dmu.h revision 0b69c2f001a429251e2d38f25aca860396551214
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2006 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
#ifndef _SYS_DMU_H
#define _SYS_DMU_H
#pragma ident "%Z%%M% %I% %E% SMI"
/*
* This file describes the interface that the DMU provides for its
* consumers.
*
* The DMU also interacts with the SPA. That interface is described in
* dmu_spa.h.
*/
#include <sys/inttypes.h>
#ifdef __cplusplus
extern "C" {
#endif
struct uio;
struct vnode;
struct spa;
struct zilog;
struct zio;
struct blkptr;
struct zap_cursor;
struct dsl_dataset;
struct dsl_pool;
struct dnode;
struct drr_begin;
struct drr_end;
struct zbookmark;
struct spa;
typedef enum dmu_object_type {
/* general: */
DMU_OT_OBJECT_DIRECTORY, /* ZAP */
DMU_OT_OBJECT_ARRAY, /* UINT64 */
DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */
DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */
DMU_OT_BPLIST, /* UINT64 */
DMU_OT_BPLIST_HDR, /* UINT64 */
/* spa: */
DMU_OT_SPACE_MAP_HEADER, /* UINT64 */
DMU_OT_SPACE_MAP, /* UINT64 */
/* zil: */
DMU_OT_INTENT_LOG, /* UINT64 */
/* dmu: */
DMU_OT_DNODE, /* DNODE */
DMU_OT_OBJSET, /* OBJSET */
/* dsl: */
DMU_OT_DSL_DIR, /* UINT64 */
DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */
DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */
DMU_OT_DSL_PROPS, /* ZAP */
DMU_OT_DSL_DATASET, /* UINT64 */
/* zpl: */
DMU_OT_ZNODE, /* ZNODE */
DMU_OT_ACL, /* ACL */
DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */
DMU_OT_DIRECTORY_CONTENTS, /* ZAP */
DMU_OT_MASTER_NODE, /* ZAP */
DMU_OT_DELETE_QUEUE, /* ZAP */
/* zvol: */
DMU_OT_ZVOL, /* UINT8 */
DMU_OT_ZVOL_PROP, /* ZAP */
/* other; for testing only! */
DMU_OT_PLAIN_OTHER, /* UINT8 */
DMU_OT_UINT64_OTHER, /* UINT64 */
DMU_OT_ZAP_OTHER, /* ZAP */
/* new object types: */
DMU_OT_ERROR_LOG, /* ZAP */
typedef enum dmu_objset_type {
DMU_OST_OTHER, /* For testing only! */
DMU_OST_ANY, /* Be careful! */
#define DS_MODE_NONE 0 /* invalid, to aid debugging */
#define DS_MODE_LEVELS 4
#define DS_MODE_READONLY 0x8
#define DS_MODE_IS_READONLY(x) ((x) & DS_MODE_READONLY)
#define DS_MODE_INCONSISTENT 0x10
#define DS_MODE_IS_INCONSISTENT(x) ((x) & DS_MODE_INCONSISTENT)
#define DS_FIND_SNAPSHOTS (1<<0)
/*
* The maximum number of bytes that can be accessed as part of one
* operation, including metadata.
*/
/*
* Public routines to create, destroy, open, and close objsets.
*/
int dmu_objset_destroy(const char *name);
int dmu_objset_rollback(const char *name);
int flags);
typedef struct dmu_buf {
void *db_data; /* data in buffer */
} dmu_buf_t;
/*
* Callback function to perform byte swapping on a block.
*/
/*
* The names of zap entries in the DIRECTORY_OBJECT of the MOS.
*/
#define DMU_POOL_DIRECTORY_OBJECT 1
#define DMU_POOL_CONFIG "config"
#define DMU_POOL_ROOT_DATASET "root_dataset"
#define DMU_POOL_SYNC_BPLIST "sync_bplist"
#define DMU_POOL_ERRLOG_SCRUB "errlog_scrub"
#define DMU_POOL_ERRLOG_LAST "errlog_last"
#define DMU_POOL_SPARES "spares"
#define DMU_POOL_DEFLATE "deflate"
/*
* Allocate an object from this objset. The range of object numbers
* available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode.
*
* The transaction must be assigned to a txg. The newly allocated
* object will be "held" in the transaction (ie. you can modify the
* newly allocated object in this transaction).
*
* dmu_object_alloc() chooses an object and returns it in *objectp.
*
* dmu_object_claim() allocates a specific object number. If that
* number is already allocated, it fails and returns EEXIST.
*
* Return 0 on success, or ENOSPC or EEXIST as specified above.
*/
/*
* Free an object from this objset.
*
* The object's data will be freed as well (ie. you don't need to call
* dmu_free(object, 0, -1, tx)).
*
* The object need not be held in the transaction.
*
* If there are any holds on this object's buffers (via dmu_buf_hold()),
* or tx holds on the object (via dmu_tx_hold_object()), you can not
* free it; it fails and returns EBUSY.
*
* If the object is not allocated, it fails and returns ENOENT.
*
* Return 0 on success, or EBUSY or ENOENT as specified above.
*/
/*
* Find the next allocated or free object.
*
* The objectp parameter is in-out. It will be updated to be the next
* object which is allocated.
*
* XXX Can only be called on a objset with no dirty data.
*
* Returns 0 on success, or ENOENT if there are no more objects.
*/
/*
* Set the data blocksize for an object.
*
* The object cannot have any blocks allcated beyond the first. If
* the first block is allocated already, the new size must be greater
* than the current block size. If these conditions are not met,
* ENOTSUP will be returned.
*
* Returns 0 on success, or EBUSY if there are any holds on the object
* contents, or ENOTSUP as described above.
*/
/*
* Set the checksum property on a dnode. The new checksum algorithm will
* apply to all newly written blocks; existing blocks will not be affected.
*/
/*
* Set the compress property on a dnode. The new compression algorithm will
* apply to all newly written blocks; existing blocks will not be affected.
*/
/*
* Decide how many copies of a given block we should make. Can be from
* 1 to SPA_DVAS_PER_BP.
*/
/*
* The bonus data is accessed more or less like a regular buffer.
* You must dmu_bonus_hold() to get the buffer, which will give you a
* dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus
* data. As with any normal buffer, you must call dmu_buf_read() to
* read db_data, dmu_buf_will_dirty() before modifying it, and the
* object must be held in an assigned transaction before calling
* dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus
* buffer as well. You must release your hold with dmu_buf_rele().
*/
int dmu_bonus_max(void);
/*
* Obtain the DMU buffer from the specified object which contains the
* specified offset. dmu_buf_hold() puts a "hold" on the buffer, so
* that it will remain in memory. You must release the hold with
* dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your
* hold. You must have a hold on any dmu_buf_t* you pass to the DMU.
*
* You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill
* on the returned buffer before reading or writing the buffer's
* db_data. The comments for those routines describe what particular
* operations are valid after calling them.
*
* The object number must be a valid, allocated object number.
*/
/*
* dmu_buf_hold_array holds the DMU buffers which contain all bytes in a
* range of an object. A pointer to an array of dmu_buf_t*'s is
* returned (in *dbpp).
*
* dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and
* frees the array. The hold on the array of buffers MUST be released
* with dmu_buf_rele_array. You can NOT release the hold on each buffer
* individually with dmu_buf_rele.
*/
/*
* Returns NULL on success, or the existing user ptr if it's already
* been set.
*
* user_ptr is for use by the user and can be obtained via dmu_buf_get_user().
*
* user_data_ptr_ptr should be NULL, or a pointer to a pointer which
* will be set to db->db_data when you are allowed to access it. Note
* that db->db_data (the pointer) can change when you do dmu_buf_read(),
* dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill().
* *user_data_ptr_ptr will be set to the new value when it changes.
*
* If non-NULL, pageout func will be called when this buffer is being
* excised from the cache, so that you can clean up the data structure
* pointed to by user_ptr.
*
* dmu_evict_user() will call the pageout func for all buffers in a
* objset with a given pageout func.
*/
/*
* set_user_ie is the same as set_user, but request immediate eviction
* when hold count goes to zero.
*/
void *user_ptr, void *user_data_ptr_ptr,
/*
* Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set.
*/
/*
* Indicate that you are going to modify the buffer's data (db_data).
*
* The transaction (tx) must be assigned to a txg (ie. you've called
* dmu_tx_assign()). The buffer's object must be held in the tx
* (ie. you've called dmu_tx_hold_object(tx, db->db_object)).
*/
/*
* You must create a transaction, then hold the objects which you will
* (or might) modify as part of this transaction. Then you must assign
* the transaction to a transaction group. Once the transaction has
* been assigned, you can modify buffers which belong to held objects as
* part of this transaction. You can't modify buffers before the
* transaction has been assigned; you can't modify buffers which don't
* belong to objects which this transaction holds; you can't hold
* objects once the transaction has been assigned. You may hold an
* object which you are going to free (with dmu_object_free()), but you
* don't have to.
*
* You can abort the transaction before it has been assigned.
*
* Note that you may hold buffers (with dmu_buf_hold) at any time,
* regardless of transaction state.
*/
#define DMU_NEW_OBJECT (-1ULL)
#define DMU_OBJECT_END (-1ULL)
/*
* Free up the data blocks for a defined range of a file. If size is
* zero, the range from offset to end-of-file is freed.
*/
/*
* Convenience functions.
*
* Canfail routines will return 0 on success, or an errno if there is a
* nonrecoverable I/O error.
*/
void *buf);
/*
* Asynchronously try to read in the data.
*/
typedef struct dmu_object_info {
/* All sizes are in bytes. */
/* Values below are number of 512-byte blocks. */
typedef struct dmu_object_type_info {
char *ot_name;
/*
* Get information on a DMU object.
*
* Return 0 on success or ENOENT if object is not allocated.
*
* If doi is NULL, just indicates whether the object exists.
*/
/*
* Get the maximum nonzero offset in the object (ie. this offset and all
* offsets following are zero).
*
* XXX Perhaps integrate this with dmu_object_info(), although that
* would then have to bring in the indirect blocks.
*/
typedef struct dmu_objset_stats {
char dds_clone_of[MAXNAMELEN];
/* How much data is there in this objset? */
/*
* Space referenced, taking into account pending writes and
* frees. Only relavent to filesystems and snapshots (not
* collections).
*/
/*
* Space "used", taking into account pending writes and frees, and
* children's reservations (in bytes). This is the amount of
* space that will be freed if this and all dependent items are
* destroyed (eg. child datasets, objsets, and snapshots). So
* for snapshots, this is the amount of space unique to this
* snapshot.
*/
/*
* Compressed and uncompressed bytes consumed. Does not take
* into account reservations. Used for computing compression
* ratio.
*/
/*
* The ds_fsid_guid is a 56-bit ID that can change to avoid
* collisions. The ds_guid is a 64-bit ID that will never
* change, so there is a small probability that it will collide.
*/
/* The dataset's administratively-set quota, in bytes. */
/* The dataset's administratively-set reservation, in bytes */
/*
* The amount of additional space that this dataset can consume.
* Takes into account quotas & reservations.
* (Assuming that no other datasets consume it first.)
*/
/*
* Used for debugging purposes
*/
/*
* Get stats on a dataset.
*/
/*
* Return the txg number for the given assigned transaction.
*/
/*
* Synchronous write.
* If a parent zio is provided this function initiates a write on the
* provided buffer as a child of the parent zio.
* In the absense of a parent zio, the write is completed synchronously.
* At write completion, blk is filled with the bp of the written block.
* Note that while the data covered by this function will be on stable
* storage when the write completes this new data does not become a
* permanent part of the file until the associated transaction commits.
*/
/*
* Find the next hole or data block in file starting at *off
* Return found offset in *off. Return ESRCH for end of file.
*/
/*
* Initial setup and final teardown.
*/
extern void dmu_init(void);
extern void dmu_fini(void);
/* CRC64 table */
#ifdef __cplusplus
}
#endif
#endif /* _SYS_DMU_H */