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
{
68 typedef struct BdrvTrackedRequest
{
72 enum BdrvTrackedRequestType type
;
75 int64_t overlap_offset
;
76 unsigned int overlap_bytes
;
78 QLIST_ENTRY(BdrvTrackedRequest
) list
;
79 Coroutine
*co
; /* owner, used for deadlock detection */
80 CoQueue wait_queue
; /* coroutines blocked on this request */
82 struct BdrvTrackedRequest
*waiting_for
;
86 const char *format_name
;
89 /* set to true if the BlockDriver is a block filter */
91 /* for snapshots block filter like Quorum can implement the
92 * following recursive callback.
93 * It's purpose is to recurse on the filter children while calling
94 * bdrv_recurse_is_first_non_filter on them.
95 * For a sample implementation look in the future Quorum block filter.
97 bool (*bdrv_recurse_is_first_non_filter
)(BlockDriverState
*bs
,
98 BlockDriverState
*candidate
);
100 int (*bdrv_probe
)(const uint8_t *buf
, int buf_size
, const char *filename
);
101 int (*bdrv_probe_device
)(const char *filename
);
103 /* Any driver implementing this callback is expected to be able to handle
104 * NULL file names in its .bdrv_open() implementation */
105 void (*bdrv_parse_filename
)(const char *filename
, QDict
*options
, Error
**errp
);
106 /* Drivers not implementing bdrv_parse_filename nor bdrv_open should have
107 * this field set to true, except ones that are defined only by their
109 * An example of the last type will be the quorum block driver.
111 bool bdrv_needs_filename
;
113 /* Set if a driver can support backing files */
114 bool supports_backing
;
116 /* For handling image reopen for split or non-split files */
117 int (*bdrv_reopen_prepare
)(BDRVReopenState
*reopen_state
,
118 BlockReopenQueue
*queue
, Error
**errp
);
119 void (*bdrv_reopen_commit
)(BDRVReopenState
*reopen_state
);
120 void (*bdrv_reopen_abort
)(BDRVReopenState
*reopen_state
);
121 void (*bdrv_join_options
)(QDict
*options
, QDict
*old_options
);
123 int (*bdrv_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
125 int (*bdrv_file_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
127 void (*bdrv_close
)(BlockDriverState
*bs
);
128 int (*bdrv_create
)(const char *filename
, QemuOpts
*opts
, Error
**errp
);
129 int (*bdrv_set_key
)(BlockDriverState
*bs
, const char *key
);
130 int (*bdrv_make_empty
)(BlockDriverState
*bs
);
132 void (*bdrv_refresh_filename
)(BlockDriverState
*bs
, QDict
*options
);
135 BlockAIOCB
*(*bdrv_aio_readv
)(BlockDriverState
*bs
,
136 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
137 BlockCompletionFunc
*cb
, void *opaque
);
138 BlockAIOCB
*(*bdrv_aio_writev
)(BlockDriverState
*bs
,
139 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
140 BlockCompletionFunc
*cb
, void *opaque
);
141 BlockAIOCB
*(*bdrv_aio_flush
)(BlockDriverState
*bs
,
142 BlockCompletionFunc
*cb
, void *opaque
);
143 BlockAIOCB
*(*bdrv_aio_pdiscard
)(BlockDriverState
*bs
,
144 int64_t offset
, int count
,
145 BlockCompletionFunc
*cb
, void *opaque
);
147 int coroutine_fn (*bdrv_co_readv
)(BlockDriverState
*bs
,
148 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
149 int coroutine_fn (*bdrv_co_preadv
)(BlockDriverState
*bs
,
150 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
151 int coroutine_fn (*bdrv_co_writev
)(BlockDriverState
*bs
,
152 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
153 int coroutine_fn (*bdrv_co_writev_flags
)(BlockDriverState
*bs
,
154 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
, int flags
);
155 int coroutine_fn (*bdrv_co_pwritev
)(BlockDriverState
*bs
,
156 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
159 * Efficiently zero a region of the disk image. Typically an image format
160 * would use a compact metadata representation to implement this. This
161 * function pointer may be NULL or return -ENOSUP and .bdrv_co_writev()
162 * will be called instead.
164 int coroutine_fn (*bdrv_co_pwrite_zeroes
)(BlockDriverState
*bs
,
165 int64_t offset
, int count
, BdrvRequestFlags flags
);
166 int coroutine_fn (*bdrv_co_pdiscard
)(BlockDriverState
*bs
,
167 int64_t offset
, int count
);
168 int64_t coroutine_fn (*bdrv_co_get_block_status
)(BlockDriverState
*bs
,
169 int64_t sector_num
, int nb_sectors
, int *pnum
,
170 BlockDriverState
**file
);
173 * Invalidate any cached meta-data.
175 void (*bdrv_invalidate_cache
)(BlockDriverState
*bs
, Error
**errp
);
176 int (*bdrv_inactivate
)(BlockDriverState
*bs
);
179 * Flushes all data for all layers by calling bdrv_co_flush for underlying
180 * layers, if needed. This function is needed for deterministic
181 * synchronization of the flush finishing callback.
183 int coroutine_fn (*bdrv_co_flush
)(BlockDriverState
*bs
);
186 * Flushes all data that was already written to the OS all the way down to
187 * the disk (for example file-posix.c calls fsync()).
189 int coroutine_fn (*bdrv_co_flush_to_disk
)(BlockDriverState
*bs
);
192 * Flushes all internal caches to the OS. The data may still sit in a
193 * writeback cache of the host OS, but it will survive a crash of the qemu
196 int coroutine_fn (*bdrv_co_flush_to_os
)(BlockDriverState
*bs
);
198 const char *protocol_name
;
199 int (*bdrv_truncate
)(BlockDriverState
*bs
, int64_t offset
);
201 int64_t (*bdrv_getlength
)(BlockDriverState
*bs
);
202 bool has_variable_length
;
203 int64_t (*bdrv_get_allocated_file_size
)(BlockDriverState
*bs
);
205 int coroutine_fn (*bdrv_co_pwritev_compressed
)(BlockDriverState
*bs
,
206 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
);
208 int (*bdrv_snapshot_create
)(BlockDriverState
*bs
,
209 QEMUSnapshotInfo
*sn_info
);
210 int (*bdrv_snapshot_goto
)(BlockDriverState
*bs
,
211 const char *snapshot_id
);
212 int (*bdrv_snapshot_delete
)(BlockDriverState
*bs
,
213 const char *snapshot_id
,
216 int (*bdrv_snapshot_list
)(BlockDriverState
*bs
,
217 QEMUSnapshotInfo
**psn_info
);
218 int (*bdrv_snapshot_load_tmp
)(BlockDriverState
*bs
,
219 const char *snapshot_id
,
222 int (*bdrv_get_info
)(BlockDriverState
*bs
, BlockDriverInfo
*bdi
);
223 ImageInfoSpecific
*(*bdrv_get_specific_info
)(BlockDriverState
*bs
);
225 int coroutine_fn (*bdrv_save_vmstate
)(BlockDriverState
*bs
,
228 int coroutine_fn (*bdrv_load_vmstate
)(BlockDriverState
*bs
,
232 int (*bdrv_change_backing_file
)(BlockDriverState
*bs
,
233 const char *backing_file
, const char *backing_fmt
);
235 /* removable device specific */
236 bool (*bdrv_is_inserted
)(BlockDriverState
*bs
);
237 int (*bdrv_media_changed
)(BlockDriverState
*bs
);
238 void (*bdrv_eject
)(BlockDriverState
*bs
, bool eject_flag
);
239 void (*bdrv_lock_medium
)(BlockDriverState
*bs
, bool locked
);
241 /* to control generic scsi devices */
242 BlockAIOCB
*(*bdrv_aio_ioctl
)(BlockDriverState
*bs
,
243 unsigned long int req
, void *buf
,
244 BlockCompletionFunc
*cb
, void *opaque
);
245 int coroutine_fn (*bdrv_co_ioctl
)(BlockDriverState
*bs
,
246 unsigned long int req
, void *buf
);
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 /* Alignment requirement, in bytes, for offset/length of I/O
328 * requests. Must be a power of 2 less than INT_MAX; defaults to
329 * 1 for drivers with modern byte interfaces, and to 512
331 uint32_t request_alignment
;
333 /* Maximum number of bytes that can be discarded at once (since it
334 * is signed, it must be < 2G, if set). Must be multiple of
335 * pdiscard_alignment, but need not be power of 2. May be 0 if no
336 * inherent 32-bit limit */
337 int32_t max_pdiscard
;
339 /* Optimal alignment for discard requests in bytes. A power of 2
340 * is best but not mandatory. Must be a multiple of
341 * bl.request_alignment, and must be less than max_pdiscard if
342 * that is set. May be 0 if bl.request_alignment is good enough */
343 uint32_t pdiscard_alignment
;
345 /* Maximum number of bytes that can zeroized at once (since it is
346 * signed, it must be < 2G, if set). Must be multiple of
347 * pwrite_zeroes_alignment. May be 0 if no inherent 32-bit limit */
348 int32_t max_pwrite_zeroes
;
350 /* Optimal alignment for write zeroes requests in bytes. A power
351 * of 2 is best but not mandatory. Must be a multiple of
352 * bl.request_alignment, and must be less than max_pwrite_zeroes
353 * if that is set. May be 0 if bl.request_alignment is good
355 uint32_t pwrite_zeroes_alignment
;
357 /* Optimal transfer length in bytes. A power of 2 is best but not
358 * mandatory. Must be a multiple of bl.request_alignment, or 0 if
359 * no preferred size */
360 uint32_t opt_transfer
;
362 /* Maximal transfer length in bytes. Need not be power of 2, but
363 * must be multiple of opt_transfer and bl.request_alignment, or 0
364 * for no 32-bit limit. For now, anything larger than INT_MAX is
366 uint32_t max_transfer
;
368 /* memory alignment, in bytes so that no bounce buffer is needed */
369 size_t min_mem_alignment
;
371 /* memory alignment, in bytes, for bounce buffer */
372 size_t opt_mem_alignment
;
374 /* maximum number of iovec elements */
378 typedef struct BdrvOpBlocker BdrvOpBlocker
;
380 typedef struct BdrvAioNotifier
{
381 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
);
382 void (*detach_aio_context
)(void *opaque
);
387 QLIST_ENTRY(BdrvAioNotifier
) list
;
390 struct BdrvChildRole
{
391 void (*inherit_options
)(int *child_flags
, QDict
*child_options
,
392 int parent_flags
, QDict
*parent_options
);
394 void (*change_media
)(BdrvChild
*child
, bool load
);
395 void (*resize
)(BdrvChild
*child
);
397 /* Returns a name that is supposedly more useful for human users than the
398 * node name for identifying the node in question (in particular, a BB
399 * name), or NULL if the parent can't provide a better name. */
400 const char* (*get_name
)(BdrvChild
*child
);
403 * If this pair of functions is implemented, the parent doesn't issue new
404 * requests after returning from .drained_begin() until .drained_end() is
407 * Note that this can be nested. If drained_begin() was called twice, new
408 * I/O is allowed only after drained_end() was called twice, too.
410 void (*drained_begin
)(BdrvChild
*child
);
411 void (*drained_end
)(BdrvChild
*child
);
414 extern const BdrvChildRole child_file
;
415 extern const BdrvChildRole child_format
;
418 BlockDriverState
*bs
;
420 const BdrvChildRole
*role
;
422 QLIST_ENTRY(BdrvChild
) next
;
423 QLIST_ENTRY(BdrvChild
) next_parent
;
427 * Note: the function bdrv_append() copies and swaps contents of
428 * BlockDriverStates, so if you add new fields to this struct, please
429 * inspect bdrv_append() to determine if the new fields need to be
432 struct BlockDriverState
{
433 int64_t total_sectors
; /* if we are reading a disk image, give its
435 int open_flags
; /* flags used to open the file, re-used for re-open */
436 bool read_only
; /* if true, the media is read only */
437 bool encrypted
; /* if true, the media is encrypted */
438 bool valid_key
; /* if true, a valid encryption key has been set */
439 bool sg
; /* if true, the device is a /dev/sg* */
440 bool probed
; /* if true, format was probed rather than specified */
442 int copy_on_read
; /* if nonzero, copy read backing sectors into image.
443 note this is a reference count */
445 CoQueue flush_queue
; /* Serializing flush queue */
446 bool active_flush_req
; /* Flush request in flight? */
447 unsigned int write_gen
; /* Current data generation */
448 unsigned int flushed_gen
; /* Flushed write generation */
450 BlockDriver
*drv
; /* NULL means no media */
453 AioContext
*aio_context
; /* event loop used for fd handlers, timers, etc */
454 /* long-running tasks intended to always use the same AioContext as this
455 * BDS may register themselves in this list to be notified of changes
456 * regarding this BDS's context */
457 QLIST_HEAD(, BdrvAioNotifier
) aio_notifiers
;
458 bool walking_aio_notifiers
; /* to make removal during iteration safe */
460 char filename
[PATH_MAX
];
461 char backing_file
[PATH_MAX
]; /* if non zero, the image is a diff of
463 char backing_format
[16]; /* if non-zero and backing_file exists */
465 QDict
*full_open_options
;
466 char exact_filename
[PATH_MAX
];
471 /* Callback before write request is processed */
472 NotifierWithReturnList before_write_notifiers
;
474 /* number of in-flight requests; overall and serialising */
475 unsigned int in_flight
;
476 unsigned int serialising_in_flight
;
480 /* Offset after the highest byte written to */
481 uint64_t wr_highest_offset
;
486 /* Flags honored during pwrite (so far: BDRV_REQ_FUA) */
487 unsigned int supported_write_flags
;
488 /* Flags honored during pwrite_zeroes (so far: BDRV_REQ_FUA,
489 * BDRV_REQ_MAY_UNMAP) */
490 unsigned int supported_zero_flags
;
492 /* the following member gives a name to every node on the bs graph. */
494 /* element of the list of named nodes building the graph */
495 QTAILQ_ENTRY(BlockDriverState
) node_list
;
496 /* element of the list of all BlockDriverStates (all_bdrv_states) */
497 QTAILQ_ENTRY(BlockDriverState
) bs_list
;
498 /* element of the list of monitor-owned BDS */
499 QTAILQ_ENTRY(BlockDriverState
) monitor_list
;
500 QLIST_HEAD(, BdrvDirtyBitmap
) dirty_bitmaps
;
503 QLIST_HEAD(, BdrvTrackedRequest
) tracked_requests
;
505 /* operation blockers */
506 QLIST_HEAD(, BdrvOpBlocker
) op_blockers
[BLOCK_OP_TYPE_MAX
];
508 /* long-running background operation */
511 /* The node that this node inherited default options from (and a reopen on
512 * which can affect this node by changing these defaults). This is always a
513 * parent node of this node. */
514 BlockDriverState
*inherits_from
;
515 QLIST_HEAD(, BdrvChild
) children
;
516 QLIST_HEAD(, BdrvChild
) parents
;
519 QDict
*explicit_options
;
520 BlockdevDetectZeroesOptions detect_zeroes
;
522 /* The error object in use for blocking operations on backing_hd */
523 Error
*backing_blocker
;
525 /* threshold limit for writes, in bytes. "High water mark". */
526 uint64_t write_threshold_offset
;
527 NotifierWithReturn write_threshold_notifier
;
529 /* counter for nested bdrv_io_plug */
535 struct BlockBackendRootState
{
538 BlockdevDetectZeroesOptions detect_zeroes
;
541 typedef enum BlockMirrorBackingMode
{
542 /* Reuse the existing backing chain from the source for the target.
543 * - sync=full: Set backing BDS to NULL.
544 * - sync=top: Use source's backing BDS.
545 * - sync=none: Use source as the backing BDS. */
546 MIRROR_SOURCE_BACKING_CHAIN
,
548 /* Open the target's backing chain completely anew */
549 MIRROR_OPEN_BACKING_CHAIN
,
551 /* Do not change the target's backing BDS after job completion */
552 MIRROR_LEAVE_BACKING_CHAIN
,
553 } BlockMirrorBackingMode
;
555 static inline BlockDriverState
*backing_bs(BlockDriverState
*bs
)
557 return bs
->backing
? bs
->backing
->bs
: NULL
;
561 /* Essential block drivers which must always be statically linked into qemu, and
562 * which therefore can be accessed without using bdrv_find_format() */
563 extern BlockDriver bdrv_file
;
564 extern BlockDriver bdrv_raw
;
565 extern BlockDriver bdrv_qcow2
;
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 *),
639 * @bs: The BlockDriverState for which an I/O operation has been completed.
641 * Wake up the main thread if it is waiting on BDRV_POLL_WHILE. During
642 * synchronous I/O on a BlockDriverState that is attached to another
643 * I/O thread, the main thread lets the I/O thread's event loop run,
644 * waiting for the I/O operation to complete. A bdrv_wakeup will wake
645 * up the main thread if necessary.
647 * Manual calls to bdrv_wakeup are rarely necessary, because
648 * bdrv_dec_in_flight already calls it.
650 void bdrv_wakeup(BlockDriverState
*bs
);
653 int is_windows_drive(const char *filename
);
658 * @job_id: The id of the newly-created job, or %NULL to use the
659 * device name of @bs.
660 * @bs: Block device to operate on.
661 * @base: Block device that will become the new base, or %NULL to
662 * flatten the whole backing file chain onto @bs.
663 * @backing_file_str: The file name that will be written to @bs as the
664 * the new backing file if the job completes. Ignored if @base is %NULL.
665 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
666 * @on_error: The action to take upon error.
667 * @errp: Error object.
669 * Start a streaming operation on @bs. Clusters that are unallocated
670 * in @bs, but allocated in any image between @base and @bs (both
671 * exclusive) will be written to @bs. At the end of a successful
672 * streaming job, the backing file of @bs will be changed to
673 * @backing_file_str in the written image and to @base in the live
676 void stream_start(const char *job_id
, BlockDriverState
*bs
,
677 BlockDriverState
*base
, const char *backing_file_str
,
678 int64_t speed
, BlockdevOnError on_error
, Error
**errp
);
682 * @job_id: The id of the newly-created job, or %NULL to use the
683 * device name of @bs.
684 * @bs: Active block device.
685 * @top: Top block device to be committed.
686 * @base: Block device that will be written into, and become the new top.
687 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
688 * @on_error: The action to take upon error.
689 * @backing_file_str: String to use as the backing file in @top's overlay
690 * @errp: Error object.
693 void commit_start(const char *job_id
, BlockDriverState
*bs
,
694 BlockDriverState
*base
, BlockDriverState
*top
, int64_t speed
,
695 BlockdevOnError on_error
, const char *backing_file_str
,
698 * commit_active_start:
699 * @job_id: The id of the newly-created job, or %NULL to use the
700 * device name of @bs.
701 * @bs: Active block device to be committed.
702 * @base: Block device that will be written into, and become the new top.
703 * @creation_flags: Flags that control the behavior of the Job lifetime.
704 * See @BlockJobCreateFlags
705 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
706 * @on_error: The action to take upon error.
707 * @cb: Completion function for the job.
708 * @opaque: Opaque pointer value passed to @cb.
709 * @errp: Error object.
710 * @auto_complete: Auto complete the job.
713 void commit_active_start(const char *job_id
, BlockDriverState
*bs
,
714 BlockDriverState
*base
, int creation_flags
,
715 int64_t speed
, BlockdevOnError on_error
,
716 BlockCompletionFunc
*cb
,
717 void *opaque
, Error
**errp
, bool auto_complete
);
720 * @job_id: The id of the newly-created job, or %NULL to use the
721 * device name of @bs.
722 * @bs: Block device to operate on.
723 * @target: Block device to write to.
724 * @replaces: Block graph node name to replace once the mirror is done. Can
725 * only be used when full mirroring is selected.
726 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
727 * @granularity: The chosen granularity for the dirty bitmap.
728 * @buf_size: The amount of data that can be in flight at one time.
729 * @mode: Whether to collapse all images in the chain to the target.
730 * @backing_mode: How to establish the target's backing chain after completion.
731 * @on_source_error: The action to take upon error reading from the source.
732 * @on_target_error: The action to take upon error writing to the target.
733 * @unmap: Whether to unmap target where source sectors only contain zeroes.
734 * @errp: Error object.
736 * Start a mirroring operation on @bs. Clusters that are allocated
737 * in @bs will be written to @target until the job is cancelled or
738 * manually completed. At the end of a successful mirroring job,
739 * @bs will be switched to read from @target.
741 void mirror_start(const char *job_id
, BlockDriverState
*bs
,
742 BlockDriverState
*target
, const char *replaces
,
743 int64_t speed
, uint32_t granularity
, int64_t buf_size
,
744 MirrorSyncMode mode
, BlockMirrorBackingMode backing_mode
,
745 BlockdevOnError on_source_error
,
746 BlockdevOnError on_target_error
,
747 bool unmap
, Error
**errp
);
751 * @job_id: The id of the newly-created job, or %NULL to use the
752 * device name of @bs.
753 * @bs: Block device to operate on.
754 * @target: Block device to write to.
755 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
756 * @sync_mode: What parts of the disk image should be copied to the destination.
757 * @sync_bitmap: The dirty bitmap if sync_mode is MIRROR_SYNC_MODE_INCREMENTAL.
758 * @on_source_error: The action to take upon error reading from the source.
759 * @on_target_error: The action to take upon error writing to the target.
760 * @creation_flags: Flags that control the behavior of the Job lifetime.
761 * See @BlockJobCreateFlags
762 * @cb: Completion function for the job.
763 * @opaque: Opaque pointer value passed to @cb.
764 * @txn: Transaction that this job is part of (may be NULL).
766 * Create a backup operation on @bs. Clusters in @bs are written to @target
767 * until the job is cancelled or manually completed.
769 BlockJob
*backup_job_create(const char *job_id
, BlockDriverState
*bs
,
770 BlockDriverState
*target
, int64_t speed
,
771 MirrorSyncMode sync_mode
,
772 BdrvDirtyBitmap
*sync_bitmap
,
774 BlockdevOnError on_source_error
,
775 BlockdevOnError on_target_error
,
777 BlockCompletionFunc
*cb
, void *opaque
,
778 BlockJobTxn
*txn
, Error
**errp
);
780 void hmp_drive_add_node(Monitor
*mon
, const char *optstr
);
782 BdrvChild
*bdrv_root_attach_child(BlockDriverState
*child_bs
,
783 const char *child_name
,
784 const BdrvChildRole
*child_role
,
786 void bdrv_root_unref_child(BdrvChild
*child
);
788 const char *bdrv_get_parent_name(const BlockDriverState
*bs
);
789 void blk_dev_change_media_cb(BlockBackend
*blk
, bool load
);
790 bool blk_dev_has_removable_media(BlockBackend
*blk
);
791 bool blk_dev_has_tray(BlockBackend
*blk
);
792 void blk_dev_eject_request(BlockBackend
*blk
, bool force
);
793 bool blk_dev_is_tray_open(BlockBackend
*blk
);
794 bool blk_dev_is_medium_locked(BlockBackend
*blk
);
796 void bdrv_set_dirty(BlockDriverState
*bs
, int64_t cur_sector
, int64_t nr_sect
);
797 bool bdrv_requests_pending(BlockDriverState
*bs
);
799 void bdrv_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
**out
);
800 void bdrv_undo_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
*in
);
802 void bdrv_inc_in_flight(BlockDriverState
*bs
);
803 void bdrv_dec_in_flight(BlockDriverState
*bs
);
805 void blockdev_close_all_bdrv_states(void);
807 #endif /* BLOCK_INT_H */