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/block.h"
28 #include "qemu/option.h"
29 #include "qemu/queue.h"
30 #include "block/coroutine.h"
31 #include "qemu/timer.h"
32 #include "qapi-types.h"
33 #include "qapi/qmp/qerror.h"
34 #include "monitor/monitor.h"
35 #include "qemu/hbitmap.h"
36 #include "block/snapshot.h"
37 #include "qemu/main-loop.h"
38 #include "qemu/throttle.h"
40 #define BLOCK_FLAG_ENCRYPT 1
41 #define BLOCK_FLAG_COMPAT6 4
42 #define BLOCK_FLAG_LAZY_REFCOUNTS 8
44 #define BLOCK_OPT_SIZE "size"
45 #define BLOCK_OPT_ENCRYPT "encryption"
46 #define BLOCK_OPT_COMPAT6 "compat6"
47 #define BLOCK_OPT_BACKING_FILE "backing_file"
48 #define BLOCK_OPT_BACKING_FMT "backing_fmt"
49 #define BLOCK_OPT_CLUSTER_SIZE "cluster_size"
50 #define BLOCK_OPT_TABLE_SIZE "table_size"
51 #define BLOCK_OPT_PREALLOC "preallocation"
52 #define BLOCK_OPT_SUBFMT "subformat"
53 #define BLOCK_OPT_COMPAT_LEVEL "compat"
54 #define BLOCK_OPT_LAZY_REFCOUNTS "lazy_refcounts"
55 #define BLOCK_OPT_ADAPTER_TYPE "adapter_type"
57 typedef struct BdrvTrackedRequest
{
62 QLIST_ENTRY(BdrvTrackedRequest
) list
;
63 Coroutine
*co
; /* owner, used for deadlock detection */
64 CoQueue wait_queue
; /* coroutines blocked on this request */
68 const char *format_name
;
70 int (*bdrv_probe
)(const uint8_t *buf
, int buf_size
, const char *filename
);
71 int (*bdrv_probe_device
)(const char *filename
);
73 /* Any driver implementing this callback is expected to be able to handle
74 * NULL file names in its .bdrv_open() implementation */
75 void (*bdrv_parse_filename
)(const char *filename
, QDict
*options
, Error
**errp
);
77 /* For handling image reopen for split or non-split files */
78 int (*bdrv_reopen_prepare
)(BDRVReopenState
*reopen_state
,
79 BlockReopenQueue
*queue
, Error
**errp
);
80 void (*bdrv_reopen_commit
)(BDRVReopenState
*reopen_state
);
81 void (*bdrv_reopen_abort
)(BDRVReopenState
*reopen_state
);
83 int (*bdrv_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
);
84 int (*bdrv_file_open
)(BlockDriverState
*bs
, QDict
*options
, int flags
);
85 int (*bdrv_read
)(BlockDriverState
*bs
, int64_t sector_num
,
86 uint8_t *buf
, int nb_sectors
);
87 int (*bdrv_write
)(BlockDriverState
*bs
, int64_t sector_num
,
88 const uint8_t *buf
, int nb_sectors
);
89 void (*bdrv_close
)(BlockDriverState
*bs
);
90 void (*bdrv_rebind
)(BlockDriverState
*bs
);
91 int (*bdrv_create
)(const char *filename
, QEMUOptionParameter
*options
);
92 int (*bdrv_set_key
)(BlockDriverState
*bs
, const char *key
);
93 int (*bdrv_make_empty
)(BlockDriverState
*bs
);
95 BlockDriverAIOCB
*(*bdrv_aio_readv
)(BlockDriverState
*bs
,
96 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
97 BlockDriverCompletionFunc
*cb
, void *opaque
);
98 BlockDriverAIOCB
*(*bdrv_aio_writev
)(BlockDriverState
*bs
,
99 int64_t sector_num
, QEMUIOVector
*qiov
, int nb_sectors
,
100 BlockDriverCompletionFunc
*cb
, void *opaque
);
101 BlockDriverAIOCB
*(*bdrv_aio_flush
)(BlockDriverState
*bs
,
102 BlockDriverCompletionFunc
*cb
, void *opaque
);
103 BlockDriverAIOCB
*(*bdrv_aio_discard
)(BlockDriverState
*bs
,
104 int64_t sector_num
, int nb_sectors
,
105 BlockDriverCompletionFunc
*cb
, void *opaque
);
107 int coroutine_fn (*bdrv_co_readv
)(BlockDriverState
*bs
,
108 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
109 int coroutine_fn (*bdrv_co_writev
)(BlockDriverState
*bs
,
110 int64_t sector_num
, int nb_sectors
, QEMUIOVector
*qiov
);
112 * Efficiently zero a region of the disk image. Typically an image format
113 * would use a compact metadata representation to implement this. This
114 * function pointer may be NULL and .bdrv_co_writev() will be called
117 int coroutine_fn (*bdrv_co_write_zeroes
)(BlockDriverState
*bs
,
118 int64_t sector_num
, int nb_sectors
);
119 int coroutine_fn (*bdrv_co_discard
)(BlockDriverState
*bs
,
120 int64_t sector_num
, int nb_sectors
);
121 int64_t coroutine_fn (*bdrv_co_get_block_status
)(BlockDriverState
*bs
,
122 int64_t sector_num
, int nb_sectors
, int *pnum
);
125 * Invalidate any cached meta-data.
127 void (*bdrv_invalidate_cache
)(BlockDriverState
*bs
);
130 * Flushes all data that was already written to the OS all the way down to
131 * the disk (for example raw-posix calls fsync()).
133 int coroutine_fn (*bdrv_co_flush_to_disk
)(BlockDriverState
*bs
);
136 * Flushes all internal caches to the OS. The data may still sit in a
137 * writeback cache of the host OS, but it will survive a crash of the qemu
140 int coroutine_fn (*bdrv_co_flush_to_os
)(BlockDriverState
*bs
);
142 const char *protocol_name
;
143 int (*bdrv_truncate
)(BlockDriverState
*bs
, int64_t offset
);
144 int64_t (*bdrv_getlength
)(BlockDriverState
*bs
);
145 int64_t (*bdrv_get_allocated_file_size
)(BlockDriverState
*bs
);
146 int (*bdrv_write_compressed
)(BlockDriverState
*bs
, int64_t sector_num
,
147 const uint8_t *buf
, int nb_sectors
);
149 int (*bdrv_snapshot_create
)(BlockDriverState
*bs
,
150 QEMUSnapshotInfo
*sn_info
);
151 int (*bdrv_snapshot_goto
)(BlockDriverState
*bs
,
152 const char *snapshot_id
);
153 int (*bdrv_snapshot_delete
)(BlockDriverState
*bs
, const char *snapshot_id
);
154 int (*bdrv_snapshot_list
)(BlockDriverState
*bs
,
155 QEMUSnapshotInfo
**psn_info
);
156 int (*bdrv_snapshot_load_tmp
)(BlockDriverState
*bs
,
157 const char *snapshot_name
);
158 int (*bdrv_get_info
)(BlockDriverState
*bs
, BlockDriverInfo
*bdi
);
160 int (*bdrv_save_vmstate
)(BlockDriverState
*bs
, QEMUIOVector
*qiov
,
162 int (*bdrv_load_vmstate
)(BlockDriverState
*bs
, uint8_t *buf
,
163 int64_t pos
, int size
);
165 int (*bdrv_change_backing_file
)(BlockDriverState
*bs
,
166 const char *backing_file
, const char *backing_fmt
);
168 /* removable device specific */
169 int (*bdrv_is_inserted
)(BlockDriverState
*bs
);
170 int (*bdrv_media_changed
)(BlockDriverState
*bs
);
171 void (*bdrv_eject
)(BlockDriverState
*bs
, bool eject_flag
);
172 void (*bdrv_lock_medium
)(BlockDriverState
*bs
, bool locked
);
174 /* to control generic scsi devices */
175 int (*bdrv_ioctl
)(BlockDriverState
*bs
, unsigned long int req
, void *buf
);
176 BlockDriverAIOCB
*(*bdrv_aio_ioctl
)(BlockDriverState
*bs
,
177 unsigned long int req
, void *buf
,
178 BlockDriverCompletionFunc
*cb
, void *opaque
);
180 /* List of options for creating images, terminated by name == NULL */
181 QEMUOptionParameter
*create_options
;
185 * Returns 0 for completed check, -errno for internal errors.
186 * The check results are stored in result.
188 int (*bdrv_check
)(BlockDriverState
* bs
, BdrvCheckResult
*result
,
191 int (*bdrv_amend_options
)(BlockDriverState
*bs
,
192 QEMUOptionParameter
*options
);
194 void (*bdrv_debug_event
)(BlockDriverState
*bs
, BlkDebugEvent event
);
196 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
197 int (*bdrv_debug_breakpoint
)(BlockDriverState
*bs
, const char *event
,
199 int (*bdrv_debug_resume
)(BlockDriverState
*bs
, const char *tag
);
200 bool (*bdrv_debug_is_suspended
)(BlockDriverState
*bs
, const char *tag
);
203 * Returns 1 if newly created images are guaranteed to contain only
204 * zeros, 0 otherwise.
206 int (*bdrv_has_zero_init
)(BlockDriverState
*bs
);
208 QLIST_ENTRY(BlockDriver
) list
;
212 * Note: the function bdrv_append() copies and swaps contents of
213 * BlockDriverStates, so if you add new fields to this struct, please
214 * inspect bdrv_append() to determine if the new fields need to be
217 struct BlockDriverState
{
218 int64_t total_sectors
; /* if we are reading a disk image, give its
220 int read_only
; /* if true, the media is read only */
221 int open_flags
; /* flags used to open the file, re-used for re-open */
222 int encrypted
; /* if true, the media is encrypted */
223 int valid_key
; /* if true, a valid encryption key has been set */
224 int sg
; /* if true, the device is a /dev/sg* */
225 int copy_on_read
; /* if true, copy read backing sectors into image
226 note this is a reference count */
228 BlockDriver
*drv
; /* NULL means no media */
231 void *dev
; /* attached device model, if any */
232 /* TODO change to DeviceState when all users are qdevified */
233 const BlockDevOps
*dev_ops
;
237 char backing_file
[1024]; /* if non zero, the image is a diff of
239 char backing_format
[16]; /* if non-zero and backing_file exists */
242 BlockDriverState
*backing_hd
;
243 BlockDriverState
*file
;
245 NotifierList close_notifiers
;
247 /* Callback before write request is processed */
248 NotifierWithReturnList before_write_notifiers
;
250 /* number of in-flight copy-on-read requests */
251 unsigned int copy_on_read_in_flight
;
254 ThrottleState throttle_state
;
255 CoQueue throttled_reqs
[2];
256 bool io_limits_enabled
;
258 /* I/O stats (display with "info blockstats"). */
259 uint64_t nr_bytes
[BDRV_MAX_IOTYPE
];
260 uint64_t nr_ops
[BDRV_MAX_IOTYPE
];
261 uint64_t total_time_ns
[BDRV_MAX_IOTYPE
];
262 uint64_t wr_highest_sector
;
264 /* Whether the disk can expand beyond total_sectors */
267 /* Whether produces zeros when read beyond eof */
268 bool zero_beyond_eof
;
270 /* the memory alignment required for the buffers handled by this driver */
271 int buffer_alignment
;
273 /* do we need to tell the quest if we have a volatile write cache? */
274 int enable_write_cache
;
276 /* NOTE: the following infos are only hints for real hardware
277 drivers. They are not used by the block driver */
278 BlockdevOnError on_read_error
, on_write_error
;
279 bool iostatus_enabled
;
280 BlockDeviceIoStatus iostatus
;
281 char device_name
[32];
282 HBitmap
*dirty_bitmap
;
284 int in_use
; /* users other than guest access, eg. block migration */
285 QTAILQ_ENTRY(BlockDriverState
) list
;
287 QLIST_HEAD(, BdrvTrackedRequest
) tracked_requests
;
289 /* long-running background operation */
295 int get_tmp_filename(char *filename
, int size
);
297 void bdrv_set_io_limits(BlockDriverState
*bs
,
298 ThrottleConfig
*cfg
);
302 * bdrv_add_before_write_notifier:
304 * Register a callback that is invoked before write requests are processed but
305 * after any throttling or waiting for overlapping requests.
307 void bdrv_add_before_write_notifier(BlockDriverState
*bs
,
308 NotifierWithReturn
*notifier
);
311 * bdrv_get_aio_context:
313 * Returns: the currently bound #AioContext
315 AioContext
*bdrv_get_aio_context(BlockDriverState
*bs
);
318 int is_windows_drive(const char *filename
);
320 void bdrv_emit_qmp_error_event(const BlockDriverState
*bdrv
,
321 enum MonitorEvent ev
,
322 BlockErrorAction action
, bool is_read
);
326 * @bs: Block device to operate on.
327 * @base: Block device that will become the new base, or %NULL to
328 * flatten the whole backing file chain onto @bs.
329 * @base_id: The file name that will be written to @bs as the new
330 * backing file if the job completes. Ignored if @base is %NULL.
331 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
332 * @on_error: The action to take upon error.
333 * @cb: Completion function for the job.
334 * @opaque: Opaque pointer value passed to @cb.
335 * @errp: Error object.
337 * Start a streaming operation on @bs. Clusters that are unallocated
338 * in @bs, but allocated in any image between @base and @bs (both
339 * exclusive) will be written to @bs. At the end of a successful
340 * streaming job, the backing file of @bs will be changed to
341 * @base_id in the written image and to @base in the live BlockDriverState.
343 void stream_start(BlockDriverState
*bs
, BlockDriverState
*base
,
344 const char *base_id
, int64_t speed
, BlockdevOnError on_error
,
345 BlockDriverCompletionFunc
*cb
,
346 void *opaque
, Error
**errp
);
350 * @bs: Top Block device
351 * @base: Block device that will be written into, and become the new top
352 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
353 * @on_error: The action to take upon error.
354 * @cb: Completion function for the job.
355 * @opaque: Opaque pointer value passed to @cb.
356 * @errp: Error object.
359 void commit_start(BlockDriverState
*bs
, BlockDriverState
*base
,
360 BlockDriverState
*top
, int64_t speed
,
361 BlockdevOnError on_error
, BlockDriverCompletionFunc
*cb
,
362 void *opaque
, Error
**errp
);
366 * @bs: Block device to operate on.
367 * @target: Block device to write to.
368 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
369 * @granularity: The chosen granularity for the dirty bitmap.
370 * @buf_size: The amount of data that can be in flight at one time.
371 * @mode: Whether to collapse all images in the chain to the target.
372 * @on_source_error: The action to take upon error reading from the source.
373 * @on_target_error: The action to take upon error writing to the target.
374 * @cb: Completion function for the job.
375 * @opaque: Opaque pointer value passed to @cb.
376 * @errp: Error object.
378 * Start a mirroring operation on @bs. Clusters that are allocated
379 * in @bs will be written to @bs until the job is cancelled or
380 * manually completed. At the end of a successful mirroring job,
381 * @bs will be switched to read from @target.
383 void mirror_start(BlockDriverState
*bs
, BlockDriverState
*target
,
384 int64_t speed
, int64_t granularity
, int64_t buf_size
,
385 MirrorSyncMode mode
, BlockdevOnError on_source_error
,
386 BlockdevOnError on_target_error
,
387 BlockDriverCompletionFunc
*cb
,
388 void *opaque
, Error
**errp
);
392 * @bs: Block device to operate on.
393 * @target: Block device to write to.
394 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
395 * @sync_mode: What parts of the disk image should be copied to the destination.
396 * @on_source_error: The action to take upon error reading from the source.
397 * @on_target_error: The action to take upon error writing to the target.
398 * @cb: Completion function for the job.
399 * @opaque: Opaque pointer value passed to @cb.
401 * Start a backup operation on @bs. Clusters in @bs are written to @target
402 * until the job is cancelled or manually completed.
404 void backup_start(BlockDriverState
*bs
, BlockDriverState
*target
,
405 int64_t speed
, MirrorSyncMode sync_mode
,
406 BlockdevOnError on_source_error
,
407 BlockdevOnError on_target_error
,
408 BlockDriverCompletionFunc
*cb
, void *opaque
,
411 #endif /* BLOCK_INT_H */