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 "block/aio-wait.h"
30 #include "qemu/queue.h"
31 #include "qemu/coroutine.h"
32 #include "qemu/stats64.h"
33 #include "qemu/timer.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_LAZY_REFCOUNTS 8
41 #define BLOCK_OPT_SIZE "size"
42 #define BLOCK_OPT_ENCRYPT "encryption"
43 #define BLOCK_OPT_ENCRYPT_FORMAT "encrypt.format"
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"
59 #define BLOCK_OPT_DATA_FILE "data_file"
60 #define BLOCK_OPT_DATA_FILE_RAW "data_file_raw"
62 #define BLOCK_PROBE_BUF_SIZE 512
64 enum BdrvTrackedRequestType
{
68 BDRV_TRACKED_TRUNCATE
,
71 typedef struct BdrvTrackedRequest
{
75 enum BdrvTrackedRequestType type
;
78 int64_t overlap_offset
;
79 uint64_t overlap_bytes
;
81 QLIST_ENTRY(BdrvTrackedRequest
) list
;
82 Coroutine
*co
; /* owner, used for deadlock detection */
83 CoQueue wait_queue
; /* coroutines blocked on this request */
85 struct BdrvTrackedRequest
*waiting_for
;
89 const char *format_name
;
92 /* set to true if the BlockDriver is a block filter. Block filters pass
93 * certain callbacks that refer to data (see block.c) to their bs->file if
94 * the driver doesn't implement them. Drivers that do not wish to forward
95 * must implement them and return -ENOTSUP.
98 /* for snapshots block filter like Quorum can implement the
99 * following recursive callback.
100 * It's purpose is to recurse on the filter children while calling
101 * bdrv_recurse_is_first_non_filter on them.
102 * For a sample implementation look in the future Quorum block filter.
104 bool (*bdrv_recurse_is_first_non_filter
)(BlockDriverState
*bs
,
105 BlockDriverState
*candidate
);
107 int (*bdrv_probe
)(const uint8_t *buf
, int buf_size
, const char *filename
);
108 int (*bdrv_probe_device
)(const char *filename
);
110 /* Any driver implementing this callback is expected to be able to handle
111 * NULL file names in its .bdrv_open() implementation */
112 void (*bdrv_parse_filename
)(const char *filename
, QDict
*options
, Error
**errp
);
113 /* Drivers not implementing bdrv_parse_filename nor bdrv_open should have
114 * this field set to true, except ones that are defined only by their
116 * An example of the last type will be the quorum block driver.
118 bool bdrv_needs_filename
;
120 /* Set if a driver can support backing files */
121 bool supports_backing
;
123 /* For handling image reopen for split or non-split files */
124 int (*bdrv_reopen_prepare
)(BDRVReopenState
*reopen_state
,
125 BlockReopenQueue
*queue
, Error
**errp
);
126 void (*bdrv_reopen_commit
)(BDRVReopenState
*reopen_state
);
127 void (*bdrv_reopen_abort
)(BDRVReopenState
*reopen_state
);
128 void (*bdrv_join_options
)(QDict
*options
, QDict
*old_options
);
130 int (*bdrv_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
133 /* Protocol drivers should implement this instead of bdrv_open */
134 int (*bdrv_file_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
,
136 void (*bdrv_close
)(BlockDriverState
*bs
);
137 int coroutine_fn (*bdrv_co_create
)(BlockdevCreateOptions
*opts
,
139 int coroutine_fn (*bdrv_co_create_opts
)(const char *filename
,
142 int (*bdrv_make_empty
)(BlockDriverState
*bs
);
145 * Refreshes the bs->exact_filename field. If that is impossible,
146 * bs->exact_filename has to be left empty.
148 void (*bdrv_refresh_filename
)(BlockDriverState
*bs
);
151 * Gathers the open options for all children into @target.
152 * A simple format driver (without backing file support) might
153 * implement this function like this:
155 * QINCREF(bs->file->bs->full_open_options);
156 * qdict_put(target, "file", bs->file->bs->full_open_options);
158 * If not specified, the generic implementation will simply put
159 * all children's options under their respective name.
161 * @backing_overridden is true when bs->backing seems not to be
162 * the child that would result from opening bs->backing_file.
163 * Therefore, if it is true, the backing child's options should be
164 * gathered; otherwise, there is no need since the backing child
165 * is the one implied by the image header.
167 * Note that ideally this function would not be needed. Every
168 * block driver which implements it is probably doing something
169 * shady regarding its runtime option structure.
171 void (*bdrv_gather_child_options
)(BlockDriverState
*bs
, QDict
*target
,
172 bool backing_overridden
);
175 * Returns an allocated string which is the directory name of this BDS: It
176 * will be used to make relative filenames absolute by prepending this
177 * function's return value to them.
179 char *(*bdrv_dirname
)(BlockDriverState
*bs
, Error
**errp
);
182 BlockAIOCB
*(*bdrv_aio_preadv
)(BlockDriverState
*bs
,
183 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
,
184 BlockCompletionFunc
*cb
, void *opaque
);
185 BlockAIOCB
*(*bdrv_aio_pwritev
)(BlockDriverState
*bs
,
186 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
,
187 BlockCompletionFunc
*cb
, void *opaque
);
188 BlockAIOCB
*(*bdrv_aio_flush
)(BlockDriverState
*bs
,
189 BlockCompletionFunc
*cb
, void *opaque
);
190 BlockAIOCB
*(*bdrv_aio_pdiscard
)(BlockDriverState
*bs
,
191 int64_t offset
, int bytes
,
192 BlockCompletionFunc
*cb
, void *opaque
);
194 int coroutine_fn (*bdrv_co_readv
)(BlockDriverState
*bs
,
195 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
198 * @offset: position in bytes to read at
199 * @bytes: number of bytes to read
200 * @qiov: the buffers to fill with read data
201 * @flags: currently unused, always 0
203 * @offset and @bytes will be a multiple of 'request_alignment',
204 * but the length of individual @qiov elements does not have to
207 * @bytes will always equal the total size of @qiov, and will be
208 * no larger than 'max_transfer'.
210 * The buffer in @qiov may point directly to guest memory.
212 int coroutine_fn (*bdrv_co_preadv
)(BlockDriverState
*bs
,
213 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
214 int coroutine_fn (*bdrv_co_writev
)(BlockDriverState
*bs
,
215 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
, int flags
);
217 * @offset: position in bytes to write at
218 * @bytes: number of bytes to write
219 * @qiov: the buffers containing data to write
220 * @flags: zero or more bits allowed by 'supported_write_flags'
222 * @offset and @bytes will be a multiple of 'request_alignment',
223 * but the length of individual @qiov elements does not have to
226 * @bytes will always equal the total size of @qiov, and will be
227 * no larger than 'max_transfer'.
229 * The buffer in @qiov may point directly to guest memory.
231 int coroutine_fn (*bdrv_co_pwritev
)(BlockDriverState
*bs
,
232 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
, int flags
);
235 * Efficiently zero a region of the disk image. Typically an image format
236 * would use a compact metadata representation to implement this. This
237 * function pointer may be NULL or return -ENOSUP and .bdrv_co_writev()
238 * will be called instead.
240 int coroutine_fn (*bdrv_co_pwrite_zeroes
)(BlockDriverState
*bs
,
241 int64_t offset
, int bytes
, BdrvRequestFlags flags
);
242 int coroutine_fn (*bdrv_co_pdiscard
)(BlockDriverState
*bs
,
243 int64_t offset
, int bytes
);
245 /* Map [offset, offset + nbytes) range onto a child of @bs to copy from,
246 * and invoke bdrv_co_copy_range_from(child, ...), or invoke
247 * bdrv_co_copy_range_to() if @bs is the leaf child to copy data from.
249 * See the comment of bdrv_co_copy_range for the parameter and return value
252 int coroutine_fn (*bdrv_co_copy_range_from
)(BlockDriverState
*bs
,
258 BdrvRequestFlags read_flags
,
259 BdrvRequestFlags write_flags
);
261 /* Map [offset, offset + nbytes) range onto a child of bs to copy data to,
262 * and invoke bdrv_co_copy_range_to(child, src, ...), or perform the copy
263 * operation if @bs is the leaf and @src has the same BlockDriver. Return
264 * -ENOTSUP if @bs is the leaf but @src has a different BlockDriver.
266 * See the comment of bdrv_co_copy_range for the parameter and return value
269 int coroutine_fn (*bdrv_co_copy_range_to
)(BlockDriverState
*bs
,
275 BdrvRequestFlags read_flags
,
276 BdrvRequestFlags write_flags
);
279 * Building block for bdrv_block_status[_above] and
280 * bdrv_is_allocated[_above]. The driver should answer only
281 * according to the current layer, and should only need to set
282 * BDRV_BLOCK_DATA, BDRV_BLOCK_ZERO, BDRV_BLOCK_OFFSET_VALID,
283 * and/or BDRV_BLOCK_RAW; if the current layer defers to a backing
284 * layer, the result should be 0 (and not BDRV_BLOCK_ZERO). See
285 * block.h for the overall meaning of the bits. As a hint, the
286 * flag want_zero is true if the caller cares more about precise
287 * mappings (favor accurate _OFFSET_VALID/_ZERO) or false for
288 * overall allocation (favor larger *pnum, perhaps by reporting
289 * _DATA instead of _ZERO). The block layer guarantees input
290 * clamped to bdrv_getlength() and aligned to request_alignment,
291 * as well as non-NULL pnum, map, and file; in turn, the driver
292 * must return an error or set pnum to an aligned non-zero value.
294 int coroutine_fn (*bdrv_co_block_status
)(BlockDriverState
*bs
,
295 bool want_zero
, int64_t offset
, int64_t bytes
, int64_t *pnum
,
296 int64_t *map
, BlockDriverState
**file
);
299 * Invalidate any cached meta-data.
301 void coroutine_fn (*bdrv_co_invalidate_cache
)(BlockDriverState
*bs
,
303 int (*bdrv_inactivate
)(BlockDriverState
*bs
);
306 * Flushes all data for all layers by calling bdrv_co_flush for underlying
307 * layers, if needed. This function is needed for deterministic
308 * synchronization of the flush finishing callback.
310 int coroutine_fn (*bdrv_co_flush
)(BlockDriverState
*bs
);
313 * Flushes all data that was already written to the OS all the way down to
314 * the disk (for example file-posix.c calls fsync()).
316 int coroutine_fn (*bdrv_co_flush_to_disk
)(BlockDriverState
*bs
);
319 * Flushes all internal caches to the OS. The data may still sit in a
320 * writeback cache of the host OS, but it will survive a crash of the qemu
323 int coroutine_fn (*bdrv_co_flush_to_os
)(BlockDriverState
*bs
);
326 * Drivers setting this field must be able to work with just a plain
327 * filename with '<protocol_name>:' as a prefix, and no other options.
328 * Options may be extracted from the filename by implementing
329 * bdrv_parse_filename.
331 const char *protocol_name
;
332 int coroutine_fn (*bdrv_co_truncate
)(BlockDriverState
*bs
, int64_t offset
,
333 PreallocMode prealloc
, Error
**errp
);
335 int64_t (*bdrv_getlength
)(BlockDriverState
*bs
);
336 bool has_variable_length
;
337 int64_t (*bdrv_get_allocated_file_size
)(BlockDriverState
*bs
);
338 BlockMeasureInfo
*(*bdrv_measure
)(QemuOpts
*opts
, BlockDriverState
*in_bs
,
341 int coroutine_fn (*bdrv_co_pwritev_compressed
)(BlockDriverState
*bs
,
342 uint64_t offset
, uint64_t bytes
, QEMUIOVector
*qiov
);
344 int (*bdrv_snapshot_create
)(BlockDriverState
*bs
,
345 QEMUSnapshotInfo
*sn_info
);
346 int (*bdrv_snapshot_goto
)(BlockDriverState
*bs
,
347 const char *snapshot_id
);
348 int (*bdrv_snapshot_delete
)(BlockDriverState
*bs
,
349 const char *snapshot_id
,
352 int (*bdrv_snapshot_list
)(BlockDriverState
*bs
,
353 QEMUSnapshotInfo
**psn_info
);
354 int (*bdrv_snapshot_load_tmp
)(BlockDriverState
*bs
,
355 const char *snapshot_id
,
358 int (*bdrv_get_info
)(BlockDriverState
*bs
, BlockDriverInfo
*bdi
);
359 ImageInfoSpecific
*(*bdrv_get_specific_info
)(BlockDriverState
*bs
,
362 int coroutine_fn (*bdrv_save_vmstate
)(BlockDriverState
*bs
,
365 int coroutine_fn (*bdrv_load_vmstate
)(BlockDriverState
*bs
,
369 int (*bdrv_change_backing_file
)(BlockDriverState
*bs
,
370 const char *backing_file
, const char *backing_fmt
);
372 /* removable device specific */
373 bool (*bdrv_is_inserted
)(BlockDriverState
*bs
);
374 void (*bdrv_eject
)(BlockDriverState
*bs
, bool eject_flag
);
375 void (*bdrv_lock_medium
)(BlockDriverState
*bs
, bool locked
);
377 /* to control generic scsi devices */
378 BlockAIOCB
*(*bdrv_aio_ioctl
)(BlockDriverState
*bs
,
379 unsigned long int req
, void *buf
,
380 BlockCompletionFunc
*cb
, void *opaque
);
381 int coroutine_fn (*bdrv_co_ioctl
)(BlockDriverState
*bs
,
382 unsigned long int req
, void *buf
);
384 /* List of options for creating images, terminated by name == NULL */
385 QemuOptsList
*create_opts
;
387 * If this driver supports reopening images this contains a
388 * NULL-terminated list of the runtime options that can be
389 * modified. If an option in this list is unspecified during
390 * reopen then it _must_ be reset to its default value or return
393 const char *const *mutable_opts
;
396 * Returns 0 for completed check, -errno for internal errors.
397 * The check results are stored in result.
399 int coroutine_fn (*bdrv_co_check
)(BlockDriverState
*bs
,
400 BdrvCheckResult
*result
,
403 int (*bdrv_amend_options
)(BlockDriverState
*bs
, QemuOpts
*opts
,
404 BlockDriverAmendStatusCB
*status_cb
,
408 void (*bdrv_debug_event
)(BlockDriverState
*bs
, BlkdebugEvent event
);
410 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
411 int (*bdrv_debug_breakpoint
)(BlockDriverState
*bs
, const char *event
,
413 int (*bdrv_debug_remove_breakpoint
)(BlockDriverState
*bs
,
415 int (*bdrv_debug_resume
)(BlockDriverState
*bs
, const char *tag
);
416 bool (*bdrv_debug_is_suspended
)(BlockDriverState
*bs
, const char *tag
);
418 void (*bdrv_refresh_limits
)(BlockDriverState
*bs
, Error
**errp
);
421 * Returns 1 if newly created images are guaranteed to contain only
422 * zeros, 0 otherwise.
424 int (*bdrv_has_zero_init
)(BlockDriverState
*bs
);
426 /* Remove fd handlers, timers, and other event loop callbacks so the event
427 * loop is no longer in use. Called with no in-flight requests and in
428 * depth-first traversal order with parents before child nodes.
430 void (*bdrv_detach_aio_context
)(BlockDriverState
*bs
);
432 /* Add fd handlers, timers, and other event loop callbacks so I/O requests
433 * can be processed again. Called with no in-flight requests and in
434 * depth-first traversal order with child nodes before parent nodes.
436 void (*bdrv_attach_aio_context
)(BlockDriverState
*bs
,
437 AioContext
*new_context
);
439 /* io queue for linux-aio */
440 void (*bdrv_io_plug
)(BlockDriverState
*bs
);
441 void (*bdrv_io_unplug
)(BlockDriverState
*bs
);
444 * Try to get @bs's logical and physical block size.
445 * On success, store them in @bsz and return zero.
446 * On failure, return negative errno.
448 int (*bdrv_probe_blocksizes
)(BlockDriverState
*bs
, BlockSizes
*bsz
);
450 * Try to get @bs's geometry (cyls, heads, sectors)
451 * On success, store them in @geo and return 0.
452 * On failure return -errno.
453 * Only drivers that want to override guest geometry implement this
454 * callback; see hd_geometry_guess().
456 int (*bdrv_probe_geometry
)(BlockDriverState
*bs
, HDGeometry
*geo
);
459 * bdrv_co_drain_begin is called if implemented in the beginning of a
460 * drain operation to drain and stop any internal sources of requests in
462 * bdrv_co_drain_end is called if implemented at the end of the drain.
464 * They should be used by the driver to e.g. manage scheduled I/O
465 * requests, or toggle an internal state. After the end of the drain new
466 * requests will continue normally.
468 void coroutine_fn (*bdrv_co_drain_begin
)(BlockDriverState
*bs
);
469 void coroutine_fn (*bdrv_co_drain_end
)(BlockDriverState
*bs
);
471 void (*bdrv_add_child
)(BlockDriverState
*parent
, BlockDriverState
*child
,
473 void (*bdrv_del_child
)(BlockDriverState
*parent
, BdrvChild
*child
,
477 * Informs the block driver that a permission change is intended. The
478 * driver checks whether the change is permissible and may take other
479 * preparations for the change (e.g. get file system locks). This operation
480 * is always followed either by a call to either .bdrv_set_perm or
481 * .bdrv_abort_perm_update.
483 * Checks whether the requested set of cumulative permissions in @perm
484 * can be granted for accessing @bs and whether no other users are using
485 * permissions other than those given in @shared (both arguments take
486 * BLK_PERM_* bitmasks).
488 * If both conditions are met, 0 is returned. Otherwise, -errno is returned
489 * and errp is set to an error describing the conflict.
491 int (*bdrv_check_perm
)(BlockDriverState
*bs
, uint64_t perm
,
492 uint64_t shared
, Error
**errp
);
495 * Called to inform the driver that the set of cumulative set of used
496 * permissions for @bs has changed to @perm, and the set of sharable
497 * permission to @shared. The driver can use this to propagate changes to
498 * its children (i.e. request permissions only if a parent actually needs
501 * This function is only invoked after bdrv_check_perm(), so block drivers
502 * may rely on preparations made in their .bdrv_check_perm implementation.
504 void (*bdrv_set_perm
)(BlockDriverState
*bs
, uint64_t perm
, uint64_t shared
);
507 * Called to inform the driver that after a previous bdrv_check_perm()
508 * call, the permission update is not performed and any preparations made
509 * for it (e.g. taken file locks) need to be undone.
511 * This function can be called even for nodes that never saw a
512 * bdrv_check_perm() call. It is a no-op then.
514 void (*bdrv_abort_perm_update
)(BlockDriverState
*bs
);
517 * Returns in @nperm and @nshared the permissions that the driver for @bs
518 * needs on its child @c, based on the cumulative permissions requested by
519 * the parents in @parent_perm and @parent_shared.
521 * If @c is NULL, return the permissions for attaching a new child for the
524 * If @reopen_queue is non-NULL, don't return the currently needed
525 * permissions, but those that will be needed after applying the
528 void (*bdrv_child_perm
)(BlockDriverState
*bs
, BdrvChild
*c
,
529 const BdrvChildRole
*role
,
530 BlockReopenQueue
*reopen_queue
,
531 uint64_t parent_perm
, uint64_t parent_shared
,
532 uint64_t *nperm
, uint64_t *nshared
);
535 * Bitmaps should be marked as 'IN_USE' in the image on reopening image
536 * as rw. This handler should realize it. It also should unset readonly
537 * field of BlockDirtyBitmap's in case of success.
539 int (*bdrv_reopen_bitmaps_rw
)(BlockDriverState
*bs
, Error
**errp
);
540 bool (*bdrv_can_store_new_dirty_bitmap
)(BlockDriverState
*bs
,
542 uint32_t granularity
,
544 void (*bdrv_remove_persistent_dirty_bitmap
)(BlockDriverState
*bs
,
549 * Register/unregister a buffer for I/O. For example, when the driver is
550 * interested to know the memory areas that will later be used in iovs, so
551 * that it can do IOMMU mapping with VFIO etc., in order to get better
552 * performance. In the case of VFIO drivers, this callback is used to do
553 * DMA mapping for hot buffers.
555 void (*bdrv_register_buf
)(BlockDriverState
*bs
, void *host
, size_t size
);
556 void (*bdrv_unregister_buf
)(BlockDriverState
*bs
, void *host
);
557 QLIST_ENTRY(BlockDriver
) list
;
559 /* Pointer to a NULL-terminated array of names of strong options
560 * that can be specified for bdrv_open(). A strong option is one
561 * that changes the data of a BDS.
562 * If this pointer is NULL, the array is considered empty.
563 * "filename" and "driver" are always considered strong. */
564 const char *const *strong_runtime_opts
;
567 typedef struct BlockLimits
{
568 /* Alignment requirement, in bytes, for offset/length of I/O
569 * requests. Must be a power of 2 less than INT_MAX; defaults to
570 * 1 for drivers with modern byte interfaces, and to 512
572 uint32_t request_alignment
;
574 /* Maximum number of bytes that can be discarded at once (since it
575 * is signed, it must be < 2G, if set). Must be multiple of
576 * pdiscard_alignment, but need not be power of 2. May be 0 if no
577 * inherent 32-bit limit */
578 int32_t max_pdiscard
;
580 /* Optimal alignment for discard requests in bytes. A power of 2
581 * is best but not mandatory. Must be a multiple of
582 * bl.request_alignment, and must be less than max_pdiscard if
583 * that is set. May be 0 if bl.request_alignment is good enough */
584 uint32_t pdiscard_alignment
;
586 /* Maximum number of bytes that can zeroized at once (since it is
587 * signed, it must be < 2G, if set). Must be multiple of
588 * pwrite_zeroes_alignment. May be 0 if no inherent 32-bit limit */
589 int32_t max_pwrite_zeroes
;
591 /* Optimal alignment for write zeroes requests in bytes. A power
592 * of 2 is best but not mandatory. Must be a multiple of
593 * bl.request_alignment, and must be less than max_pwrite_zeroes
594 * if that is set. May be 0 if bl.request_alignment is good
596 uint32_t pwrite_zeroes_alignment
;
598 /* Optimal transfer length in bytes. A power of 2 is best but not
599 * mandatory. Must be a multiple of bl.request_alignment, or 0 if
600 * no preferred size */
601 uint32_t opt_transfer
;
603 /* Maximal transfer length in bytes. Need not be power of 2, but
604 * must be multiple of opt_transfer and bl.request_alignment, or 0
605 * for no 32-bit limit. For now, anything larger than INT_MAX is
607 uint32_t max_transfer
;
609 /* memory alignment, in bytes so that no bounce buffer is needed */
610 size_t min_mem_alignment
;
612 /* memory alignment, in bytes, for bounce buffer */
613 size_t opt_mem_alignment
;
615 /* maximum number of iovec elements */
619 typedef struct BdrvOpBlocker BdrvOpBlocker
;
621 typedef struct BdrvAioNotifier
{
622 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
);
623 void (*detach_aio_context
)(void *opaque
);
628 QLIST_ENTRY(BdrvAioNotifier
) list
;
631 struct BdrvChildRole
{
632 /* If true, bdrv_replace_node() doesn't change the node this BdrvChild
636 /* If true, the parent is a BlockDriverState and bdrv_next_all_states()
637 * will return it. This information is used for drain_all, where every node
638 * will be drained separately, so the drain only needs to be propagated to
639 * non-BDS parents. */
642 void (*inherit_options
)(int *child_flags
, QDict
*child_options
,
643 int parent_flags
, QDict
*parent_options
);
645 void (*change_media
)(BdrvChild
*child
, bool load
);
646 void (*resize
)(BdrvChild
*child
);
648 /* Returns a name that is supposedly more useful for human users than the
649 * node name for identifying the node in question (in particular, a BB
650 * name), or NULL if the parent can't provide a better name. */
651 const char *(*get_name
)(BdrvChild
*child
);
653 /* Returns a malloced string that describes the parent of the child for a
654 * human reader. This could be a node-name, BlockBackend name, qdev ID or
655 * QOM path of the device owning the BlockBackend, job type and ID etc. The
656 * caller is responsible for freeing the memory. */
657 char *(*get_parent_desc
)(BdrvChild
*child
);
660 * If this pair of functions is implemented, the parent doesn't issue new
661 * requests after returning from .drained_begin() until .drained_end() is
664 * These functions must not change the graph (and therefore also must not
665 * call aio_poll(), which could change the graph indirectly).
667 * If drained_end() schedules background operations, it must atomically
668 * increment *drained_end_counter for each such operation and atomically
669 * decrement it once the operation has settled.
671 * Note that this can be nested. If drained_begin() was called twice, new
672 * I/O is allowed only after drained_end() was called twice, too.
674 void (*drained_begin
)(BdrvChild
*child
);
675 void (*drained_end
)(BdrvChild
*child
, int *drained_end_counter
);
678 * Returns whether the parent has pending requests for the child. This
679 * callback is polled after .drained_begin() has been called until all
680 * activity on the child has stopped.
682 bool (*drained_poll
)(BdrvChild
*child
);
684 /* Notifies the parent that the child has been activated/inactivated (e.g.
685 * when migration is completing) and it can start/stop requesting
686 * permissions and doing I/O on it. */
687 void (*activate
)(BdrvChild
*child
, Error
**errp
);
688 int (*inactivate
)(BdrvChild
*child
);
690 void (*attach
)(BdrvChild
*child
);
691 void (*detach
)(BdrvChild
*child
);
693 /* Notifies the parent that the filename of its child has changed (e.g.
694 * because the direct child was removed from the backing chain), so that it
695 * can update its reference. */
696 int (*update_filename
)(BdrvChild
*child
, BlockDriverState
*new_base
,
697 const char *filename
, Error
**errp
);
699 bool (*can_set_aio_ctx
)(BdrvChild
*child
, AioContext
*ctx
,
700 GSList
**ignore
, Error
**errp
);
701 void (*set_aio_ctx
)(BdrvChild
*child
, AioContext
*ctx
, GSList
**ignore
);
704 extern const BdrvChildRole child_file
;
705 extern const BdrvChildRole child_format
;
706 extern const BdrvChildRole child_backing
;
709 BlockDriverState
*bs
;
711 const BdrvChildRole
*role
;
715 * Granted permissions for operating on this BdrvChild (BLK_PERM_* bitmask)
720 * Permissions that can still be granted to other users of @bs while this
721 * BdrvChild is still attached to it. (BLK_PERM_* bitmask)
723 uint64_t shared_perm
;
725 /* backup of permissions during permission update procedure */
726 bool has_backup_perm
;
727 uint64_t backup_perm
;
728 uint64_t backup_shared_perm
;
731 * This link is frozen: the child can neither be replaced nor
732 * detached from the parent.
737 * How many times the parent of this child has been drained
738 * (through role->drained_*).
739 * Usually, this is equal to bs->quiesce_counter (potentially
740 * reduced by bdrv_drain_all_count). It may differ while the
741 * child is entering or leaving a drained section.
743 int parent_quiesce_counter
;
745 QLIST_ENTRY(BdrvChild
) next
;
746 QLIST_ENTRY(BdrvChild
) next_parent
;
750 * Note: the function bdrv_append() copies and swaps contents of
751 * BlockDriverStates, so if you add new fields to this struct, please
752 * inspect bdrv_append() to determine if the new fields need to be
755 struct BlockDriverState
{
756 /* Protected by big QEMU lock or read-only after opening. No special
757 * locking needed during I/O...
759 int open_flags
; /* flags used to open the file, re-used for re-open */
760 bool read_only
; /* if true, the media is read only */
761 bool encrypted
; /* if true, the media is encrypted */
762 bool sg
; /* if true, the device is a /dev/sg* */
763 bool probed
; /* if true, format was probed rather than specified */
764 bool force_share
; /* if true, always allow all shared permissions */
765 bool implicit
; /* if true, this filter node was automatically inserted */
767 BlockDriver
*drv
; /* NULL means no media */
770 AioContext
*aio_context
; /* event loop used for fd handlers, timers, etc */
771 /* long-running tasks intended to always use the same AioContext as this
772 * BDS may register themselves in this list to be notified of changes
773 * regarding this BDS's context */
774 QLIST_HEAD(, BdrvAioNotifier
) aio_notifiers
;
775 bool walking_aio_notifiers
; /* to make removal during iteration safe */
777 char filename
[PATH_MAX
];
778 char backing_file
[PATH_MAX
]; /* if non zero, the image is a diff of
780 /* The backing filename indicated by the image header; if we ever
781 * open this file, then this is replaced by the resulting BDS's
782 * filename (i.e. after a bdrv_refresh_filename() run). */
783 char auto_backing_file
[PATH_MAX
];
784 char backing_format
[16]; /* if non-zero and backing_file exists */
786 QDict
*full_open_options
;
787 char exact_filename
[PATH_MAX
];
795 /* Flags honored during pwrite (so far: BDRV_REQ_FUA,
796 * BDRV_REQ_WRITE_UNCHANGED).
797 * If a driver does not support BDRV_REQ_WRITE_UNCHANGED, those
798 * writes will be issued as normal writes without the flag set.
799 * This is important to note for drivers that do not explicitly
800 * request a WRITE permission for their children and instead take
801 * the same permissions as their parent did (this is commonly what
802 * block filters do). Such drivers have to be aware that the
803 * parent may have taken a WRITE_UNCHANGED permission only and is
804 * issuing such requests. Drivers either must make sure that
805 * these requests do not result in plain WRITE accesses (usually
806 * by supporting BDRV_REQ_WRITE_UNCHANGED, and then forwarding
807 * every incoming write request as-is, including potentially that
808 * flag), or they have to explicitly take the WRITE permission for
810 unsigned int supported_write_flags
;
811 /* Flags honored during pwrite_zeroes (so far: BDRV_REQ_FUA,
812 * BDRV_REQ_MAY_UNMAP, BDRV_REQ_WRITE_UNCHANGED) */
813 unsigned int supported_zero_flags
;
815 /* the following member gives a name to every node on the bs graph. */
817 /* element of the list of named nodes building the graph */
818 QTAILQ_ENTRY(BlockDriverState
) node_list
;
819 /* element of the list of all BlockDriverStates (all_bdrv_states) */
820 QTAILQ_ENTRY(BlockDriverState
) bs_list
;
821 /* element of the list of monitor-owned BDS */
822 QTAILQ_ENTRY(BlockDriverState
) monitor_list
;
825 /* operation blockers */
826 QLIST_HEAD(, BdrvOpBlocker
) op_blockers
[BLOCK_OP_TYPE_MAX
];
828 /* The node that this node inherited default options from (and a reopen on
829 * which can affect this node by changing these defaults). This is always a
830 * parent node of this node. */
831 BlockDriverState
*inherits_from
;
832 QLIST_HEAD(, BdrvChild
) children
;
833 QLIST_HEAD(, BdrvChild
) parents
;
836 QDict
*explicit_options
;
837 BlockdevDetectZeroesOptions detect_zeroes
;
839 /* The error object in use for blocking operations on backing_hd */
840 Error
*backing_blocker
;
842 /* Protected by AioContext lock */
844 /* If we are reading a disk image, give its size in sectors.
845 * Generally read-only; it is written to by load_snapshot and
846 * save_snaphost, but the block layer is quiescent during those.
848 int64_t total_sectors
;
850 /* Callback before write request is processed */
851 NotifierWithReturnList before_write_notifiers
;
853 /* threshold limit for writes, in bytes. "High water mark". */
854 uint64_t write_threshold_offset
;
855 NotifierWithReturn write_threshold_notifier
;
857 /* Writing to the list requires the BQL _and_ the dirty_bitmap_mutex.
858 * Reading from the list can be done with either the BQL or the
859 * dirty_bitmap_mutex. Modifying a bitmap only requires
860 * dirty_bitmap_mutex. */
861 QemuMutex dirty_bitmap_mutex
;
862 QLIST_HEAD(, BdrvDirtyBitmap
) dirty_bitmaps
;
864 /* Offset after the highest byte written to */
865 Stat64 wr_highest_offset
;
867 /* If true, copy read backing sectors into image. Can be >1 if more
868 * than one client has requested copy-on-read. Accessed with atomic
873 /* number of in-flight requests; overall and serialising.
874 * Accessed with atomic ops.
876 unsigned int in_flight
;
877 unsigned int serialising_in_flight
;
879 /* counter for nested bdrv_io_plug.
880 * Accessed with atomic ops.
884 /* do we need to tell the quest if we have a volatile write cache? */
885 int enable_write_cache
;
887 /* Accessed with atomic ops. */
889 int recursive_quiesce_counter
;
891 unsigned int write_gen
; /* Current data generation */
893 /* Protected by reqs_lock. */
895 QLIST_HEAD(, BdrvTrackedRequest
) tracked_requests
;
896 CoQueue flush_queue
; /* Serializing flush queue */
897 bool active_flush_req
; /* Flush request in flight? */
899 /* Only read/written by whoever has set active_flush_req to true. */
900 unsigned int flushed_gen
; /* Flushed write generation */
902 /* BdrvChild links to this node may never be frozen */
906 struct BlockBackendRootState
{
909 BlockdevDetectZeroesOptions detect_zeroes
;
912 typedef enum BlockMirrorBackingMode
{
913 /* Reuse the existing backing chain from the source for the target.
914 * - sync=full: Set backing BDS to NULL.
915 * - sync=top: Use source's backing BDS.
916 * - sync=none: Use source as the backing BDS. */
917 MIRROR_SOURCE_BACKING_CHAIN
,
919 /* Open the target's backing chain completely anew */
920 MIRROR_OPEN_BACKING_CHAIN
,
922 /* Do not change the target's backing BDS after job completion */
923 MIRROR_LEAVE_BACKING_CHAIN
,
924 } BlockMirrorBackingMode
;
926 static inline BlockDriverState
*backing_bs(BlockDriverState
*bs
)
928 return bs
->backing
? bs
->backing
->bs
: NULL
;
932 /* Essential block drivers which must always be statically linked into qemu, and
933 * which therefore can be accessed without using bdrv_find_format() */
934 extern BlockDriver bdrv_file
;
935 extern BlockDriver bdrv_raw
;
936 extern BlockDriver bdrv_qcow2
;
938 int coroutine_fn
bdrv_co_preadv(BdrvChild
*child
,
939 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
940 BdrvRequestFlags flags
);
941 int coroutine_fn
bdrv_co_pwritev(BdrvChild
*child
,
942 int64_t offset
, unsigned int bytes
, QEMUIOVector
*qiov
,
943 BdrvRequestFlags flags
);
945 static inline int coroutine_fn
bdrv_co_pread(BdrvChild
*child
,
946 int64_t offset
, unsigned int bytes
, void *buf
, BdrvRequestFlags flags
)
948 QEMUIOVector qiov
= QEMU_IOVEC_INIT_BUF(qiov
, buf
, bytes
);
950 return bdrv_co_preadv(child
, offset
, bytes
, &qiov
, flags
);
953 static inline int coroutine_fn
bdrv_co_pwrite(BdrvChild
*child
,
954 int64_t offset
, unsigned int bytes
, void *buf
, BdrvRequestFlags flags
)
956 QEMUIOVector qiov
= QEMU_IOVEC_INIT_BUF(qiov
, buf
, bytes
);
958 return bdrv_co_pwritev(child
, offset
, bytes
, &qiov
, flags
);
961 extern unsigned int bdrv_drain_all_count
;
962 void bdrv_apply_subtree_drain(BdrvChild
*child
, BlockDriverState
*new_parent
);
963 void bdrv_unapply_subtree_drain(BdrvChild
*child
, BlockDriverState
*old_parent
);
965 int get_tmp_filename(char *filename
, int size
);
966 BlockDriver
*bdrv_probe_all(const uint8_t *buf
, int buf_size
,
967 const char *filename
);
969 void bdrv_parse_filename_strip_prefix(const char *filename
, const char *prefix
,
974 * bdrv_add_before_write_notifier:
976 * Register a callback that is invoked before write requests are processed but
977 * after any throttling or waiting for overlapping requests.
979 void bdrv_add_before_write_notifier(BlockDriverState
*bs
,
980 NotifierWithReturn
*notifier
);
983 * bdrv_add_aio_context_notifier:
985 * If a long-running job intends to be always run in the same AioContext as a
986 * certain BDS, it may use this function to be notified of changes regarding the
987 * association of the BDS to an AioContext.
989 * attached_aio_context() is called after the target BDS has been attached to a
990 * new AioContext; detach_aio_context() is called before the target BDS is being
991 * detached from its old AioContext.
993 void bdrv_add_aio_context_notifier(BlockDriverState
*bs
,
994 void (*attached_aio_context
)(AioContext
*new_context
, void *opaque
),
995 void (*detach_aio_context
)(void *opaque
), void *opaque
);
998 * bdrv_remove_aio_context_notifier:
1000 * Unsubscribe of change notifications regarding the BDS's AioContext. The
1001 * parameters given here have to be the same as those given to
1002 * bdrv_add_aio_context_notifier().
1004 void bdrv_remove_aio_context_notifier(BlockDriverState
*bs
,
1005 void (*aio_context_attached
)(AioContext
*,
1007 void (*aio_context_detached
)(void *),
1012 * @bs: The BlockDriverState for which an I/O operation has been completed.
1014 * Wake up the main thread if it is waiting on BDRV_POLL_WHILE. During
1015 * synchronous I/O on a BlockDriverState that is attached to another
1016 * I/O thread, the main thread lets the I/O thread's event loop run,
1017 * waiting for the I/O operation to complete. A bdrv_wakeup will wake
1018 * up the main thread if necessary.
1020 * Manual calls to bdrv_wakeup are rarely necessary, because
1021 * bdrv_dec_in_flight already calls it.
1023 void bdrv_wakeup(BlockDriverState
*bs
);
1026 int is_windows_drive(const char *filename
);
1031 * @job_id: The id of the newly-created job, or %NULL to use the
1032 * device name of @bs.
1033 * @bs: Block device to operate on.
1034 * @base: Block device that will become the new base, or %NULL to
1035 * flatten the whole backing file chain onto @bs.
1036 * @backing_file_str: The file name that will be written to @bs as the
1037 * the new backing file if the job completes. Ignored if @base is %NULL.
1038 * @creation_flags: Flags that control the behavior of the Job lifetime.
1039 * See @BlockJobCreateFlags
1040 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
1041 * @on_error: The action to take upon error.
1042 * @errp: Error object.
1044 * Start a streaming operation on @bs. Clusters that are unallocated
1045 * in @bs, but allocated in any image between @base and @bs (both
1046 * exclusive) will be written to @bs. At the end of a successful
1047 * streaming job, the backing file of @bs will be changed to
1048 * @backing_file_str in the written image and to @base in the live
1051 void stream_start(const char *job_id
, BlockDriverState
*bs
,
1052 BlockDriverState
*base
, const char *backing_file_str
,
1053 int creation_flags
, int64_t speed
,
1054 BlockdevOnError on_error
, Error
**errp
);
1058 * @job_id: The id of the newly-created job, or %NULL to use the
1059 * device name of @bs.
1060 * @bs: Active block device.
1061 * @top: Top block device to be committed.
1062 * @base: Block device that will be written into, and become the new top.
1063 * @creation_flags: Flags that control the behavior of the Job lifetime.
1064 * See @BlockJobCreateFlags
1065 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
1066 * @on_error: The action to take upon error.
1067 * @backing_file_str: String to use as the backing file in @top's overlay
1068 * @filter_node_name: The node name that should be assigned to the filter
1069 * driver that the commit job inserts into the graph above @top. NULL means
1070 * that a node name should be autogenerated.
1071 * @errp: Error object.
1074 void commit_start(const char *job_id
, BlockDriverState
*bs
,
1075 BlockDriverState
*base
, BlockDriverState
*top
,
1076 int creation_flags
, int64_t speed
,
1077 BlockdevOnError on_error
, const char *backing_file_str
,
1078 const char *filter_node_name
, Error
**errp
);
1080 * commit_active_start:
1081 * @job_id: The id of the newly-created job, or %NULL to use the
1082 * device name of @bs.
1083 * @bs: Active block device to be committed.
1084 * @base: Block device that will be written into, and become the new top.
1085 * @creation_flags: Flags that control the behavior of the Job lifetime.
1086 * See @BlockJobCreateFlags
1087 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
1088 * @on_error: The action to take upon error.
1089 * @filter_node_name: The node name that should be assigned to the filter
1090 * driver that the commit job inserts into the graph above @bs. NULL means that
1091 * a node name should be autogenerated.
1092 * @cb: Completion function for the job.
1093 * @opaque: Opaque pointer value passed to @cb.
1094 * @auto_complete: Auto complete the job.
1095 * @errp: Error object.
1098 BlockJob
*commit_active_start(const char *job_id
, BlockDriverState
*bs
,
1099 BlockDriverState
*base
, int creation_flags
,
1100 int64_t speed
, BlockdevOnError on_error
,
1101 const char *filter_node_name
,
1102 BlockCompletionFunc
*cb
, void *opaque
,
1103 bool auto_complete
, Error
**errp
);
1106 * @job_id: The id of the newly-created job, or %NULL to use the
1107 * device name of @bs.
1108 * @bs: Block device to operate on.
1109 * @target: Block device to write to.
1110 * @replaces: Block graph node name to replace once the mirror is done. Can
1111 * only be used when full mirroring is selected.
1112 * @creation_flags: Flags that control the behavior of the Job lifetime.
1113 * See @BlockJobCreateFlags
1114 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
1115 * @granularity: The chosen granularity for the dirty bitmap.
1116 * @buf_size: The amount of data that can be in flight at one time.
1117 * @mode: Whether to collapse all images in the chain to the target.
1118 * @backing_mode: How to establish the target's backing chain after completion.
1119 * @on_source_error: The action to take upon error reading from the source.
1120 * @on_target_error: The action to take upon error writing to the target.
1121 * @unmap: Whether to unmap target where source sectors only contain zeroes.
1122 * @filter_node_name: The node name that should be assigned to the filter
1123 * driver that the mirror job inserts into the graph above @bs. NULL means that
1124 * a node name should be autogenerated.
1125 * @copy_mode: When to trigger writes to the target.
1126 * @errp: Error object.
1128 * Start a mirroring operation on @bs. Clusters that are allocated
1129 * in @bs will be written to @target until the job is cancelled or
1130 * manually completed. At the end of a successful mirroring job,
1131 * @bs will be switched to read from @target.
1133 void mirror_start(const char *job_id
, BlockDriverState
*bs
,
1134 BlockDriverState
*target
, const char *replaces
,
1135 int creation_flags
, int64_t speed
,
1136 uint32_t granularity
, int64_t buf_size
,
1137 MirrorSyncMode mode
, BlockMirrorBackingMode backing_mode
,
1138 BlockdevOnError on_source_error
,
1139 BlockdevOnError on_target_error
,
1140 bool unmap
, const char *filter_node_name
,
1141 MirrorCopyMode copy_mode
, Error
**errp
);
1144 * backup_job_create:
1145 * @job_id: The id of the newly-created job, or %NULL to use the
1146 * device name of @bs.
1147 * @bs: Block device to operate on.
1148 * @target: Block device to write to.
1149 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
1150 * @sync_mode: What parts of the disk image should be copied to the destination.
1151 * @sync_bitmap: The dirty bitmap if sync_mode is MIRROR_SYNC_MODE_INCREMENTAL.
1152 * @on_source_error: The action to take upon error reading from the source.
1153 * @on_target_error: The action to take upon error writing to the target.
1154 * @creation_flags: Flags that control the behavior of the Job lifetime.
1155 * See @BlockJobCreateFlags
1156 * @cb: Completion function for the job.
1157 * @opaque: Opaque pointer value passed to @cb.
1158 * @txn: Transaction that this job is part of (may be NULL).
1160 * Create a backup operation on @bs. Clusters in @bs are written to @target
1161 * until the job is cancelled or manually completed.
1163 BlockJob
*backup_job_create(const char *job_id
, BlockDriverState
*bs
,
1164 BlockDriverState
*target
, int64_t speed
,
1165 MirrorSyncMode sync_mode
,
1166 BdrvDirtyBitmap
*sync_bitmap
,
1168 BlockdevOnError on_source_error
,
1169 BlockdevOnError on_target_error
,
1171 BlockCompletionFunc
*cb
, void *opaque
,
1172 JobTxn
*txn
, Error
**errp
);
1174 void hmp_drive_add_node(Monitor
*mon
, const char *optstr
);
1176 BdrvChild
*bdrv_root_attach_child(BlockDriverState
*child_bs
,
1177 const char *child_name
,
1178 const BdrvChildRole
*child_role
,
1180 uint64_t perm
, uint64_t shared_perm
,
1181 void *opaque
, Error
**errp
);
1182 void bdrv_root_unref_child(BdrvChild
*child
);
1185 * Sets a BdrvChild's permissions. Avoid if the parent is a BDS; use
1186 * bdrv_child_refresh_perms() instead and make the parent's
1187 * .bdrv_child_perm() implementation return the correct values.
1189 int bdrv_child_try_set_perm(BdrvChild
*c
, uint64_t perm
, uint64_t shared
,
1193 * Calls bs->drv->bdrv_child_perm() and updates the child's permission
1194 * masks with the result.
1195 * Drivers should invoke this function whenever an event occurs that
1196 * makes their .bdrv_child_perm() implementation return different
1197 * values than before, but which will not result in the block layer
1198 * automatically refreshing the permissions.
1200 int bdrv_child_refresh_perms(BlockDriverState
*bs
, BdrvChild
*c
, Error
**errp
);
1202 /* Default implementation for BlockDriver.bdrv_child_perm() that can be used by
1203 * block filters: Forward CONSISTENT_READ, WRITE, WRITE_UNCHANGED and RESIZE to
1205 void bdrv_filter_default_perms(BlockDriverState
*bs
, BdrvChild
*c
,
1206 const BdrvChildRole
*role
,
1207 BlockReopenQueue
*reopen_queue
,
1208 uint64_t perm
, uint64_t shared
,
1209 uint64_t *nperm
, uint64_t *nshared
);
1211 /* Default implementation for BlockDriver.bdrv_child_perm() that can be used by
1212 * (non-raw) image formats: Like above for bs->backing, but for bs->file it
1213 * requires WRITE | RESIZE for read-write images, always requires
1214 * CONSISTENT_READ and doesn't share WRITE. */
1215 void bdrv_format_default_perms(BlockDriverState
*bs
, BdrvChild
*c
,
1216 const BdrvChildRole
*role
,
1217 BlockReopenQueue
*reopen_queue
,
1218 uint64_t perm
, uint64_t shared
,
1219 uint64_t *nperm
, uint64_t *nshared
);
1222 * Default implementation for drivers to pass bdrv_co_block_status() to
1225 int coroutine_fn
bdrv_co_block_status_from_file(BlockDriverState
*bs
,
1231 BlockDriverState
**file
);
1233 * Default implementation for drivers to pass bdrv_co_block_status() to
1234 * their backing file.
1236 int coroutine_fn
bdrv_co_block_status_from_backing(BlockDriverState
*bs
,
1242 BlockDriverState
**file
);
1243 const char *bdrv_get_parent_name(const BlockDriverState
*bs
);
1244 void blk_dev_change_media_cb(BlockBackend
*blk
, bool load
, Error
**errp
);
1245 bool blk_dev_has_removable_media(BlockBackend
*blk
);
1246 bool blk_dev_has_tray(BlockBackend
*blk
);
1247 void blk_dev_eject_request(BlockBackend
*blk
, bool force
);
1248 bool blk_dev_is_tray_open(BlockBackend
*blk
);
1249 bool blk_dev_is_medium_locked(BlockBackend
*blk
);
1251 void bdrv_set_dirty(BlockDriverState
*bs
, int64_t offset
, int64_t bytes
);
1253 void bdrv_clear_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
**out
);
1254 void bdrv_restore_dirty_bitmap(BdrvDirtyBitmap
*bitmap
, HBitmap
*backup
);
1256 void bdrv_inc_in_flight(BlockDriverState
*bs
);
1257 void bdrv_dec_in_flight(BlockDriverState
*bs
);
1259 void blockdev_close_all_bdrv_states(void);
1261 int coroutine_fn
bdrv_co_copy_range_from(BdrvChild
*src
, uint64_t src_offset
,
1262 BdrvChild
*dst
, uint64_t dst_offset
,
1264 BdrvRequestFlags read_flags
,
1265 BdrvRequestFlags write_flags
);
1266 int coroutine_fn
bdrv_co_copy_range_to(BdrvChild
*src
, uint64_t src_offset
,
1267 BdrvChild
*dst
, uint64_t dst_offset
,
1269 BdrvRequestFlags read_flags
,
1270 BdrvRequestFlags write_flags
);
1272 int refresh_total_sectors(BlockDriverState
*bs
, int64_t hint
);
1274 #endif /* BLOCK_INT_H */