2 * QEMU System Emulator block driver
4 * Copyright (c) 2003 Fabrice Bellard
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to deal
8 * in the Software without restriction, including without limitation the rights
9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 * copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
27 #include "block/accounting.h"
28 #include "block/block.h"
29 #include "qemu/option.h"
30 #include "qemu/queue.h"
31 #include "qemu/coroutine.h"
32 #include "qemu/timer.h"
33 #include "qapi-types.h"
34 #include "qemu/hbitmap.h"
35 #include "block/snapshot.h"
36 #include "qemu/main-loop.h"
37 #include "qemu/throttle.h"
39 #define BLOCK_FLAG_ENCRYPT 1
40 #define BLOCK_FLAG_LAZY_REFCOUNTS 8
42 #define BLOCK_OPT_SIZE "size"
43 #define BLOCK_OPT_ENCRYPT "encryption"
44 #define BLOCK_OPT_COMPAT6 "compat6"
45 #define BLOCK_OPT_HWVERSION "hwversion"
46 #define BLOCK_OPT_BACKING_FILE "backing_file"
47 #define BLOCK_OPT_BACKING_FMT "backing_fmt"
48 #define BLOCK_OPT_CLUSTER_SIZE "cluster_size"
49 #define BLOCK_OPT_TABLE_SIZE "table_size"
50 #define BLOCK_OPT_PREALLOC "preallocation"
51 #define BLOCK_OPT_SUBFMT "subformat"
52 #define BLOCK_OPT_COMPAT_LEVEL "compat"
53 #define BLOCK_OPT_LAZY_REFCOUNTS "lazy_refcounts"
54 #define BLOCK_OPT_ADAPTER_TYPE "adapter_type"
55 #define BLOCK_OPT_REDUNDANCY "redundancy"
56 #define BLOCK_OPT_NOCOW "nocow"
57 #define BLOCK_OPT_OBJECT_SIZE "object_size"
58 #define BLOCK_OPT_REFCOUNT_BITS "refcount_bits"
60 #define BLOCK_PROBE_BUF_SIZE 512
62 enum BdrvTrackedRequestType
{
70 typedef struct BdrvTrackedRequest
{
74 enum BdrvTrackedRequestType type
;
77 int64_t overlap_offset
;
78 unsigned int overlap_bytes
;
80 QLIST_ENTRY(BdrvTrackedRequest
) list
;
81 Coroutine
*co
; /* owner, used for deadlock detection */
82 CoQueue wait_queue
; /* coroutines blocked on this request */
84 struct BdrvTrackedRequest
*waiting_for
;
88 const char *format_name
;
91 /* set to true if the BlockDriver is a block filter */
93 /* for snapshots block filter like Quorum can implement the
94 * following recursive callback.
95 * It's purpose is to recurse on the filter children while calling
96 * bdrv_recurse_is_first_non_filter on them.
97 * For a sample implementation look in the future Quorum block filter.
99 bool (*bdrv_recurse_is_first_non_filter
)(BlockDriverState
*bs
,
100 BlockDriverState
*candidate
);
102 int (*bdrv_probe
)(const uint8_t *buf
, int buf_size
, const char *filename
);
103 int (*bdrv_probe_device
)(const char *filename
);
105 /* Any driver implementing this callback is expected to be able to handle
106 * NULL file names in its .bdrv_open() implementation */
107 void (*bdrv_parse_filename
)(const char *filename
, QDict
*options
, Error
**errp
);
108 /* Drivers not implementing bdrv_parse_filename nor bdrv_open should have
109 * this field set to true, except ones that are defined only by their
111 * An example of the last type will be the quorum block driver.
113 bool bdrv_needs_filename
;
115 /* Set if a driver can support backing files */
116 bool supports_backing
;
118 /* For handling image reopen for split or non-split files */
119 int (*bdrv_reopen_prepare
)(BDRVReopenState
*reopen_state
,
120 BlockReopenQueue
*queue
, Error
**errp
);
121 void (*bdrv_reopen_commit
)(BDRVReopenState
*reopen_state
);
122 void (*bdrv_reopen_abort
)(BDRVReopenState
*reopen_state
);
123 void (*bdrv_join_options
)(QDict
*options
, QDict
*old_options
);
125 int (*bdrv_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
127 int (*bdrv_file_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
129 void (*bdrv_close
)(BlockDriverState
*bs
);
130 int (*bdrv_create
)(const char *filename
, QemuOpts
*opts
, Error
**errp
);
131 int (*bdrv_set_key
)(BlockDriverState
*bs
, const char *key
);
132 int (*bdrv_make_empty
)(BlockDriverState
*bs
);
134 void (*bdrv_refresh_filename
)(BlockDriverState
*bs
, QDict
*options
);
137 BlockAIOCB
*(*bdrv_aio_readv
)(BlockDriverState
*bs
,
138 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
139 BlockCompletionFunc
*cb
, void *opaque
);
140 BlockAIOCB
*(*bdrv_aio_writev
)(BlockDriverState
*bs
,
141 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
142 BlockCompletionFunc
*cb
, void *opaque
);
143 BlockAIOCB
*(*bdrv_aio_flush
)(BlockDriverState
*bs
,
144 BlockCompletionFunc
*cb
, void *opaque
);
145 BlockAIOCB
*(*bdrv_aio_discard
)(BlockDriverState
*bs
,
146 int64_t sector_num
, int nb_sectors
,
147 BlockCompletionFunc
*cb
, void *opaque
);
149 int coroutine_fn (*bdrv_co_readv
)(BlockDriverState
*bs
,
150 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
151 int coroutine_fn (*bdrv_co_preadv
)(BlockDriverState
*bs
,
152 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
153 int coroutine_fn (*bdrv_co_writev
)(BlockDriverState
*bs
,
154 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
155 int coroutine_fn (*bdrv_co_writev_flags
)(BlockDriverState
*bs
,
156 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
, int flags
);
157 int coroutine_fn (*bdrv_co_pwritev
)(BlockDriverState
*bs
,
158 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
161 * Efficiently zero a region of the disk image. Typically an image format
162 * would use a compact metadata representation to implement this. This
163 * function pointer may be NULL or return -ENOSUP and .bdrv_co_writev()
164 * will be called instead.
166 int coroutine_fn (*bdrv_co_pwrite_zeroes
)(BlockDriverState
*bs
,
167 int64_t offset
, int count
, BdrvRequestFlags flags
);
168 int coroutine_fn (*bdrv_co_discard
)(BlockDriverState
*bs
,
169 int64_t sector_num
, int nb_sectors
);
170 int64_t coroutine_fn (*bdrv_co_get_block_status
)(BlockDriverState
*bs
,
171 int64_t sector_num
, int nb_sectors
, int *pnum
,
172 BlockDriverState
**file
);
175 * Invalidate any cached meta-data.
177 void (*bdrv_invalidate_cache
)(BlockDriverState
*bs
, Error
**errp
);
178 int (*bdrv_inactivate
)(BlockDriverState
*bs
);
181 * Flushes all data for all layers by calling bdrv_co_flush for underlying
182 * layers, if needed. This function is needed for deterministic
183 * synchronization of the flush finishing callback.
185 int coroutine_fn (*bdrv_co_flush
)(BlockDriverState
*bs
);
188 * Flushes all data that was already written to the OS all the way down to
189 * the disk (for example raw-posix calls fsync()).
191 int coroutine_fn (*bdrv_co_flush_to_disk
)(BlockDriverState
*bs
);
194 * Flushes all internal caches to the OS. The data may still sit in a
195 * writeback cache of the host OS, but it will survive a crash of the qemu
198 int coroutine_fn (*bdrv_co_flush_to_os
)(BlockDriverState
*bs
);
200 const char *protocol_name
;
201 int (*bdrv_truncate
)(BlockDriverState
*bs
, int64_t offset
);
203 int64_t (*bdrv_getlength
)(BlockDriverState
*bs
);
204 bool has_variable_length
;
205 int64_t (*bdrv_get_allocated_file_size
)(BlockDriverState
*bs
);
207 int (*bdrv_write_compressed
)(BlockDriverState
*bs
, int64_t sector_num
,
208 const uint8_t *buf
, int nb_sectors
);
210 int (*bdrv_snapshot_create
)(BlockDriverState
*bs
,
211 QEMUSnapshotInfo
*sn_info
);
212 int (*bdrv_snapshot_goto
)(BlockDriverState
*bs
,
213 const char *snapshot_id
);
214 int (*bdrv_snapshot_delete
)(BlockDriverState
*bs
,
215 const char *snapshot_id
,
218 int (*bdrv_snapshot_list
)(BlockDriverState
*bs
,
219 QEMUSnapshotInfo
**psn_info
);
220 int (*bdrv_snapshot_load_tmp
)(BlockDriverState
*bs
,
221 const char *snapshot_id
,
224 int (*bdrv_get_info
)(BlockDriverState
*bs
, BlockDriverInfo
*bdi
);
225 ImageInfoSpecific
*(*bdrv_get_specific_info
)(BlockDriverState
*bs
);
227 int coroutine_fn (*bdrv_save_vmstate
)(BlockDriverState
*bs
,
230 int coroutine_fn (*bdrv_load_vmstate
)(BlockDriverState
*bs
,
234 int (*bdrv_change_backing_file
)(BlockDriverState
*bs
,
235 const char *backing_file
, const char *backing_fmt
);
237 /* removable device specific */
238 bool (*bdrv_is_inserted
)(BlockDriverState
*bs
);
239 int (*bdrv_media_changed
)(BlockDriverState
*bs
);
240 void (*bdrv_eject
)(BlockDriverState
*bs
, bool eject_flag
);
241 void (*bdrv_lock_medium
)(BlockDriverState
*bs
, bool locked
);
243 /* to control generic scsi devices */
244 BlockAIOCB
*(*bdrv_aio_ioctl
)(BlockDriverState
*bs
,
245 unsigned long int req
, void *buf
,
246 BlockCompletionFunc
*cb
, void *opaque
);
248 /* List of options for creating images, terminated by name == NULL */
249 QemuOptsList
*create_opts
;
252 * Returns 0 for completed check, -errno for internal errors.
253 * The check results are stored in result.
255 int (*bdrv_check
)(BlockDriverState
* bs
, BdrvCheckResult
*result
,
258 int (*bdrv_amend_options
)(BlockDriverState
*bs
, QemuOpts
*opts
,
259 BlockDriverAmendStatusCB
*status_cb
,
262 void (*bdrv_debug_event
)(BlockDriverState
*bs
, BlkdebugEvent event
);
264 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
265 int (*bdrv_debug_breakpoint
)(BlockDriverState
*bs
, const char *event
,
267 int (*bdrv_debug_remove_breakpoint
)(BlockDriverState
*bs
,
269 int (*bdrv_debug_resume
)(BlockDriverState
*bs
, const char *tag
);
270 bool (*bdrv_debug_is_suspended
)(BlockDriverState
*bs
, const char *tag
);
272 void (*bdrv_refresh_limits
)(BlockDriverState
*bs
, Error
**errp
);
275 * Returns 1 if newly created images are guaranteed to contain only
276 * zeros, 0 otherwise.
278 int (*bdrv_has_zero_init
)(BlockDriverState
*bs
);
280 /* Remove fd handlers, timers, and other event loop callbacks so the event
281 * loop is no longer in use. Called with no in-flight requests and in
282 * depth-first traversal order with parents before child nodes.
284 void (*bdrv_detach_aio_context
)(BlockDriverState
*bs
);
286 /* Add fd handlers, timers, and other event loop callbacks so I/O requests
287 * can be processed again. Called with no in-flight requests and in
288 * depth-first traversal order with child nodes before parent nodes.
290 void (*bdrv_attach_aio_context
)(BlockDriverState
*bs
,
291 AioContext
*new_context
);
293 /* io queue for linux-aio */
294 void (*bdrv_io_plug
)(BlockDriverState
*bs
);
295 void (*bdrv_io_unplug
)(BlockDriverState
*bs
);
298 * Try to get @bs's logical and physical block size.
299 * On success, store them in @bsz and return zero.
300 * On failure, return negative errno.
302 int (*bdrv_probe_blocksizes
)(BlockDriverState
*bs
, BlockSizes
*bsz
);
304 * Try to get @bs's geometry (cyls, heads, sectors)
305 * On success, store them in @geo and return 0.
306 * On failure return -errno.
307 * Only drivers that want to override guest geometry implement this
308 * callback; see hd_geometry_guess().
310 int (*bdrv_probe_geometry
)(BlockDriverState
*bs
, HDGeometry
*geo
);
313 * Drain and stop any internal sources of requests in the driver, and
314 * remain so until next I/O callback (e.g. bdrv_co_writev) is called.
316 void (*bdrv_drain
)(BlockDriverState
*bs
);
318 void (*bdrv_add_child
)(BlockDriverState
*parent
, BlockDriverState
*child
,
320 void (*bdrv_del_child
)(BlockDriverState
*parent
, BdrvChild
*child
,
323 QLIST_ENTRY(BlockDriver
) list
;
326 typedef struct BlockLimits
{
327 /* maximum number of bytes that can be discarded at once (since it
328 * is signed, it must be < 2G, if set), should be multiple of
329 * pdiscard_alignment, but need not be power of 2. May be 0 if no
330 * inherent 32-bit limit */
331 int32_t max_pdiscard
;
333 /* optimal alignment for discard requests in bytes, must be power
334 * of 2, less than max_pdiscard if that is set, and multiple of
335 * bs->request_alignment. May be 0 if bs->request_alignment is
337 uint32_t pdiscard_alignment
;
339 /* maximum number of bytes that can zeroized at once (since it is
340 * signed, it must be < 2G, if set), should be multiple of
341 * pwrite_zeroes_alignment. May be 0 if no inherent 32-bit limit */
342 int32_t max_pwrite_zeroes
;
344 /* optimal alignment for write zeroes requests in bytes, must be
345 * power of 2, less than max_pwrite_zeroes if that is set, and
346 * multiple of bs->request_alignment. May be 0 if
347 * bs->request_alignment is good enough */
348 uint32_t pwrite_zeroes_alignment
;
350 /* optimal transfer length in bytes (must be power of 2, and
351 * multiple of bs->request_alignment), or 0 if no preferred size */
352 uint32_t opt_transfer
;
354 /* maximal transfer length in bytes (need not be power of 2, but
355 * should be multiple of opt_transfer), or 0 for no 32-bit limit.
356 * For now, anything larger than INT_MAX is clamped down. */
357 uint32_t max_transfer
;
359 /* memory alignment so that no bounce buffer is needed */
360 size_t min_mem_alignment
;
362 /* memory alignment for bounce buffer */
363 size_t opt_mem_alignment
;
365 /* maximum number of iovec elements */
369 typedef struct BdrvOpBlocker BdrvOpBlocker
;
371 typedef struct BdrvAioNotifier
{
372 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
);
373 void (*detach_aio_context
)(void *opaque
);
378 QLIST_ENTRY(BdrvAioNotifier
) list
;
381 struct BdrvChildRole
{
382 void (*inherit_options
)(int *child_flags
, QDict
*child_options
,
383 int parent_flags
, QDict
*parent_options
);
385 void (*change_media
)(BdrvChild
*child
, bool load
);
386 void (*resize
)(BdrvChild
*child
);
388 /* Returns a name that is supposedly more useful for human users than the
389 * node name for identifying the node in question (in particular, a BB
390 * name), or NULL if the parent can't provide a better name. */
391 const char* (*get_name
)(BdrvChild
*child
);
394 * If this pair of functions is implemented, the parent doesn't issue new
395 * requests after returning from .drained_begin() until .drained_end() is
398 * Note that this can be nested. If drained_begin() was called twice, new
399 * I/O is allowed only after drained_end() was called twice, too.
401 void (*drained_begin
)(BdrvChild
*child
);
402 void (*drained_end
)(BdrvChild
*child
);
405 extern const BdrvChildRole child_file
;
406 extern const BdrvChildRole child_format
;
409 BlockDriverState
*bs
;
411 const BdrvChildRole
*role
;
413 QLIST_ENTRY(BdrvChild
) next
;
414 QLIST_ENTRY(BdrvChild
) next_parent
;
418 * Note: the function bdrv_append() copies and swaps contents of
419 * BlockDriverStates, so if you add new fields to this struct, please
420 * inspect bdrv_append() to determine if the new fields need to be
423 struct BlockDriverState
{
424 int64_t total_sectors
; /* if we are reading a disk image, give its
426 int read_only
; /* if true, the media is read only */
427 int open_flags
; /* flags used to open the file, re-used for re-open */
428 int encrypted
; /* if true, the media is encrypted */
429 int valid_key
; /* if true, a valid encryption key has been set */
430 int sg
; /* if true, the device is a /dev/sg* */
431 int copy_on_read
; /* if true, copy read backing sectors into image
432 note this is a reference count */
435 BlockDriver
*drv
; /* NULL means no media */
438 AioContext
*aio_context
; /* event loop used for fd handlers, timers, etc */
439 /* long-running tasks intended to always use the same AioContext as this
440 * BDS may register themselves in this list to be notified of changes
441 * regarding this BDS's context */
442 QLIST_HEAD(, BdrvAioNotifier
) aio_notifiers
;
443 bool walking_aio_notifiers
; /* to make removal during iteration safe */
445 char filename
[PATH_MAX
];
446 char backing_file
[PATH_MAX
]; /* if non zero, the image is a diff of
448 char backing_format
[16]; /* if non-zero and backing_file exists */
450 QDict
*full_open_options
;
451 char exact_filename
[PATH_MAX
];
456 /* Callback before write request is processed */
457 NotifierWithReturnList before_write_notifiers
;
459 /* number of in-flight serialising requests */
460 unsigned int serialising_in_flight
;
462 /* Offset after the highest byte written to */
463 uint64_t wr_highest_offset
;
468 /* Alignment requirement for offset/length of I/O requests */
469 unsigned int request_alignment
;
470 /* Flags honored during pwrite (so far: BDRV_REQ_FUA) */
471 unsigned int supported_write_flags
;
472 /* Flags honored during pwrite_zeroes (so far: BDRV_REQ_FUA,
473 * BDRV_REQ_MAY_UNMAP) */
474 unsigned int supported_zero_flags
;
476 /* the following member gives a name to every node on the bs graph. */
478 /* element of the list of named nodes building the graph */
479 QTAILQ_ENTRY(BlockDriverState
) node_list
;
480 /* element of the list of all BlockDriverStates (all_bdrv_states) */
481 QTAILQ_ENTRY(BlockDriverState
) bs_list
;
482 /* element of the list of monitor-owned BDS */
483 QTAILQ_ENTRY(BlockDriverState
) monitor_list
;
484 QLIST_HEAD(, BdrvDirtyBitmap
) dirty_bitmaps
;
487 QLIST_HEAD(, BdrvTrackedRequest
) tracked_requests
;
489 /* operation blockers */
490 QLIST_HEAD(, BdrvOpBlocker
) op_blockers
[BLOCK_OP_TYPE_MAX
];
492 /* long-running background operation */
495 /* The node that this node inherited default options from (and a reopen on
496 * which can affect this node by changing these defaults). This is always a
497 * parent node of this node. */
498 BlockDriverState
*inherits_from
;
499 QLIST_HEAD(, BdrvChild
) children
;
500 QLIST_HEAD(, BdrvChild
) parents
;
503 QDict
*explicit_options
;
504 BlockdevDetectZeroesOptions detect_zeroes
;
506 /* The error object in use for blocking operations on backing_hd */
507 Error
*backing_blocker
;
509 /* threshold limit for writes, in bytes. "High water mark". */
510 uint64_t write_threshold_offset
;
511 NotifierWithReturn write_threshold_notifier
;
513 /* counters for nested bdrv_io_plug and bdrv_io_unplugged_begin */
515 unsigned io_plug_disabled
;
520 struct BlockBackendRootState
{
523 BlockdevDetectZeroesOptions detect_zeroes
;
526 typedef enum BlockMirrorBackingMode
{
527 /* Reuse the existing backing chain from the source for the target.
528 * - sync=full: Set backing BDS to NULL.
529 * - sync=top: Use source's backing BDS.
530 * - sync=none: Use source as the backing BDS. */
531 MIRROR_SOURCE_BACKING_CHAIN
,
533 /* Open the target's backing chain completely anew */
534 MIRROR_OPEN_BACKING_CHAIN
,
536 /* Do not change the target's backing BDS after job completion */
537 MIRROR_LEAVE_BACKING_CHAIN
,
538 } BlockMirrorBackingMode
;
540 static inline BlockDriverState
*backing_bs(BlockDriverState
*bs
)
542 return bs
->backing
? bs
->backing
->bs
: NULL
;
546 /* Essential block drivers which must always be statically linked into qemu, and
547 * which therefore can be accessed without using bdrv_find_format() */
548 extern BlockDriver bdrv_file
;
549 extern BlockDriver bdrv_raw
;
550 extern BlockDriver bdrv_qcow2
;
553 * bdrv_setup_io_funcs:
555 * Prepare a #BlockDriver for I/O request processing by populating
556 * unimplemented coroutine and AIO interfaces with generic wrapper functions
557 * that fall back to implemented interfaces.
559 void bdrv_setup_io_funcs(BlockDriver
*bdrv
);
561 int coroutine_fn
bdrv_co_preadv(BlockDriverState
*bs
,
562 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
563 BdrvRequestFlags flags
);
564 int coroutine_fn
bdrv_co_pwritev(BlockDriverState
*bs
,
565 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
566 BdrvRequestFlags flags
);
568 int get_tmp_filename(char *filename
, int size
);
569 BlockDriver
*bdrv_probe_all(const uint8_t *buf
, int buf_size
,
570 const char *filename
);
574 * bdrv_add_before_write_notifier:
576 * Register a callback that is invoked before write requests are processed but
577 * after any throttling or waiting for overlapping requests.
579 void bdrv_add_before_write_notifier(BlockDriverState
*bs
,
580 NotifierWithReturn
*notifier
);
583 * bdrv_detach_aio_context:
585 * May be called from .bdrv_detach_aio_context() to detach children from the
586 * current #AioContext. This is only needed by block drivers that manage their
587 * own children. Both ->file and ->backing are automatically handled and
588 * block drivers should not call this function on them explicitly.
590 void bdrv_detach_aio_context(BlockDriverState
*bs
);
593 * bdrv_attach_aio_context:
595 * May be called from .bdrv_attach_aio_context() to attach children to the new
596 * #AioContext. This is only needed by block drivers that manage their own
597 * children. Both ->file and ->backing are automatically handled and block
598 * drivers should not call this function on them explicitly.
600 void bdrv_attach_aio_context(BlockDriverState
*bs
,
601 AioContext
*new_context
);
604 * bdrv_add_aio_context_notifier:
606 * If a long-running job intends to be always run in the same AioContext as a
607 * certain BDS, it may use this function to be notified of changes regarding the
608 * association of the BDS to an AioContext.
610 * attached_aio_context() is called after the target BDS has been attached to a
611 * new AioContext; detach_aio_context() is called before the target BDS is being
612 * detached from its old AioContext.
614 void bdrv_add_aio_context_notifier(BlockDriverState
*bs
,
615 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
),
616 void (*detach_aio_context
)(void *opaque
), void *opaque
);
619 * bdrv_remove_aio_context_notifier:
621 * Unsubscribe of change notifications regarding the BDS's AioContext. The
622 * parameters given here have to be the same as those given to
623 * bdrv_add_aio_context_notifier().
625 void bdrv_remove_aio_context_notifier(BlockDriverState
*bs
,
626 void (*aio_context_attached
)(AioContext
*,
628 void (*aio_context_detached
)(void *),
632 int is_windows_drive(const char *filename
);
637 * @bs: Block device to operate on.
638 * @base: Block device that will become the new base, or %NULL to
639 * flatten the whole backing file chain onto @bs.
640 * @base_id: The file name that will be written to @bs as the new
641 * backing file if the job completes. Ignored if @base is %NULL.
642 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
643 * @on_error: The action to take upon error.
644 * @cb: Completion function for the job.
645 * @opaque: Opaque pointer value passed to @cb.
646 * @errp: Error object.
648 * Start a streaming operation on @bs. Clusters that are unallocated
649 * in @bs, but allocated in any image between @base and @bs (both
650 * exclusive) will be written to @bs. At the end of a successful
651 * streaming job, the backing file of @bs will be changed to
652 * @base_id in the written image and to @base in the live BlockDriverState.
654 void stream_start(BlockDriverState
*bs
, BlockDriverState
*base
,
655 const char *base_id
, int64_t speed
, BlockdevOnError on_error
,
656 BlockCompletionFunc
*cb
,
657 void *opaque
, Error
**errp
);
661 * @bs: Active block device.
662 * @top: Top block device to be committed.
663 * @base: Block device that will be written into, and become the new top.
664 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
665 * @on_error: The action to take upon error.
666 * @cb: Completion function for the job.
667 * @opaque: Opaque pointer value passed to @cb.
668 * @backing_file_str: String to use as the backing file in @top's overlay
669 * @errp: Error object.
672 void commit_start(BlockDriverState
*bs
, BlockDriverState
*base
,
673 BlockDriverState
*top
, int64_t speed
,
674 BlockdevOnError on_error
, BlockCompletionFunc
*cb
,
675 void *opaque
, const char *backing_file_str
, Error
**errp
);
677 * commit_active_start:
678 * @bs: Active block device to be committed.
679 * @base: Block device that will be written into, and become the new top.
680 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
681 * @on_error: The action to take upon error.
682 * @cb: Completion function for the job.
683 * @opaque: Opaque pointer value passed to @cb.
684 * @errp: Error object.
687 void commit_active_start(BlockDriverState
*bs
, BlockDriverState
*base
,
689 BlockdevOnError on_error
,
690 BlockCompletionFunc
*cb
,
691 void *opaque
, Error
**errp
);
694 * @bs: Block device to operate on.
695 * @target: Block device to write to.
696 * @replaces: Block graph node name to replace once the mirror is done. Can
697 * only be used when full mirroring is selected.
698 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
699 * @granularity: The chosen granularity for the dirty bitmap.
700 * @buf_size: The amount of data that can be in flight at one time.
701 * @mode: Whether to collapse all images in the chain to the target.
702 * @backing_mode: How to establish the target's backing chain after completion.
703 * @on_source_error: The action to take upon error reading from the source.
704 * @on_target_error: The action to take upon error writing to the target.
705 * @unmap: Whether to unmap target where source sectors only contain zeroes.
706 * @cb: Completion function for the job.
707 * @opaque: Opaque pointer value passed to @cb.
708 * @errp: Error object.
710 * Start a mirroring operation on @bs. Clusters that are allocated
711 * in @bs will be written to @bs until the job is cancelled or
712 * manually completed. At the end of a successful mirroring job,
713 * @bs will be switched to read from @target.
715 void mirror_start(BlockDriverState
*bs
, BlockDriverState
*target
,
716 const char *replaces
,
717 int64_t speed
, uint32_t granularity
, int64_t buf_size
,
718 MirrorSyncMode mode
, BlockMirrorBackingMode backing_mode
,
719 BlockdevOnError on_source_error
,
720 BlockdevOnError on_target_error
,
722 BlockCompletionFunc
*cb
,
723 void *opaque
, Error
**errp
);
727 * @bs: Block device to operate on.
728 * @target: Block device to write to.
729 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
730 * @sync_mode: What parts of the disk image should be copied to the destination.
731 * @sync_bitmap: The dirty bitmap if sync_mode is MIRROR_SYNC_MODE_INCREMENTAL.
732 * @on_source_error: The action to take upon error reading from the source.
733 * @on_target_error: The action to take upon error writing to the target.
734 * @cb: Completion function for the job.
735 * @opaque: Opaque pointer value passed to @cb.
736 * @txn: Transaction that this job is part of (may be NULL).
738 * Start a backup operation on @bs. Clusters in @bs are written to @target
739 * until the job is cancelled or manually completed.
741 void backup_start(BlockDriverState
*bs
, BlockDriverState
*target
,
742 int64_t speed
, MirrorSyncMode sync_mode
,
743 BdrvDirtyBitmap
*sync_bitmap
,
744 BlockdevOnError on_source_error
,
745 BlockdevOnError on_target_error
,
746 BlockCompletionFunc
*cb
, void *opaque
,
747 BlockJobTxn
*txn
, Error
**errp
);
749 void hmp_drive_add_node(Monitor
*mon
, const char *optstr
);
751 BdrvChild
*bdrv_root_attach_child(BlockDriverState
*child_bs
,
752 const char *child_name
,
753 const BdrvChildRole
*child_role
,
755 void bdrv_root_unref_child(BdrvChild
*child
);
757 const char *bdrv_get_parent_name(const BlockDriverState
*bs
);
758 void blk_dev_change_media_cb(BlockBackend
*blk
, bool load
);
759 bool blk_dev_has_removable_media(BlockBackend
*blk
);
760 bool blk_dev_has_tray(BlockBackend
*blk
);
761 void blk_dev_eject_request(BlockBackend
*blk
, bool force
);
762 bool blk_dev_is_tray_open(BlockBackend
*blk
);
763 bool blk_dev_is_medium_locked(BlockBackend
*blk
);
765 void bdrv_set_dirty(BlockDriverState
*bs
, int64_t cur_sector
, int nr_sectors
);
766 bool bdrv_requests_pending(BlockDriverState
*bs
);
768 void bdrv_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
**out
);
769 void bdrv_undo_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
*in
);
771 void blockdev_close_all_bdrv_states(void);
773 #endif /* BLOCK_INT_H */