snapshot: distinguish id and name in load_tmp
[qemu/kevin.git] / include / block / block_int.h
blobec0797ebc7de96d9325673c7c940d49fbe893ca0
1 /*
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
22 * THE SOFTWARE.
24 #ifndef BLOCK_INT_H
25 #define BLOCK_INT_H
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"
56 #define BLOCK_OPT_REDUNDANCY "redundancy"
58 typedef struct BdrvTrackedRequest {
59 BlockDriverState *bs;
60 int64_t sector_num;
61 int nb_sectors;
62 bool is_write;
63 QLIST_ENTRY(BdrvTrackedRequest) list;
64 Coroutine *co; /* owner, used for deadlock detection */
65 CoQueue wait_queue; /* coroutines blocked on this request */
66 } BdrvTrackedRequest;
68 struct BlockDriver {
69 const char *format_name;
70 int instance_size;
72 /* if not defined external snapshots are allowed
73 * future block filters will query their children to build the response
75 ExtSnapshotPerm (*bdrv_check_ext_snapshot)(BlockDriverState *bs);
77 int (*bdrv_probe)(const uint8_t *buf, int buf_size, const char *filename);
78 int (*bdrv_probe_device)(const char *filename);
80 /* Any driver implementing this callback is expected to be able to handle
81 * NULL file names in its .bdrv_open() implementation */
82 void (*bdrv_parse_filename)(const char *filename, QDict *options, Error **errp);
83 /* Drivers not implementing bdrv_parse_filename nor bdrv_open should have
84 * this field set to true, except ones that are defined only by their
85 * child's bs.
86 * An example of the last type will be the quorum block driver.
88 bool bdrv_needs_filename;
90 /* For handling image reopen for split or non-split files */
91 int (*bdrv_reopen_prepare)(BDRVReopenState *reopen_state,
92 BlockReopenQueue *queue, Error **errp);
93 void (*bdrv_reopen_commit)(BDRVReopenState *reopen_state);
94 void (*bdrv_reopen_abort)(BDRVReopenState *reopen_state);
96 int (*bdrv_open)(BlockDriverState *bs, QDict *options, int flags,
97 Error **errp);
98 int (*bdrv_file_open)(BlockDriverState *bs, QDict *options, int flags,
99 Error **errp);
100 int (*bdrv_read)(BlockDriverState *bs, int64_t sector_num,
101 uint8_t *buf, int nb_sectors);
102 int (*bdrv_write)(BlockDriverState *bs, int64_t sector_num,
103 const uint8_t *buf, int nb_sectors);
104 void (*bdrv_close)(BlockDriverState *bs);
105 void (*bdrv_rebind)(BlockDriverState *bs);
106 int (*bdrv_create)(const char *filename, QEMUOptionParameter *options,
107 Error **errp);
108 int (*bdrv_set_key)(BlockDriverState *bs, const char *key);
109 int (*bdrv_make_empty)(BlockDriverState *bs);
110 /* aio */
111 BlockDriverAIOCB *(*bdrv_aio_readv)(BlockDriverState *bs,
112 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
113 BlockDriverCompletionFunc *cb, void *opaque);
114 BlockDriverAIOCB *(*bdrv_aio_writev)(BlockDriverState *bs,
115 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
116 BlockDriverCompletionFunc *cb, void *opaque);
117 BlockDriverAIOCB *(*bdrv_aio_flush)(BlockDriverState *bs,
118 BlockDriverCompletionFunc *cb, void *opaque);
119 BlockDriverAIOCB *(*bdrv_aio_discard)(BlockDriverState *bs,
120 int64_t sector_num, int nb_sectors,
121 BlockDriverCompletionFunc *cb, void *opaque);
123 int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs,
124 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov);
125 int coroutine_fn (*bdrv_co_writev)(BlockDriverState *bs,
126 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov);
128 * Efficiently zero a region of the disk image. Typically an image format
129 * would use a compact metadata representation to implement this. This
130 * function pointer may be NULL and .bdrv_co_writev() will be called
131 * instead.
133 int coroutine_fn (*bdrv_co_write_zeroes)(BlockDriverState *bs,
134 int64_t sector_num, int nb_sectors, BdrvRequestFlags flags);
135 int coroutine_fn (*bdrv_co_discard)(BlockDriverState *bs,
136 int64_t sector_num, int nb_sectors);
137 int64_t coroutine_fn (*bdrv_co_get_block_status)(BlockDriverState *bs,
138 int64_t sector_num, int nb_sectors, int *pnum);
141 * Invalidate any cached meta-data.
143 void (*bdrv_invalidate_cache)(BlockDriverState *bs);
146 * Flushes all data that was already written to the OS all the way down to
147 * the disk (for example raw-posix calls fsync()).
149 int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs);
152 * Flushes all internal caches to the OS. The data may still sit in a
153 * writeback cache of the host OS, but it will survive a crash of the qemu
154 * process.
156 int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs);
158 const char *protocol_name;
159 int (*bdrv_truncate)(BlockDriverState *bs, int64_t offset);
161 int64_t (*bdrv_getlength)(BlockDriverState *bs);
162 bool has_variable_length;
163 int64_t (*bdrv_get_allocated_file_size)(BlockDriverState *bs);
165 int (*bdrv_write_compressed)(BlockDriverState *bs, int64_t sector_num,
166 const uint8_t *buf, int nb_sectors);
168 int (*bdrv_snapshot_create)(BlockDriverState *bs,
169 QEMUSnapshotInfo *sn_info);
170 int (*bdrv_snapshot_goto)(BlockDriverState *bs,
171 const char *snapshot_id);
172 int (*bdrv_snapshot_delete)(BlockDriverState *bs,
173 const char *snapshot_id,
174 const char *name,
175 Error **errp);
176 int (*bdrv_snapshot_list)(BlockDriverState *bs,
177 QEMUSnapshotInfo **psn_info);
178 int (*bdrv_snapshot_load_tmp)(BlockDriverState *bs,
179 const char *snapshot_id,
180 const char *name,
181 Error **errp);
182 int (*bdrv_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi);
183 ImageInfoSpecific *(*bdrv_get_specific_info)(BlockDriverState *bs);
185 int (*bdrv_save_vmstate)(BlockDriverState *bs, QEMUIOVector *qiov,
186 int64_t pos);
187 int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf,
188 int64_t pos, int size);
190 int (*bdrv_change_backing_file)(BlockDriverState *bs,
191 const char *backing_file, const char *backing_fmt);
193 /* removable device specific */
194 int (*bdrv_is_inserted)(BlockDriverState *bs);
195 int (*bdrv_media_changed)(BlockDriverState *bs);
196 void (*bdrv_eject)(BlockDriverState *bs, bool eject_flag);
197 void (*bdrv_lock_medium)(BlockDriverState *bs, bool locked);
199 /* to control generic scsi devices */
200 int (*bdrv_ioctl)(BlockDriverState *bs, unsigned long int req, void *buf);
201 BlockDriverAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs,
202 unsigned long int req, void *buf,
203 BlockDriverCompletionFunc *cb, void *opaque);
205 /* List of options for creating images, terminated by name == NULL */
206 QEMUOptionParameter *create_options;
210 * Returns 0 for completed check, -errno for internal errors.
211 * The check results are stored in result.
213 int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result,
214 BdrvCheckMode fix);
216 int (*bdrv_amend_options)(BlockDriverState *bs,
217 QEMUOptionParameter *options);
219 void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event);
221 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
222 int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event,
223 const char *tag);
224 int (*bdrv_debug_remove_breakpoint)(BlockDriverState *bs,
225 const char *tag);
226 int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag);
227 bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag);
230 * Returns 1 if newly created images are guaranteed to contain only
231 * zeros, 0 otherwise.
233 int (*bdrv_has_zero_init)(BlockDriverState *bs);
235 QLIST_ENTRY(BlockDriver) list;
238 typedef struct BlockLimits {
239 /* maximum number of sectors that can be discarded at once */
240 int max_discard;
242 /* optimal alignment for discard requests in sectors */
243 int64_t discard_alignment;
245 /* maximum number of sectors that can zeroized at once */
246 int max_write_zeroes;
248 /* optimal alignment for write zeroes requests in sectors */
249 int64_t write_zeroes_alignment;
250 } BlockLimits;
253 * Note: the function bdrv_append() copies and swaps contents of
254 * BlockDriverStates, so if you add new fields to this struct, please
255 * inspect bdrv_append() to determine if the new fields need to be
256 * copied as well.
258 struct BlockDriverState {
259 int64_t total_sectors; /* if we are reading a disk image, give its
260 size in sectors */
261 int read_only; /* if true, the media is read only */
262 int open_flags; /* flags used to open the file, re-used for re-open */
263 int encrypted; /* if true, the media is encrypted */
264 int valid_key; /* if true, a valid encryption key has been set */
265 int sg; /* if true, the device is a /dev/sg* */
266 int copy_on_read; /* if true, copy read backing sectors into image
267 note this is a reference count */
269 BlockDriver *drv; /* NULL means no media */
270 void *opaque;
272 void *dev; /* attached device model, if any */
273 /* TODO change to DeviceState when all users are qdevified */
274 const BlockDevOps *dev_ops;
275 void *dev_opaque;
277 char filename[1024];
278 char backing_file[1024]; /* if non zero, the image is a diff of
279 this file image */
280 char backing_format[16]; /* if non-zero and backing_file exists */
281 int is_temporary;
283 BlockDriverState *backing_hd;
284 BlockDriverState *file;
286 NotifierList close_notifiers;
288 /* Callback before write request is processed */
289 NotifierWithReturnList before_write_notifiers;
291 /* number of in-flight copy-on-read requests */
292 unsigned int copy_on_read_in_flight;
294 /* I/O throttling */
295 ThrottleState throttle_state;
296 CoQueue throttled_reqs[2];
297 bool io_limits_enabled;
299 /* I/O stats (display with "info blockstats"). */
300 uint64_t nr_bytes[BDRV_MAX_IOTYPE];
301 uint64_t nr_ops[BDRV_MAX_IOTYPE];
302 uint64_t total_time_ns[BDRV_MAX_IOTYPE];
303 uint64_t wr_highest_sector;
305 /* I/O Limits */
306 BlockLimits bl;
308 /* Whether the disk can expand beyond total_sectors */
309 int growable;
311 /* Whether produces zeros when read beyond eof */
312 bool zero_beyond_eof;
314 /* the memory alignment required for the buffers handled by this driver */
315 int buffer_alignment;
317 /* do we need to tell the quest if we have a volatile write cache? */
318 int enable_write_cache;
320 /* NOTE: the following infos are only hints for real hardware
321 drivers. They are not used by the block driver */
322 BlockdevOnError on_read_error, on_write_error;
323 bool iostatus_enabled;
324 BlockDeviceIoStatus iostatus;
325 char device_name[32];
326 QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
327 int refcnt;
328 int in_use; /* users other than guest access, eg. block migration */
329 QTAILQ_ENTRY(BlockDriverState) list;
331 QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
333 /* long-running background operation */
334 BlockJob *job;
336 QDict *options;
339 int get_tmp_filename(char *filename, int size);
341 void bdrv_set_io_limits(BlockDriverState *bs,
342 ThrottleConfig *cfg);
346 * bdrv_add_before_write_notifier:
348 * Register a callback that is invoked before write requests are processed but
349 * after any throttling or waiting for overlapping requests.
351 void bdrv_add_before_write_notifier(BlockDriverState *bs,
352 NotifierWithReturn *notifier);
355 * bdrv_get_aio_context:
357 * Returns: the currently bound #AioContext
359 AioContext *bdrv_get_aio_context(BlockDriverState *bs);
361 #ifdef _WIN32
362 int is_windows_drive(const char *filename);
363 #endif
364 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
365 enum MonitorEvent ev,
366 BlockErrorAction action, bool is_read);
369 * stream_start:
370 * @bs: Block device to operate on.
371 * @base: Block device that will become the new base, or %NULL to
372 * flatten the whole backing file chain onto @bs.
373 * @base_id: The file name that will be written to @bs as the new
374 * backing file if the job completes. Ignored if @base is %NULL.
375 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
376 * @on_error: The action to take upon error.
377 * @cb: Completion function for the job.
378 * @opaque: Opaque pointer value passed to @cb.
379 * @errp: Error object.
381 * Start a streaming operation on @bs. Clusters that are unallocated
382 * in @bs, but allocated in any image between @base and @bs (both
383 * exclusive) will be written to @bs. At the end of a successful
384 * streaming job, the backing file of @bs will be changed to
385 * @base_id in the written image and to @base in the live BlockDriverState.
387 void stream_start(BlockDriverState *bs, BlockDriverState *base,
388 const char *base_id, int64_t speed, BlockdevOnError on_error,
389 BlockDriverCompletionFunc *cb,
390 void *opaque, Error **errp);
393 * commit_start:
394 * @bs: Top Block device
395 * @base: Block device that will be written into, and become the new top
396 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
397 * @on_error: The action to take upon error.
398 * @cb: Completion function for the job.
399 * @opaque: Opaque pointer value passed to @cb.
400 * @errp: Error object.
403 void commit_start(BlockDriverState *bs, BlockDriverState *base,
404 BlockDriverState *top, int64_t speed,
405 BlockdevOnError on_error, BlockDriverCompletionFunc *cb,
406 void *opaque, Error **errp);
409 * mirror_start:
410 * @bs: Block device to operate on.
411 * @target: Block device to write to.
412 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
413 * @granularity: The chosen granularity for the dirty bitmap.
414 * @buf_size: The amount of data that can be in flight at one time.
415 * @mode: Whether to collapse all images in the chain to the target.
416 * @on_source_error: The action to take upon error reading from the source.
417 * @on_target_error: The action to take upon error writing to the target.
418 * @cb: Completion function for the job.
419 * @opaque: Opaque pointer value passed to @cb.
420 * @errp: Error object.
422 * Start a mirroring operation on @bs. Clusters that are allocated
423 * in @bs will be written to @bs until the job is cancelled or
424 * manually completed. At the end of a successful mirroring job,
425 * @bs will be switched to read from @target.
427 void mirror_start(BlockDriverState *bs, BlockDriverState *target,
428 int64_t speed, int64_t granularity, int64_t buf_size,
429 MirrorSyncMode mode, BlockdevOnError on_source_error,
430 BlockdevOnError on_target_error,
431 BlockDriverCompletionFunc *cb,
432 void *opaque, Error **errp);
435 * backup_start:
436 * @bs: Block device to operate on.
437 * @target: Block device to write to.
438 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
439 * @sync_mode: What parts of the disk image should be copied to the destination.
440 * @on_source_error: The action to take upon error reading from the source.
441 * @on_target_error: The action to take upon error writing to the target.
442 * @cb: Completion function for the job.
443 * @opaque: Opaque pointer value passed to @cb.
445 * Start a backup operation on @bs. Clusters in @bs are written to @target
446 * until the job is cancelled or manually completed.
448 void backup_start(BlockDriverState *bs, BlockDriverState *target,
449 int64_t speed, MirrorSyncMode sync_mode,
450 BlockdevOnError on_source_error,
451 BlockdevOnError on_target_error,
452 BlockDriverCompletionFunc *cb, void *opaque,
453 Error **errp);
455 #endif /* BLOCK_INT_H */