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
);
226 int (*bdrv_update
)(BlockDriverState
*bs
, int argc
, char **argv
);
228 int coroutine_fn (*bdrv_save_vmstate
)(BlockDriverState
*bs
,
231 int coroutine_fn (*bdrv_load_vmstate
)(BlockDriverState
*bs
,
235 int (*bdrv_change_backing_file
)(BlockDriverState
*bs
,
236 const char *backing_file
, const char *backing_fmt
);
238 /* removable device specific */
239 bool (*bdrv_is_inserted
)(BlockDriverState
*bs
);
240 int (*bdrv_media_changed
)(BlockDriverState
*bs
);
241 void (*bdrv_eject
)(BlockDriverState
*bs
, bool eject_flag
);
242 void (*bdrv_lock_medium
)(BlockDriverState
*bs
, bool locked
);
244 /* to control generic scsi devices */
245 BlockAIOCB
*(*bdrv_aio_ioctl
)(BlockDriverState
*bs
,
246 unsigned long int req
, void *buf
,
247 BlockCompletionFunc
*cb
, void *opaque
);
249 /* List of options for creating images, terminated by name == NULL */
250 QemuOptsList
*create_opts
;
253 * Returns 0 for completed check, -errno for internal errors.
254 * The check results are stored in result.
256 int (*bdrv_check
)(BlockDriverState
* bs
, BdrvCheckResult
*result
,
259 int (*bdrv_amend_options
)(BlockDriverState
*bs
, QemuOpts
*opts
,
260 BlockDriverAmendStatusCB
*status_cb
,
263 void (*bdrv_debug_event
)(BlockDriverState
*bs
, BlkdebugEvent event
);
265 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
266 int (*bdrv_debug_breakpoint
)(BlockDriverState
*bs
, const char *event
,
268 int (*bdrv_debug_remove_breakpoint
)(BlockDriverState
*bs
,
270 int (*bdrv_debug_resume
)(BlockDriverState
*bs
, const char *tag
);
271 bool (*bdrv_debug_is_suspended
)(BlockDriverState
*bs
, const char *tag
);
273 void (*bdrv_refresh_limits
)(BlockDriverState
*bs
, Error
**errp
);
276 * Returns 1 if newly created images are guaranteed to contain only
277 * zeros, 0 otherwise.
279 int (*bdrv_has_zero_init
)(BlockDriverState
*bs
);
281 /* Remove fd handlers, timers, and other event loop callbacks so the event
282 * loop is no longer in use. Called with no in-flight requests and in
283 * depth-first traversal order with parents before child nodes.
285 void (*bdrv_detach_aio_context
)(BlockDriverState
*bs
);
287 /* Add fd handlers, timers, and other event loop callbacks so I/O requests
288 * can be processed again. Called with no in-flight requests and in
289 * depth-first traversal order with child nodes before parent nodes.
291 void (*bdrv_attach_aio_context
)(BlockDriverState
*bs
,
292 AioContext
*new_context
);
294 /* io queue for linux-aio */
295 void (*bdrv_io_plug
)(BlockDriverState
*bs
);
296 void (*bdrv_io_unplug
)(BlockDriverState
*bs
);
299 * Try to get @bs's logical and physical block size.
300 * On success, store them in @bsz and return zero.
301 * On failure, return negative errno.
303 int (*bdrv_probe_blocksizes
)(BlockDriverState
*bs
, BlockSizes
*bsz
);
305 * Try to get @bs's geometry (cyls, heads, sectors)
306 * On success, store them in @geo and return 0.
307 * On failure return -errno.
308 * Only drivers that want to override guest geometry implement this
309 * callback; see hd_geometry_guess().
311 int (*bdrv_probe_geometry
)(BlockDriverState
*bs
, HDGeometry
*geo
);
314 * Drain and stop any internal sources of requests in the driver, and
315 * remain so until next I/O callback (e.g. bdrv_co_writev) is called.
317 void (*bdrv_drain
)(BlockDriverState
*bs
);
319 void (*bdrv_add_child
)(BlockDriverState
*parent
, BlockDriverState
*child
,
321 void (*bdrv_del_child
)(BlockDriverState
*parent
, BdrvChild
*child
,
324 QLIST_ENTRY(BlockDriver
) list
;
327 typedef struct BlockLimits
{
328 /* Alignment requirement, in bytes, for offset/length of I/O
329 * requests. Must be a power of 2 less than INT_MAX; defaults to
330 * 1 for drivers with modern byte interfaces, and to 512
332 uint32_t request_alignment
;
334 /* maximum number of bytes that can be discarded at once (since it
335 * is signed, it must be < 2G, if set), should be multiple of
336 * pdiscard_alignment, but need not be power of 2. May be 0 if no
337 * inherent 32-bit limit */
338 int32_t max_pdiscard
;
340 /* optimal alignment for discard requests in bytes, must be power
341 * of 2, less than max_pdiscard if that is set, and multiple of
342 * bl.request_alignment. May be 0 if bl.request_alignment is good
344 uint32_t pdiscard_alignment
;
346 /* maximum number of bytes that can zeroized at once (since it is
347 * signed, it must be < 2G, if set), should be multiple of
348 * pwrite_zeroes_alignment. May be 0 if no inherent 32-bit limit */
349 int32_t max_pwrite_zeroes
;
351 /* optimal alignment for write zeroes requests in bytes, must be
352 * power of 2, less than max_pwrite_zeroes if that is set, and
353 * multiple of bl.request_alignment. May be 0 if
354 * bl.request_alignment is good enough */
355 uint32_t pwrite_zeroes_alignment
;
357 /* optimal transfer length in bytes (must be power of 2, and
358 * multiple of bl.request_alignment), or 0 if no preferred size */
359 uint32_t opt_transfer
;
361 /* maximal transfer length in bytes (need not be power of 2, but
362 * should be multiple of opt_transfer), or 0 for no 32-bit limit.
363 * For now, anything larger than INT_MAX is clamped down. */
364 uint32_t max_transfer
;
366 /* memory alignment, in bytes so that no bounce buffer is needed */
367 size_t min_mem_alignment
;
369 /* memory alignment, in bytes, for bounce buffer */
370 size_t opt_mem_alignment
;
372 /* maximum number of iovec elements */
376 typedef struct BdrvOpBlocker BdrvOpBlocker
;
378 typedef struct BdrvAioNotifier
{
379 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
);
380 void (*detach_aio_context
)(void *opaque
);
385 QLIST_ENTRY(BdrvAioNotifier
) list
;
388 struct BdrvChildRole
{
389 void (*inherit_options
)(int *child_flags
, QDict
*child_options
,
390 int parent_flags
, QDict
*parent_options
);
392 void (*change_media
)(BdrvChild
*child
, bool load
);
393 void (*resize
)(BdrvChild
*child
);
395 /* Returns a name that is supposedly more useful for human users than the
396 * node name for identifying the node in question (in particular, a BB
397 * name), or NULL if the parent can't provide a better name. */
398 const char* (*get_name
)(BdrvChild
*child
);
401 * If this pair of functions is implemented, the parent doesn't issue new
402 * requests after returning from .drained_begin() until .drained_end() is
405 * Note that this can be nested. If drained_begin() was called twice, new
406 * I/O is allowed only after drained_end() was called twice, too.
408 void (*drained_begin
)(BdrvChild
*child
);
409 void (*drained_end
)(BdrvChild
*child
);
412 extern const BdrvChildRole child_file
;
413 extern const BdrvChildRole child_format
;
416 BlockDriverState
*bs
;
418 const BdrvChildRole
*role
;
420 QLIST_ENTRY(BdrvChild
) next
;
421 QLIST_ENTRY(BdrvChild
) next_parent
;
425 * Note: the function bdrv_append() copies and swaps contents of
426 * BlockDriverStates, so if you add new fields to this struct, please
427 * inspect bdrv_append() to determine if the new fields need to be
430 struct BlockDriverState
{
431 int64_t total_sectors
; /* if we are reading a disk image, give its
433 int open_flags
; /* flags used to open the file, re-used for re-open */
434 bool read_only
; /* if true, the media is read only */
435 bool encrypted
; /* if true, the media is encrypted */
436 bool valid_key
; /* if true, a valid encryption key has been set */
437 bool sg
; /* if true, the device is a /dev/sg* */
438 bool probed
; /* if true, format was probed rather than specified */
440 int copy_on_read
; /* if nonzero, copy read backing sectors into image.
441 note this is a reference count */
443 BlockDriver
*drv
; /* NULL means no media */
446 AioContext
*aio_context
; /* event loop used for fd handlers, timers, etc */
447 /* long-running tasks intended to always use the same AioContext as this
448 * BDS may register themselves in this list to be notified of changes
449 * regarding this BDS's context */
450 QLIST_HEAD(, BdrvAioNotifier
) aio_notifiers
;
451 bool walking_aio_notifiers
; /* to make removal during iteration safe */
453 char filename
[PATH_MAX
];
454 char backing_file
[PATH_MAX
]; /* if non zero, the image is a diff of
456 char backing_format
[16]; /* if non-zero and backing_file exists */
458 QDict
*full_open_options
;
459 char exact_filename
[PATH_MAX
];
464 /* Callback before write request is processed */
465 NotifierWithReturnList before_write_notifiers
;
467 /* number of in-flight serialising requests */
468 unsigned int serialising_in_flight
;
470 /* Offset after the highest byte written to */
471 uint64_t wr_highest_offset
;
476 /* Flags honored during pwrite (so far: BDRV_REQ_FUA) */
477 unsigned int supported_write_flags
;
478 /* Flags honored during pwrite_zeroes (so far: BDRV_REQ_FUA,
479 * BDRV_REQ_MAY_UNMAP) */
480 unsigned int supported_zero_flags
;
482 /* the following member gives a name to every node on the bs graph. */
484 /* element of the list of named nodes building the graph */
485 QTAILQ_ENTRY(BlockDriverState
) node_list
;
486 /* element of the list of all BlockDriverStates (all_bdrv_states) */
487 QTAILQ_ENTRY(BlockDriverState
) bs_list
;
488 /* element of the list of monitor-owned BDS */
489 QTAILQ_ENTRY(BlockDriverState
) monitor_list
;
490 QLIST_HEAD(, BdrvDirtyBitmap
) dirty_bitmaps
;
493 QLIST_HEAD(, BdrvTrackedRequest
) tracked_requests
;
495 /* operation blockers */
496 QLIST_HEAD(, BdrvOpBlocker
) op_blockers
[BLOCK_OP_TYPE_MAX
];
498 /* long-running background operation */
501 /* The node that this node inherited default options from (and a reopen on
502 * which can affect this node by changing these defaults). This is always a
503 * parent node of this node. */
504 BlockDriverState
*inherits_from
;
505 QLIST_HEAD(, BdrvChild
) children
;
506 QLIST_HEAD(, BdrvChild
) parents
;
509 QDict
*explicit_options
;
510 BlockdevDetectZeroesOptions detect_zeroes
;
512 /* The error object in use for blocking operations on backing_hd */
513 Error
*backing_blocker
;
515 /* threshold limit for writes, in bytes. "High water mark". */
516 uint64_t write_threshold_offset
;
517 NotifierWithReturn write_threshold_notifier
;
519 /* counters for nested bdrv_io_plug and bdrv_io_unplugged_begin */
521 unsigned io_plug_disabled
;
526 struct BlockBackendRootState
{
529 BlockdevDetectZeroesOptions detect_zeroes
;
532 typedef enum BlockMirrorBackingMode
{
533 /* Reuse the existing backing chain from the source for the target.
534 * - sync=full: Set backing BDS to NULL.
535 * - sync=top: Use source's backing BDS.
536 * - sync=none: Use source as the backing BDS. */
537 MIRROR_SOURCE_BACKING_CHAIN
,
539 /* Open the target's backing chain completely anew */
540 MIRROR_OPEN_BACKING_CHAIN
,
542 /* Do not change the target's backing BDS after job completion */
543 MIRROR_LEAVE_BACKING_CHAIN
,
544 } BlockMirrorBackingMode
;
546 static inline BlockDriverState
*backing_bs(BlockDriverState
*bs
)
548 return bs
->backing
? bs
->backing
->bs
: NULL
;
552 /* Essential block drivers which must always be statically linked into qemu, and
553 * which therefore can be accessed without using bdrv_find_format() */
554 extern BlockDriver bdrv_file
;
555 extern BlockDriver bdrv_raw
;
556 extern BlockDriver bdrv_qcow2
;
559 * bdrv_setup_io_funcs:
561 * Prepare a #BlockDriver for I/O request processing by populating
562 * unimplemented coroutine and AIO interfaces with generic wrapper functions
563 * that fall back to implemented interfaces.
565 void bdrv_setup_io_funcs(BlockDriver
*bdrv
);
567 int coroutine_fn
bdrv_co_preadv(BdrvChild
*child
,
568 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
569 BdrvRequestFlags flags
);
570 int coroutine_fn
bdrv_co_pwritev(BdrvChild
*child
,
571 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
572 BdrvRequestFlags flags
);
574 int get_tmp_filename(char *filename
, int size
);
575 BlockDriver
*bdrv_probe_all(const uint8_t *buf
, int buf_size
,
576 const char *filename
);
580 * bdrv_add_before_write_notifier:
582 * Register a callback that is invoked before write requests are processed but
583 * after any throttling or waiting for overlapping requests.
585 void bdrv_add_before_write_notifier(BlockDriverState
*bs
,
586 NotifierWithReturn
*notifier
);
589 * bdrv_detach_aio_context:
591 * May be called from .bdrv_detach_aio_context() to detach children from the
592 * current #AioContext. This is only needed by block drivers that manage their
593 * own children. Both ->file and ->backing are automatically handled and
594 * block drivers should not call this function on them explicitly.
596 void bdrv_detach_aio_context(BlockDriverState
*bs
);
599 * bdrv_attach_aio_context:
601 * May be called from .bdrv_attach_aio_context() to attach children to the new
602 * #AioContext. This is only needed by block drivers that manage their own
603 * children. Both ->file and ->backing are automatically handled and block
604 * drivers should not call this function on them explicitly.
606 void bdrv_attach_aio_context(BlockDriverState
*bs
,
607 AioContext
*new_context
);
610 * bdrv_add_aio_context_notifier:
612 * If a long-running job intends to be always run in the same AioContext as a
613 * certain BDS, it may use this function to be notified of changes regarding the
614 * association of the BDS to an AioContext.
616 * attached_aio_context() is called after the target BDS has been attached to a
617 * new AioContext; detach_aio_context() is called before the target BDS is being
618 * detached from its old AioContext.
620 void bdrv_add_aio_context_notifier(BlockDriverState
*bs
,
621 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
),
622 void (*detach_aio_context
)(void *opaque
), void *opaque
);
625 * bdrv_remove_aio_context_notifier:
627 * Unsubscribe of change notifications regarding the BDS's AioContext. The
628 * parameters given here have to be the same as those given to
629 * bdrv_add_aio_context_notifier().
631 void bdrv_remove_aio_context_notifier(BlockDriverState
*bs
,
632 void (*aio_context_attached
)(AioContext
*,
634 void (*aio_context_detached
)(void *),
638 int is_windows_drive(const char *filename
);
643 * @job_id: The id of the newly-created job, or %NULL to use the
644 * device name of @bs.
645 * @bs: Block device to operate on.
646 * @base: Block device that will become the new base, or %NULL to
647 * flatten the whole backing file chain onto @bs.
648 * @backing_file_str: The file name that will be written to @bs as the
649 * the new backing file if the job completes. Ignored if @base is %NULL.
650 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
651 * @on_error: The action to take upon error.
652 * @cb: Completion function for the job.
653 * @opaque: Opaque pointer value passed to @cb.
654 * @errp: Error object.
656 * Start a streaming operation on @bs. Clusters that are unallocated
657 * in @bs, but allocated in any image between @base and @bs (both
658 * exclusive) will be written to @bs. At the end of a successful
659 * streaming job, the backing file of @bs will be changed to
660 * @backing_file_str in the written image and to @base in the live
663 void stream_start(const char *job_id
, BlockDriverState
*bs
,
664 BlockDriverState
*base
, const char *backing_file_str
,
665 int64_t speed
, BlockdevOnError on_error
,
666 BlockCompletionFunc
*cb
, void *opaque
, Error
**errp
);
670 * @job_id: The id of the newly-created job, or %NULL to use the
671 * device name of @bs.
672 * @bs: Active block device.
673 * @top: Top block device to be committed.
674 * @base: Block device that will be written into, and become the new top.
675 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
676 * @on_error: The action to take upon error.
677 * @cb: Completion function for the job.
678 * @opaque: Opaque pointer value passed to @cb.
679 * @backing_file_str: String to use as the backing file in @top's overlay
680 * @errp: Error object.
683 void commit_start(const char *job_id
, BlockDriverState
*bs
,
684 BlockDriverState
*base
, BlockDriverState
*top
, int64_t speed
,
685 BlockdevOnError on_error
, BlockCompletionFunc
*cb
,
686 void *opaque
, const char *backing_file_str
, Error
**errp
);
688 * commit_active_start:
689 * @job_id: The id of the newly-created job, or %NULL to use the
690 * device name of @bs.
691 * @bs: Active block device to be committed.
692 * @base: Block device that will be written into, and become the new top.
693 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
694 * @on_error: The action to take upon error.
695 * @cb: Completion function for the job.
696 * @opaque: Opaque pointer value passed to @cb.
697 * @errp: Error object.
700 void commit_active_start(const char *job_id
, BlockDriverState
*bs
,
701 BlockDriverState
*base
, int64_t speed
,
702 BlockdevOnError on_error
,
703 BlockCompletionFunc
*cb
,
704 void *opaque
, Error
**errp
);
707 * @job_id: The id of the newly-created job, or %NULL to use the
708 * device name of @bs.
709 * @bs: Block device to operate on.
710 * @target: Block device to write to.
711 * @replaces: Block graph node name to replace once the mirror is done. Can
712 * only be used when full mirroring is selected.
713 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
714 * @granularity: The chosen granularity for the dirty bitmap.
715 * @buf_size: The amount of data that can be in flight at one time.
716 * @mode: Whether to collapse all images in the chain to the target.
717 * @backing_mode: How to establish the target's backing chain after completion.
718 * @on_source_error: The action to take upon error reading from the source.
719 * @on_target_error: The action to take upon error writing to the target.
720 * @unmap: Whether to unmap target where source sectors only contain zeroes.
721 * @cb: Completion function for the job.
722 * @opaque: Opaque pointer value passed to @cb.
723 * @errp: Error object.
725 * Start a mirroring operation on @bs. Clusters that are allocated
726 * in @bs will be written to @bs until the job is cancelled or
727 * manually completed. At the end of a successful mirroring job,
728 * @bs will be switched to read from @target.
730 void mirror_start(const char *job_id
, BlockDriverState
*bs
,
731 BlockDriverState
*target
, const char *replaces
,
732 int64_t speed
, uint32_t granularity
, int64_t buf_size
,
733 MirrorSyncMode mode
, BlockMirrorBackingMode backing_mode
,
734 BlockdevOnError on_source_error
,
735 BlockdevOnError on_target_error
,
737 BlockCompletionFunc
*cb
,
738 void *opaque
, Error
**errp
);
742 * @job_id: The id of the newly-created job, or %NULL to use the
743 * device name of @bs.
744 * @bs: Block device to operate on.
745 * @target: Block device to write to.
746 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
747 * @sync_mode: What parts of the disk image should be copied to the destination.
748 * @sync_bitmap: The dirty bitmap if sync_mode is MIRROR_SYNC_MODE_INCREMENTAL.
749 * @on_source_error: The action to take upon error reading from the source.
750 * @on_target_error: The action to take upon error writing to the target.
751 * @cb: Completion function for the job.
752 * @opaque: Opaque pointer value passed to @cb.
753 * @txn: Transaction that this job is part of (may be NULL).
755 * Start a backup operation on @bs. Clusters in @bs are written to @target
756 * until the job is cancelled or manually completed.
758 void backup_start(const char *job_id
, BlockDriverState
*bs
,
759 BlockDriverState
*target
, int64_t speed
,
760 MirrorSyncMode sync_mode
, BdrvDirtyBitmap
*sync_bitmap
,
761 BlockdevOnError on_source_error
,
762 BlockdevOnError on_target_error
,
763 BlockCompletionFunc
*cb
, void *opaque
,
764 BlockJobTxn
*txn
, Error
**errp
);
766 void hmp_drive_add_node(Monitor
*mon
, const char *optstr
);
768 BdrvChild
*bdrv_root_attach_child(BlockDriverState
*child_bs
,
769 const char *child_name
,
770 const BdrvChildRole
*child_role
,
772 void bdrv_root_unref_child(BdrvChild
*child
);
774 const char *bdrv_get_parent_name(const BlockDriverState
*bs
);
775 void blk_dev_change_media_cb(BlockBackend
*blk
, bool load
);
776 bool blk_dev_has_removable_media(BlockBackend
*blk
);
777 bool blk_dev_has_tray(BlockBackend
*blk
);
778 void blk_dev_eject_request(BlockBackend
*blk
, bool force
);
779 bool blk_dev_is_tray_open(BlockBackend
*blk
);
780 bool blk_dev_is_medium_locked(BlockBackend
*blk
);
782 void bdrv_set_dirty(BlockDriverState
*bs
, int64_t cur_sector
, int nr_sectors
);
783 bool bdrv_requests_pending(BlockDriverState
*bs
);
785 void bdrv_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
**out
);
786 void bdrv_undo_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
*in
);
788 void blockdev_close_all_bdrv_states(void);
790 #endif /* BLOCK_INT_H */