), 226ad135b5dSChristopher Siden DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE), 227fa9e4066Sahrens } dmu_object_type_t; 228fa9e4066Sahrens 2293b2aab18SMatthew Ahrens typedef enum txg_how { 2303b2aab18SMatthew Ahrens TXG_WAIT = 1, 2313b2aab18SMatthew Ahrens TXG_NOWAIT, 23269962b56SMatthew Ahrens TXG_WAITED, 2333b2aab18SMatthew Ahrens } txg_how_t; 2343b2aab18SMatthew Ahrens 235fa9e4066Sahrens void byteswap_uint64_array(void *buf, size_t size); 236fa9e4066Sahrens void byteswap_uint32_array(void *buf, size_t size); 237fa9e4066Sahrens void byteswap_uint16_array(void *buf, size_t size); 238fa9e4066Sahrens void byteswap_uint8_array(void *buf, size_t size); 239fa9e4066Sahrens void zap_byteswap(void *buf, size_t size); 240da6c28aaSamw void zfs_oldacl_byteswap(void *buf, size_t size); 241fa9e4066Sahrens void zfs_acl_byteswap(void *buf, size_t size); 242fa9e4066Sahrens void zfs_znode_byteswap(void *buf, size_t size); 243fa9e4066Sahrens 2440b69c2f0Sahrens #define DS_FIND_SNAPSHOTS (1<<0) 2450b69c2f0Sahrens #define DS_FIND_CHILDREN (1<<1) 246fa9e4066Sahrens 247fa9e4066Sahrens /* 248fa9e4066Sahrens * The maximum number of bytes that can be accessed as part of one 249fa9e4066Sahrens * operation, including metadata. 250fa9e4066Sahrens */ 251fa9e4066Sahrens #define DMU_MAX_ACCESS (10<<20) /* 10MB */ 252cdb0ab79Smaybee #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ 253fa9e4066Sahrens 25414843421SMatthew Ahrens #define DMU_USERUSED_OBJECT (-1ULL) 25514843421SMatthew Ahrens #define DMU_GROUPUSED_OBJECT (-2ULL) 25614843421SMatthew Ahrens 2570a586ceaSMark Shellenbaum /* 2580a586ceaSMark Shellenbaum * artificial blkids for bonus buffer and spill blocks 2590a586ceaSMark Shellenbaum */ 2600a586ceaSMark Shellenbaum #define DMU_BONUS_BLKID (-1ULL) 2610a586ceaSMark Shellenbaum #define DMU_SPILL_BLKID (-2ULL) 262fa9e4066Sahrens /* 263fa9e4066Sahrens * Public routines to create, destroy, open, and close objsets. 264fa9e4066Sahrens */ 265503ad85cSMatthew Ahrens int dmu_objset_hold(const char *name, void *tag, objset_t **osp); 266503ad85cSMatthew Ahrens int dmu_objset_own(const char *name, dmu_objset_type_t type, 267503ad85cSMatthew Ahrens boolean_t readonly, void *tag, objset_t **osp); 268503ad85cSMatthew Ahrens void dmu_objset_rele(objset_t *os, void *tag); 269503ad85cSMatthew Ahrens void dmu_objset_disown(objset_t *os, void *tag); 270503ad85cSMatthew Ahrens int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); 271503ad85cSMatthew Ahrens 2723b2aab18SMatthew Ahrens void dmu_objset_evict_dbufs(objset_t *os); 273ae46e4c7SMatthew Ahrens int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, 274ecd6cf80Smarks void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); 2753b2aab18SMatthew Ahrens int dmu_objset_clone(const char *name, const char *origin); 2763b2aab18SMatthew Ahrens int dsl_destroy_snapshots_nvl(struct nvlist *snaps, boolean_t defer, 2774445fffbSMatthew Ahrens struct nvlist *errlist); 2784445fffbSMatthew Ahrens int dmu_objset_snapshot_one(const char *fsname, const char *snapname); 2794445fffbSMatthew Ahrens int dmu_objset_snapshot_tmp(const char *, const char *, int); 280fd136879SMatthew Ahrens int dmu_objset_find(char *name, int func(const char *, void *), void *arg, 281fa9e4066Sahrens int flags); 282fa9e4066Sahrens void dmu_objset_byteswap(void *buf, size_t size); 2833b2aab18SMatthew Ahrens int dsl_dataset_rename_snapshot(const char *fsname, 2843b2aab18SMatthew Ahrens const char *oldsnapname, const char *newsnapname, boolean_t recursive); 285fa9e4066Sahrens 286fa9e4066Sahrens typedef struct dmu_buf { 287fa9e4066Sahrens uint64_t db_object; /* object that this buffer is part of */ 288fa9e4066Sahrens uint64_t db_offset; /* byte offset in this object */ 289fa9e4066Sahrens uint64_t db_size; /* size of buffer in bytes */ 290fa9e4066Sahrens void *db_data; /* data in buffer */ 291fa9e4066Sahrens } dmu_buf_t; 292fa9e4066Sahrens 293fa9e4066Sahrens typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); 294fa9e4066Sahrens 29599653d4eSeschrock /* 29699653d4eSeschrock * The names of zap entries in the DIRECTORY_OBJECT of the MOS. 29799653d4eSeschrock */ 298fa9e4066Sahrens #define DMU_POOL_DIRECTORY_OBJECT 1 299fa9e4066Sahrens #define DMU_POOL_CONFIG "config" 300ad135b5dSChristopher Siden #define DMU_POOL_FEATURES_FOR_WRITE "features_for_write" 301ad135b5dSChristopher Siden #define DMU_POOL_FEATURES_FOR_READ "features_for_read" 302ad135b5dSChristopher Siden #define DMU_POOL_FEATURE_DESCRIPTIONS "feature_descriptions" 30343466aaeSMax Grossman #define DMU_POOL_FEATURE_ENABLED_TXG "feature_enabled_txg" 304fa9e4066Sahrens #define DMU_POOL_ROOT_DATASET "root_dataset" 305cde58dbcSMatthew Ahrens #define DMU_POOL_SYNC_BPOBJ "sync_bplist" 306ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" 307ea8dc4b6Seschrock #define DMU_POOL_ERRLOG_LAST "errlog_last" 30899653d4eSeschrock #define DMU_POOL_SPARES "spares" 30999653d4eSeschrock #define DMU_POOL_DEFLATE "deflate" 31006eeb2adSek #define DMU_POOL_HISTORY "history" 311b1b8ab34Slling #define DMU_POOL_PROPS "pool_props" 312fa94a07fSbrendan #define DMU_POOL_L2CACHE "l2cache" 313ca45db41SChris Kirby #define DMU_POOL_TMP_USERREFS "tmp_userrefs" 314b24ab676SJeff Bonwick #define DMU_POOL_DDT "DDT-%s-%s-%s" 315b24ab676SJeff Bonwick #define DMU_POOL_DDT_STATS "DDT-statistics" 3163f9d6ad7SLin Ling #define DMU_POOL_CREATION_VERSION "creation_version" 3173f9d6ad7SLin Ling #define DMU_POOL_SCAN "scan" 318cde58dbcSMatthew Ahrens #define DMU_POOL_FREE_BPOBJ "free_bpobj" 319ad135b5dSChristopher Siden #define DMU_POOL_BPTREE_OBJ "bptree_obj" 320f1745736SMatthew Ahrens #define DMU_POOL_EMPTY_BPOBJ "empty_bpobj" 321088f3894Sahrens 322fa9e4066Sahrens /* 323fa9e4066Sahrens * Allocate an object from this objset. The range of object numbers 324fa9e4066Sahrens * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. 325fa9e4066Sahrens * 326fa9e4066Sahrens * The transaction must be assigned to a txg. The newly allocated 327fa9e4066Sahrens * object will be "held" in the transaction (ie. you can modify the 328fa9e4066Sahrens * newly allocated object in this transaction). 329fa9e4066Sahrens * 330fa9e4066Sahrens * dmu_object_alloc() chooses an object and returns it in *objectp. 331fa9e4066Sahrens * 332fa9e4066Sahrens * dmu_object_claim() allocates a specific object number. If that 333fa9e4066Sahrens * number is already allocated, it fails and returns EEXIST. 334fa9e4066Sahrens * 335fa9e4066Sahrens * Return 0 on success, or ENOSPC or EEXIST as specified above. 336fa9e4066Sahrens */ 337fa9e4066Sahrens uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, 338fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 339fa9e4066Sahrens int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, 340fa9e4066Sahrens int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); 341fa9e4066Sahrens int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, 3422bf405a2SMark Maybee int blocksize, dmu_object_type_t bonustype, int bonuslen); 343fa9e4066Sahrens 344fa9e4066Sahrens /* 345fa9e4066Sahrens * Free an object from this objset. 346fa9e4066Sahrens * 347fa9e4066Sahrens * The object's data will be freed as well (ie. you don't need to call 348fa9e4066Sahrens * dmu_free(object, 0, -1, tx)). 349fa9e4066Sahrens * 350fa9e4066Sahrens * The object need not be held in the transaction. 351fa9e4066Sahrens * 352fa9e4066Sahrens * If there are any holds on this object's buffers (via dmu_buf_hold()), 353fa9e4066Sahrens * or tx holds on the object (via dmu_tx_hold_object()), you can not 354fa9e4066Sahrens * free it; it fails and returns EBUSY. 355fa9e4066Sahrens * 356fa9e4066Sahrens * If the object is not allocated, it fails and returns ENOENT. 357fa9e4066Sahrens * 358fa9e4066Sahrens * Return 0 on success, or EBUSY or ENOENT as specified above. 359fa9e4066Sahrens */ 360fa9e4066Sahrens int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); 361fa9e4066Sahrens 362fa9e4066Sahrens /* 363fa9e4066Sahrens * Find the next allocated or free object. 364fa9e4066Sahrens * 365fa9e4066Sahrens * The objectp parameter is in-out. It will be updated to be the next 3666754306eSahrens * object which is allocated. Ignore objects which have not been 3676754306eSahrens * modified since txg. 368fa9e4066Sahrens * 369fa9e4066Sahrens * XXX Can only be called on a objset with no dirty data. 370fa9e4066Sahrens * 371fa9e4066Sahrens * Returns 0 on success, or ENOENT if there are no more objects. 372fa9e4066Sahrens */ 3736754306eSahrens int dmu_object_next(objset_t *os, uint64_t *objectp, 3746754306eSahrens boolean_t hole, uint64_t txg); 375fa9e4066Sahrens 376fa9e4066Sahrens /* 377fa9e4066Sahrens * Set the data blocksize for an object. 378fa9e4066Sahrens * 379fa9e4066Sahrens * The object cannot have any blocks allcated beyond the first. If 380fa9e4066Sahrens * the first block is allocated already, the new size must be greater 381fa9e4066Sahrens * than the current block size. If these conditions are not met, 382fa9e4066Sahrens * ENOTSUP will be returned. 383fa9e4066Sahrens * 384fa9e4066Sahrens * Returns 0 on success, or EBUSY if there are any holds on the object 385fa9e4066Sahrens * contents, or ENOTSUP as described above. 386fa9e4066Sahrens */ 387fa9e4066Sahrens int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, 388fa9e4066Sahrens int ibs, dmu_tx_t *tx); 389fa9e4066Sahrens 390fa9e4066Sahrens /* 391fa9e4066Sahrens * Set the checksum property on a dnode. The new checksum algorithm will 392fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 393fa9e4066Sahrens */ 394fa9e4066Sahrens void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, 395fa9e4066Sahrens dmu_tx_t *tx); 396fa9e4066Sahrens 397fa9e4066Sahrens /* 398fa9e4066Sahrens * Set the compress property on a dnode. The new compression algorithm will 399fa9e4066Sahrens * apply to all newly written blocks; existing blocks will not be affected. 400fa9e4066Sahrens */ 401fa9e4066Sahrens void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, 402fa9e4066Sahrens dmu_tx_t *tx); 403fa9e4066Sahrens 4045d7b4d43SMatthew Ahrens void 4055d7b4d43SMatthew Ahrens dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, 4065d7b4d43SMatthew Ahrens void *data, uint8_t etype, uint8_t comp, int uncompressed_size, 4075d7b4d43SMatthew Ahrens int compressed_size, int byteorder, dmu_tx_t *tx); 4085d7b4d43SMatthew Ahrens 40944cd46caSbillm /* 410b24ab676SJeff Bonwick * Decide how to write a block: checksum, compression, number of copies, etc. 41144cd46caSbillm */ 412b24ab676SJeff Bonwick #define WP_NOFILL 0x1 413b24ab676SJeff Bonwick #define WP_DMU_SYNC 0x2 4140a586ceaSMark Shellenbaum #define WP_SPILL 0x4 415b24ab676SJeff Bonwick 416b24ab676SJeff Bonwick void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, 417b24ab676SJeff Bonwick struct zio_prop *zp); 418fa9e4066Sahrens /* 419fa9e4066Sahrens * The bonus data is accessed more or less like a regular buffer. 420fa9e4066Sahrens * You must dmu_bonus_hold() to get the buffer, which will give you a 421fa9e4066Sahrens * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus 422fa9e4066Sahrens * data. As with any normal buffer, you must call dmu_buf_read() to 423fa9e4066Sahrens * read db_data, dmu_buf_will_dirty() before modifying it, and the 424fa9e4066Sahrens * object must be held in an assigned transaction before calling 425fa9e4066Sahrens * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus 426fa9e4066Sahrens * buffer as well. You must release your hold with dmu_buf_rele(). 4273e30c24aSWill Andrews * 4283e30c24aSWill Andrews * Returns ENOENT, EIO, or 0. 429fa9e4066Sahrens */ 430ea8dc4b6Seschrock int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 431fa9e4066Sahrens int dmu_bonus_max(void); 4321934e92fSmaybee int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); 4330a586ceaSMark Shellenbaum int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); 434744947dcSTom Erickson dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); 4350a586ceaSMark Shellenbaum int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); 4360a586ceaSMark Shellenbaum 4370a586ceaSMark Shellenbaum /* 4380a586ceaSMark Shellenbaum * Special spill buffer support used by "SA" framework 4390a586ceaSMark Shellenbaum */ 4400a586ceaSMark Shellenbaum 4410a586ceaSMark Shellenbaum int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); 4420a586ceaSMark Shellenbaum int dmu_spill_hold_by_dnode(struct dnode *dn, uint32_t flags, 4430a586ceaSMark Shellenbaum void *tag, dmu_buf_t **dbp); 4440a586ceaSMark Shellenbaum int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); 445fa9e4066Sahrens 446fa9e4066Sahrens /* 447fa9e4066Sahrens * Obtain the DMU buffer from the specified object which contains the 448fa9e4066Sahrens * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so 449fa9e4066Sahrens * that it will remain in memory. You must release the hold with 450fa9e4066Sahrens * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your 451fa9e4066Sahrens * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. 452fa9e4066Sahrens * 453fa9e4066Sahrens * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill 454fa9e4066Sahrens * on the returned buffer before reading or writing the buffer's 455fa9e4066Sahrens * db_data. The comments for those routines describe what particular 456fa9e4066Sahrens * operations are valid after calling them. 457fa9e4066Sahrens * 458fa9e4066Sahrens * The object number must be a valid, allocated object number. 459fa9e4066Sahrens */ 460ea8dc4b6Seschrock int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, 46147cb52daSJeff Bonwick void *tag, dmu_buf_t **, int flags); 462fa9e4066Sahrens void dmu_buf_add_ref(dmu_buf_t *db, void* tag); 463ea8dc4b6Seschrock void dmu_buf_rele(dmu_buf_t *db, void *tag); 464fa9e4066Sahrens uint64_t dmu_buf_refcount(dmu_buf_t *db); 465fa9e4066Sahrens 466fa9e4066Sahrens /* 467fa9e4066Sahrens * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a 468fa9e4066Sahrens * range of an object. A pointer to an array of dmu_buf_t*'s is 469fa9e4066Sahrens * returned (in *dbpp). 470fa9e4066Sahrens * 471fa9e4066Sahrens * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and 472fa9e4066Sahrens * frees the array. The hold on the array of buffers MUST be released 473fa9e4066Sahrens * with dmu_buf_rele_array. You can NOT release the hold on each buffer 474fa9e4066Sahrens * individually with dmu_buf_rele. 475fa9e4066Sahrens */ 47613506d1eSmaybee int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, 47713506d1eSmaybee uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); 478ea8dc4b6Seschrock void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); 479fa9e4066Sahrens 480fa9e4066Sahrens /* 481fa9e4066Sahrens * Returns NULL on success, or the existing user ptr if it's already 482fa9e4066Sahrens * been set. 483fa9e4066Sahrens * 484fa9e4066Sahrens * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). 485fa9e4066Sahrens * 486fa9e4066Sahrens * user_data_ptr_ptr should be NULL, or a pointer to a pointer which 487fa9e4066Sahrens * will be set to db->db_data when you are allowed to access it. Note 488fa9e4066Sahrens * that db->db_data (the pointer) can change when you do dmu_buf_read(), 489fa9e4066Sahrens * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). 490fa9e4066Sahrens * *user_data_ptr_ptr will be set to the new value when it changes. 491fa9e4066Sahrens * 492fa9e4066Sahrens * If non-NULL, pageout func will be called when this buffer is being 493fa9e4066Sahrens * excised from the cache, so that you can clean up the data structure 494fa9e4066Sahrens * pointed to by user_ptr. 495fa9e4066Sahrens * 496fa9e4066Sahrens * dmu_evict_user() will call the pageout func for all buffers in a 497fa9e4066Sahrens * objset with a given pageout func. 498fa9e4066Sahrens */ 499fa9e4066Sahrens void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, 500fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 501fa9e4066Sahrens /* 502fa9e4066Sahrens * set_user_ie is the same as set_user, but request immediate eviction 503fa9e4066Sahrens * when hold count goes to zero. 504fa9e4066Sahrens */ 505fa9e4066Sahrens void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, 506fa9e4066Sahrens void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); 507fa9e4066Sahrens void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, 508fa9e4066Sahrens void *user_ptr, void *user_data_ptr_ptr, 509fa9e4066Sahrens dmu_buf_evict_func_t *pageout_func); 510fa9e4066Sahrens void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); 511fa9e4066Sahrens 512fa9e4066Sahrens /* 513fa9e4066Sahrens * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. 514fa9e4066Sahrens */ 515fa9e4066Sahrens void *dmu_buf_get_user(dmu_buf_t *db); 516fa9e4066Sahrens 51780901aeaSGeorge Wilson /* 51880901aeaSGeorge Wilson * Returns the blkptr associated with this dbuf, or NULL if not set. 51980901aeaSGeorge Wilson */ 52080901aeaSGeorge Wilson struct blkptr *dmu_buf_get_blkptr(dmu_buf_t *db); 52180901aeaSGeorge Wilson 522fa9e4066Sahrens /* 523fa9e4066Sahrens * Indicate that you are going to modify the buffer's data (db_data). 524fa9e4066Sahrens * 525fa9e4066Sahrens * The transaction (tx) must be assigned to a txg (ie. you've called 526fa9e4066Sahrens * dmu_tx_assign()). The buffer's object must be held in the tx 527fa9e4066Sahrens * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). 528fa9e4066Sahrens */ 529fa9e4066Sahrens void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); 530fa9e4066Sahrens 5313d692628SSanjeev Bagewadi /* 5323d692628SSanjeev Bagewadi * Tells if the given dbuf is freeable. 5333d692628SSanjeev Bagewadi */ 5343d692628SSanjeev Bagewadi boolean_t dmu_buf_freeable(dmu_buf_t *); 5353d692628SSanjeev Bagewadi 536fa9e4066Sahrens /* 537fa9e4066Sahrens * You must create a transaction, then hold the objects which you will 538fa9e4066Sahrens * (or might) modify as part of this transaction. Then you must assign 539fa9e4066Sahrens * the transaction to a transaction group. Once the transaction has 540fa9e4066Sahrens * been assigned, you can modify buffers which belong to held objects as 541fa9e4066Sahrens * part of this transaction. You can't modify buffers before the 542fa9e4066Sahrens * transaction has been assigned; you can't modify buffers which don't 543fa9e4066Sahrens * belong to objects which this transaction holds; you can't hold 544fa9e4066Sahrens * objects once the transaction has been assigned. You may hold an 545fa9e4066Sahrens * object which you are going to free (with dmu_object_free()), but you 546fa9e4066Sahrens * don't have to. 547fa9e4066Sahrens * 548fa9e4066Sahrens * You can abort the transaction before it has been assigned. 549fa9e4066Sahrens * 550fa9e4066Sahrens * Note that you may hold buffers (with dmu_buf_hold) at any time, 551fa9e4066Sahrens * regardless of transaction state. 552fa9e4066Sahrens */ 553fa9e4066Sahrens 554fa9e4066Sahrens #define DMU_NEW_OBJECT (-1ULL) 555fa9e4066Sahrens #define DMU_OBJECT_END (-1ULL) 556fa9e4066Sahrens 557fa9e4066Sahrens dmu_tx_t *dmu_tx_create(objset_t *os); 558fa9e4066Sahrens void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); 559fa9e4066Sahrens void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, 560fa9e4066Sahrens uint64_t len); 56114843421SMatthew Ahrens void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); 562fa9e4066Sahrens void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); 5630a586ceaSMark Shellenbaum void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); 5640a586ceaSMark Shellenbaum void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); 5650a586ceaSMark Shellenbaum void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); 566fa9e4066Sahrens void dmu_tx_abort(dmu_tx_t *tx); 5673b2aab18SMatthew Ahrens int dmu_tx_assign(dmu_tx_t *tx, enum txg_how txg_how); 5688a2f1b91Sahrens void dmu_tx_wait(dmu_tx_t *tx); 569fa9e4066Sahrens void dmu_tx_commit(dmu_tx_t *tx); 570fa9e4066Sahrens 571d20e665cSRicardo M. Correia /* 572d20e665cSRicardo M. Correia * To register a commit callback, dmu_tx_callback_register() must be called. 573d20e665cSRicardo M. Correia * 574d20e665cSRicardo M. Correia * dcb_data is a pointer to caller private data that is passed on as a 575d20e665cSRicardo M. Correia * callback parameter. The caller is responsible for properly allocating and 576d20e665cSRicardo M. Correia * freeing it. 577d20e665cSRicardo M. Correia * 578d20e665cSRicardo M. Correia * When registering a callback, the transaction must be already created, but 579d20e665cSRicardo M. Correia * it cannot be committed or aborted. It can be assigned to a txg or not. 580d20e665cSRicardo M. Correia * 581d20e665cSRicardo M. Correia * The callback will be called after the transaction has been safely written 582d20e665cSRicardo M. Correia * to stable storage and will also be called if the dmu_tx is aborted. 583d20e665cSRicardo M. Correia * If there is any error which prevents the transaction from being committed to 584d20e665cSRicardo M. Correia * disk, the callback will be called with a value of error != 0. 585d20e665cSRicardo M. Correia */ 586d20e665cSRicardo M. Correia typedef void dmu_tx_callback_func_t(void *dcb_data, int error); 587d20e665cSRicardo M. Correia 588d20e665cSRicardo M. Correia void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, 589d20e665cSRicardo M. Correia void *dcb_data); 590d20e665cSRicardo M. Correia 591fa9e4066Sahrens /* 592fa9e4066Sahrens * Free up the data blocks for a defined range of a file. If size is 593ad135b5dSChristopher Siden * -1, the range from offset to end-of-file is freed. 594fa9e4066Sahrens */ 595ea8dc4b6Seschrock int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, 596fa9e4066Sahrens uint64_t size, dmu_tx_t *tx); 597cdb0ab79Smaybee int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, 598cdb0ab79Smaybee uint64_t size); 599713d6c20SMatthew Ahrens int dmu_free_long_object(objset_t *os, uint64_t object); 600fa9e4066Sahrens 601fa9e4066Sahrens /* 602fa9e4066Sahrens * Convenience functions. 603fa9e4066Sahrens * 604fa9e4066Sahrens * Canfail routines will return 0 on success, or an errno if there is a 605fa9e4066Sahrens * nonrecoverable I/O error. 606fa9e4066Sahrens */ 6077bfdf011SNeil Perrin #define DMU_READ_PREFETCH 0 /* prefetch */ 6087bfdf011SNeil Perrin #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ 609ea8dc4b6Seschrock int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 6107bfdf011SNeil Perrin void *buf, uint32_t flags); 611fa9e4066Sahrens void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 612fa9e4066Sahrens const void *buf, dmu_tx_t *tx); 61382c9918fSTim Haley void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, 61482c9918fSTim Haley dmu_tx_t *tx); 615feb08c6bSbillm int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); 616feb08c6bSbillm int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, 617feb08c6bSbillm dmu_tx_t *tx); 61894d1a210STim Haley int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, 61994d1a210STim Haley dmu_tx_t *tx); 62044eda4d7Smaybee int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, 62144eda4d7Smaybee uint64_t size, struct page *pp, dmu_tx_t *tx); 6222fdbea25SAleksandr Guzovskiy struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); 6232fdbea25SAleksandr Guzovskiy void dmu_return_arcbuf(struct arc_buf *buf); 6242fdbea25SAleksandr Guzovskiy void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, 6252fdbea25SAleksandr Guzovskiy dmu_tx_t *tx); 626c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_init(struct xuio *uio, int niov); 627c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void dmu_xuio_fini(struct xuio *uio); 628c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, 629c242f9a0Schunli zhang - Sun Microsystems - Irvine United States size_t n); 630c242f9a0Schunli zhang - Sun Microsystems - Irvine United States int dmu_xuio_cnt(struct xuio *uio); 631c242f9a0Schunli zhang - Sun Microsystems - Irvine United States struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); 632c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void dmu_xuio_clear(struct xuio *uio, int i); 633c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void xuio_stat_wbuf_copied(); 634c242f9a0Schunli zhang - Sun Microsystems - Irvine United States void xuio_stat_wbuf_nocopy(); 635fa9e4066Sahrens 636416e0cd8Sek extern int zfs_prefetch_disable; 637416e0cd8Sek 638fa9e4066Sahrens /* 639fa9e4066Sahrens * Asynchronously try to read in the data. 640fa9e4066Sahrens */ 641fa9e4066Sahrens void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, 642fa9e4066Sahrens uint64_t len); 643fa9e4066Sahrens 644fa9e4066Sahrens typedef struct dmu_object_info { 645b24ab676SJeff Bonwick /* All sizes are in bytes unless otherwise indicated. */ 646fa9e4066Sahrens uint32_t doi_data_block_size; 647fa9e4066Sahrens uint32_t doi_metadata_block_size; 648fa9e4066Sahrens dmu_object_type_t doi_type; 649fa9e4066Sahrens dmu_object_type_t doi_bonus_type; 650b24ab676SJeff Bonwick uint64_t doi_bonus_size; 651fa9e4066Sahrens uint8_t doi_indirection; /* 2 = dnode->indirect->data */ 652fa9e4066Sahrens uint8_t doi_checksum; 653fa9e4066Sahrens uint8_t doi_compress; 654fa9e4066Sahrens uint8_t doi_pad[5]; 655b24ab676SJeff Bonwick uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ 656b24ab676SJeff Bonwick uint64_t doi_max_offset; 657b24ab676SJeff Bonwick uint64_t doi_fill_count; /* number of non-empty blocks */ 658fa9e4066Sahrens } dmu_object_info_t; 659fa9e4066Sahrens 660e45ce728Sahrens typedef void arc_byteswap_func_t(void *buf, size_t size); 661e45ce728Sahrens 662fa9e4066Sahrens typedef struct dmu_object_type_info { 663ad135b5dSChristopher Siden dmu_object_byteswap_t ot_byteswap; 664fa9e4066Sahrens boolean_t ot_metadata; 665fa9e4066Sahrens char *ot_name; 666fa9e4066Sahrens } dmu_object_type_info_t; 667fa9e4066Sahrens 668ad135b5dSChristopher Siden typedef struct dmu_object_byteswap_info { 669ad135b5dSChristopher Siden arc_byteswap_func_t *ob_func; 670ad135b5dSChristopher Siden char *ob_name; 671ad135b5dSChristopher Siden } dmu_object_byteswap_info_t; 672ad135b5dSChristopher Siden 673fa9e4066Sahrens extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; 674ad135b5dSChristopher Siden extern const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS]; 675fa9e4066Sahrens 676fa9e4066Sahrens /* 677fa9e4066Sahrens * Get information on a DMU object. 678fa9e4066Sahrens * 679fa9e4066Sahrens * Return 0 on success or ENOENT if object is not allocated. 680fa9e4066Sahrens * 681fa9e4066Sahrens * If doi is NULL, just indicates whether the object exists. 682fa9e4066Sahrens */ 683fa9e4066Sahrens int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); 6843e30c24aSWill Andrews /* Like dmu_object_info, but faster if you have a held dnode in hand. */ 685fa9e4066Sahrens void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); 6863e30c24aSWill Andrews /* Like dmu_object_info, but faster if you have a held dbuf in hand. */ 687fa9e4066Sahrens void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); 6883e30c24aSWill Andrews /* 6893e30c24aSWill Andrews * Like dmu_object_info_from_db, but faster still when you only care about 6903e30c24aSWill Andrews * the size. This is specifically optimized for zfs_getattr(). 6913e30c24aSWill Andrews */ 692fa9e4066Sahrens void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, 693fa9e4066Sahrens u_longlong_t *nblk512); 694fa9e4066Sahrens 695fa9e4066Sahrens typedef struct dmu_objset_stats { 696a2eea2e1Sahrens uint64_t dds_num_clones; /* number of clones of this */ 697a2eea2e1Sahrens uint64_t dds_creation_txg; 6983cb34c60Sahrens uint64_t dds_guid; 699fa9e4066Sahrens dmu_objset_type_t dds_type; 700fa9e4066Sahrens uint8_t dds_is_snapshot; 70131fd60d3Sahrens uint8_t dds_inconsistent; 7023cb34c60Sahrens char dds_origin[MAXNAMELEN]; 703a2eea2e1Sahrens } dmu_objset_stats_t; 704fa9e4066Sahrens 705a2eea2e1Sahrens /* 706a2eea2e1Sahrens * Get stats on a dataset. 707a2eea2e1Sahrens */ 708a2eea2e1Sahrens void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); 709fa9e4066Sahrens 710a2eea2e1Sahrens /* 711a2eea2e1Sahrens * Add entries to the nvlist for all the objset's properties. See 712a2eea2e1Sahrens * zfs_prop_table[] and zfs(1m) for details on the properties. 713a2eea2e1Sahrens */ 714a2eea2e1Sahrens void dmu_objset_stats(objset_t *os, struct nvlist *nv); 715fa9e4066Sahrens 716a2eea2e1Sahrens /* 717a2eea2e1Sahrens * Get the space usage statistics for statvfs(). 718a2eea2e1Sahrens * 719a2eea2e1Sahrens * refdbytes is the amount of space "referenced" by this objset. 720a2eea2e1Sahrens * availbytes is the amount of space available to this objset, taking 721a2eea2e1Sahrens * into account quotas & reservations, assuming that no other objsets 722a2eea2e1Sahrens * use the space first. These values correspond to the 'referenced' and 723a2eea2e1Sahrens * 'available' properties, described in the zfs(1m) manpage. 724a2eea2e1Sahrens * 725a2eea2e1Sahrens * usedobjs and availobjs are the number of objects currently allocated, 726a2eea2e1Sahrens * and available. 727a2eea2e1Sahrens */ 728a2eea2e1Sahrens void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, 729a2eea2e1Sahrens uint64_t *usedobjsp, uint64_t *availobjsp); 730fa9e4066Sahrens 731fa9e4066Sahrens /* 732a2eea2e1Sahrens * The fsid_guid is a 56-bit ID that can change to avoid collisions. 733a2eea2e1Sahrens * (Contrast with the ds_guid which is a 64-bit ID that will never 734a2eea2e1Sahrens * change, so there is a small probability that it will collide.) 735fa9e4066Sahrens */ 736a2eea2e1Sahrens uint64_t dmu_objset_fsid_guid(objset_t *os); 737fa9e4066Sahrens 73871eb0538SChris Kirby /* 73971eb0538SChris Kirby * Get the [cm]time for an objset's snapshot dir 74071eb0538SChris Kirby */ 74171eb0538SChris Kirby timestruc_t dmu_objset_snap_cmtime(objset_t *os); 74271eb0538SChris Kirby 743fa9e4066Sahrens int dmu_objset_is_snapshot(objset_t *os); 744fa9e4066Sahrens 745fa9e4066Sahrens extern struct spa *dmu_objset_spa(objset_t *os); 746fa9e4066Sahrens extern struct zilog *dmu_objset_zil(objset_t *os); 747fa9e4066Sahrens extern struct dsl_pool *dmu_objset_pool(objset_t *os); 748fa9e4066Sahrens extern struct dsl_dataset *dmu_objset_ds(objset_t *os); 749fa9e4066Sahrens extern void dmu_objset_name(objset_t *os, char *buf); 750fa9e4066Sahrens extern dmu_objset_type_t dmu_objset_type(objset_t *os); 751fa9e4066Sahrens extern uint64_t dmu_objset_id(objset_t *os); 752edf345e6SMatthew Ahrens extern zfs_sync_type_t dmu_objset_syncprop(objset_t *os); 753edf345e6SMatthew Ahrens extern zfs_logbias_op_t dmu_objset_logbias(objset_t *os); 754fa9e4066Sahrens extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, 755b38f0970Sck uint64_t *id, uint64_t *offp, boolean_t *case_conflict); 756ab04eb8eStimh extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, 757ab04eb8eStimh int maxlen, boolean_t *conflict); 75887e5029aSahrens extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, 75987e5029aSahrens uint64_t *idp, uint64_t *offp); 76014843421SMatthew Ahrens 7619966ca11SMatthew Ahrens typedef int objset_used_cb_t(dmu_object_type_t bonustype, 7629966ca11SMatthew Ahrens void *bonus, uint64_t *userp, uint64_t *groupp); 76314843421SMatthew Ahrens extern void dmu_objset_register_type(dmu_objset_type_t ost, 76414843421SMatthew Ahrens objset_used_cb_t *cb); 765f18faf3fSek extern void dmu_objset_set_user(objset_t *os, void *user_ptr); 766f18faf3fSek extern void *dmu_objset_get_user(objset_t *os); 767fa9e4066Sahrens 768fa9e4066Sahrens /* 769fa9e4066Sahrens * Return the txg number for the given assigned transaction. 770fa9e4066Sahrens */ 77187e5029aSahrens uint64_t dmu_tx_get_txg(dmu_tx_t *tx); 772fa9e4066Sahrens 773fa9e4066Sahrens /* 774fa9e4066Sahrens * Synchronous write. 775c5c6ffa0Smaybee * If a parent zio is provided this function initiates a write on the 776c5c6ffa0Smaybee * provided buffer as a child of the parent zio. 777da6c28aaSamw * In the absence of a parent zio, the write is completed synchronously. 778c5c6ffa0Smaybee * At write completion, blk is filled with the bp of the written block. 779fa9e4066Sahrens * Note that while the data covered by this function will be on stable 780c5c6ffa0Smaybee * storage when the write completes this new data does not become a 781fa9e4066Sahrens * permanent part of the file until the associated transaction commits. 782fa9e4066Sahrens */ 783b24ab676SJeff Bonwick 784b24ab676SJeff Bonwick /* 785b24ab676SJeff Bonwick * {zfs,zvol,ztest}_get_done() args 786b24ab676SJeff Bonwick */ 787b24ab676SJeff Bonwick typedef struct zgd { 788b24ab676SJeff Bonwick struct zilog *zgd_zilog; 789b24ab676SJeff Bonwick struct blkptr *zgd_bp; 790b24ab676SJeff Bonwick dmu_buf_t *zgd_db; 791b24ab676SJeff Bonwick struct rl *zgd_rl; 792b24ab676SJeff Bonwick void *zgd_private; 793b24ab676SJeff Bonwick } zgd_t; 794b24ab676SJeff Bonwick 795b24ab676SJeff Bonwick typedef void dmu_sync_cb_t(zgd_t *arg, int error); 796b24ab676SJeff Bonwick int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); 797fa9e4066Sahrens 798fa9e4066Sahrens /* 799fa9e4066Sahrens * Find the next hole or data block in file starting at *off 800fa9e4066Sahrens * Return found offset in *off. Return ESRCH for end of file. 801fa9e4066Sahrens */ 802fa9e4066Sahrens int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, 803fa9e4066Sahrens uint64_t *off); 804fa9e4066Sahrens 805fa9e4066Sahrens /* 806fa9e4066Sahrens * Initial setup and final teardown. 807fa9e4066Sahrens */ 808fa9e4066Sahrens extern void dmu_init(void); 809fa9e4066Sahrens extern void dmu_fini(void); 810fa9e4066Sahrens 811fa9e4066Sahrens typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, 812fa9e4066Sahrens uint64_t object, uint64_t offset, int len); 813fa9e4066Sahrens void dmu_traverse_objset(objset_t *os, uint64_t txg_start, 814fa9e4066Sahrens dmu_traverse_cb_t cb, void *arg); 815fa9e4066Sahrens 8163b2aab18SMatthew Ahrens int dmu_diff(const char *tosnap_name, const char *fromsnap_name, 8173b2aab18SMatthew Ahrens struct vnode *vp, offset_t *offp); 81899d5e173STim Haley 819fa9e4066Sahrens /* CRC64 table */ 820fa9e4066Sahrens #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ 821fa9e4066Sahrens extern uint64_t zfs_crc64_table[256]; 822fa9e4066Sahrens 823e4be62a2SSaso Kiselkov extern int zfs_mdcomp_disable; 824e4be62a2SSaso Kiselkov 825fa9e4066Sahrens #ifdef __cplusplus 826fa9e4066Sahrens } 827fa9e4066Sahrens #endif 828fa9e4066Sahrens 829fa9e4066Sahrens #endif /* _SYS_DMU_H */ 830