sheepdog: support user-defined redundancy option
[qemu/ar7.git] / include / block / block_int.h
blobd0f70c4596ac54af7418d5d596c147766736e428
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_name);
180 int (*bdrv_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi);
181 ImageInfoSpecific *(*bdrv_get_specific_info)(BlockDriverState *bs);
183 int (*bdrv_save_vmstate)(BlockDriverState *bs, QEMUIOVector *qiov,
184 int64_t pos);
185 int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf,
186 int64_t pos, int size);
188 int (*bdrv_change_backing_file)(BlockDriverState *bs,
189 const char *backing_file, const char *backing_fmt);
191 /* removable device specific */
192 int (*bdrv_is_inserted)(BlockDriverState *bs);
193 int (*bdrv_media_changed)(BlockDriverState *bs);
194 void (*bdrv_eject)(BlockDriverState *bs, bool eject_flag);
195 void (*bdrv_lock_medium)(BlockDriverState *bs, bool locked);
197 /* to control generic scsi devices */
198 int (*bdrv_ioctl)(BlockDriverState *bs, unsigned long int req, void *buf);
199 BlockDriverAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs,
200 unsigned long int req, void *buf,
201 BlockDriverCompletionFunc *cb, void *opaque);
203 /* List of options for creating images, terminated by name == NULL */
204 QEMUOptionParameter *create_options;
208 * Returns 0 for completed check, -errno for internal errors.
209 * The check results are stored in result.
211 int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result,
212 BdrvCheckMode fix);
214 int (*bdrv_amend_options)(BlockDriverState *bs,
215 QEMUOptionParameter *options);
217 void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event);
219 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
220 int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event,
221 const char *tag);
222 int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag);
223 bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag);
226 * Returns 1 if newly created images are guaranteed to contain only
227 * zeros, 0 otherwise.
229 int (*bdrv_has_zero_init)(BlockDriverState *bs);
231 QLIST_ENTRY(BlockDriver) list;
234 typedef struct BlockLimits {
235 /* maximum number of sectors that can be discarded at once */
236 int max_discard;
238 /* optimal alignment for discard requests in sectors */
239 int64_t discard_alignment;
241 /* maximum number of sectors that can zeroized at once */
242 int max_write_zeroes;
244 /* optimal alignment for write zeroes requests in sectors */
245 int64_t write_zeroes_alignment;
246 } BlockLimits;
249 * Note: the function bdrv_append() copies and swaps contents of
250 * BlockDriverStates, so if you add new fields to this struct, please
251 * inspect bdrv_append() to determine if the new fields need to be
252 * copied as well.
254 struct BlockDriverState {
255 int64_t total_sectors; /* if we are reading a disk image, give its
256 size in sectors */
257 int read_only; /* if true, the media is read only */
258 int open_flags; /* flags used to open the file, re-used for re-open */
259 int encrypted; /* if true, the media is encrypted */
260 int valid_key; /* if true, a valid encryption key has been set */
261 int sg; /* if true, the device is a /dev/sg* */
262 int copy_on_read; /* if true, copy read backing sectors into image
263 note this is a reference count */
265 BlockDriver *drv; /* NULL means no media */
266 void *opaque;
268 void *dev; /* attached device model, if any */
269 /* TODO change to DeviceState when all users are qdevified */
270 const BlockDevOps *dev_ops;
271 void *dev_opaque;
273 char filename[1024];
274 char backing_file[1024]; /* if non zero, the image is a diff of
275 this file image */
276 char backing_format[16]; /* if non-zero and backing_file exists */
277 int is_temporary;
279 BlockDriverState *backing_hd;
280 BlockDriverState *file;
282 NotifierList close_notifiers;
284 /* Callback before write request is processed */
285 NotifierWithReturnList before_write_notifiers;
287 /* number of in-flight copy-on-read requests */
288 unsigned int copy_on_read_in_flight;
290 /* I/O throttling */
291 ThrottleState throttle_state;
292 CoQueue throttled_reqs[2];
293 bool io_limits_enabled;
295 /* I/O stats (display with "info blockstats"). */
296 uint64_t nr_bytes[BDRV_MAX_IOTYPE];
297 uint64_t nr_ops[BDRV_MAX_IOTYPE];
298 uint64_t total_time_ns[BDRV_MAX_IOTYPE];
299 uint64_t wr_highest_sector;
301 /* I/O Limits */
302 BlockLimits bl;
304 /* Whether the disk can expand beyond total_sectors */
305 int growable;
307 /* Whether produces zeros when read beyond eof */
308 bool zero_beyond_eof;
310 /* the memory alignment required for the buffers handled by this driver */
311 int buffer_alignment;
313 /* do we need to tell the quest if we have a volatile write cache? */
314 int enable_write_cache;
316 /* NOTE: the following infos are only hints for real hardware
317 drivers. They are not used by the block driver */
318 BlockdevOnError on_read_error, on_write_error;
319 bool iostatus_enabled;
320 BlockDeviceIoStatus iostatus;
321 char device_name[32];
322 QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
323 int refcnt;
324 int in_use; /* users other than guest access, eg. block migration */
325 QTAILQ_ENTRY(BlockDriverState) list;
327 QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
329 /* long-running background operation */
330 BlockJob *job;
332 QDict *options;
335 int get_tmp_filename(char *filename, int size);
337 void bdrv_set_io_limits(BlockDriverState *bs,
338 ThrottleConfig *cfg);
342 * bdrv_add_before_write_notifier:
344 * Register a callback that is invoked before write requests are processed but
345 * after any throttling or waiting for overlapping requests.
347 void bdrv_add_before_write_notifier(BlockDriverState *bs,
348 NotifierWithReturn *notifier);
351 * bdrv_get_aio_context:
353 * Returns: the currently bound #AioContext
355 AioContext *bdrv_get_aio_context(BlockDriverState *bs);
357 #ifdef _WIN32
358 int is_windows_drive(const char *filename);
359 #endif
360 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
361 enum MonitorEvent ev,
362 BlockErrorAction action, bool is_read);
365 * stream_start:
366 * @bs: Block device to operate on.
367 * @base: Block device that will become the new base, or %NULL to
368 * flatten the whole backing file chain onto @bs.
369 * @base_id: The file name that will be written to @bs as the new
370 * backing file if the job completes. Ignored if @base is %NULL.
371 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
372 * @on_error: The action to take upon error.
373 * @cb: Completion function for the job.
374 * @opaque: Opaque pointer value passed to @cb.
375 * @errp: Error object.
377 * Start a streaming operation on @bs. Clusters that are unallocated
378 * in @bs, but allocated in any image between @base and @bs (both
379 * exclusive) will be written to @bs. At the end of a successful
380 * streaming job, the backing file of @bs will be changed to
381 * @base_id in the written image and to @base in the live BlockDriverState.
383 void stream_start(BlockDriverState *bs, BlockDriverState *base,
384 const char *base_id, int64_t speed, BlockdevOnError on_error,
385 BlockDriverCompletionFunc *cb,
386 void *opaque, Error **errp);
389 * commit_start:
390 * @bs: Top Block device
391 * @base: Block device that will be written into, and become the new top
392 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
393 * @on_error: The action to take upon error.
394 * @cb: Completion function for the job.
395 * @opaque: Opaque pointer value passed to @cb.
396 * @errp: Error object.
399 void commit_start(BlockDriverState *bs, BlockDriverState *base,
400 BlockDriverState *top, int64_t speed,
401 BlockdevOnError on_error, BlockDriverCompletionFunc *cb,
402 void *opaque, Error **errp);
405 * mirror_start:
406 * @bs: Block device to operate on.
407 * @target: Block device to write to.
408 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
409 * @granularity: The chosen granularity for the dirty bitmap.
410 * @buf_size: The amount of data that can be in flight at one time.
411 * @mode: Whether to collapse all images in the chain to the target.
412 * @on_source_error: The action to take upon error reading from the source.
413 * @on_target_error: The action to take upon error writing to the target.
414 * @cb: Completion function for the job.
415 * @opaque: Opaque pointer value passed to @cb.
416 * @errp: Error object.
418 * Start a mirroring operation on @bs. Clusters that are allocated
419 * in @bs will be written to @bs until the job is cancelled or
420 * manually completed. At the end of a successful mirroring job,
421 * @bs will be switched to read from @target.
423 void mirror_start(BlockDriverState *bs, BlockDriverState *target,
424 int64_t speed, int64_t granularity, int64_t buf_size,
425 MirrorSyncMode mode, BlockdevOnError on_source_error,
426 BlockdevOnError on_target_error,
427 BlockDriverCompletionFunc *cb,
428 void *opaque, Error **errp);
431 * backup_start:
432 * @bs: Block device to operate on.
433 * @target: Block device to write to.
434 * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
435 * @sync_mode: What parts of the disk image should be copied to the destination.
436 * @on_source_error: The action to take upon error reading from the source.
437 * @on_target_error: The action to take upon error writing to the target.
438 * @cb: Completion function for the job.
439 * @opaque: Opaque pointer value passed to @cb.
441 * Start a backup operation on @bs. Clusters in @bs are written to @target
442 * until the job is cancelled or manually completed.
444 void backup_start(BlockDriverState *bs, BlockDriverState *target,
445 int64_t speed, MirrorSyncMode sync_mode,
446 BlockdevOnError on_source_error,
447 BlockdevOnError on_target_error,
448 BlockDriverCompletionFunc *cb, void *opaque,
449 Error **errp);
451 #endif /* BLOCK_INT_H */