1 /* 2 * QEMU System Emulator block driver 3 * 4 * Copyright (c) 2003 Fabrice Bellard 5 * 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: 12 * 13 * The above copyright notice and this permission notice shall be included in 14 * all copies or substantial portions of the Software. 15 * 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. 23 */ 24 #ifndef BLOCK_INT_H 25 #define BLOCK_INT_H 26 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" 39 40 #define BLOCK_FLAG_ENCRYPT 1 41 #define BLOCK_FLAG_COMPAT6 4 42 #define BLOCK_FLAG_LAZY_REFCOUNTS 8 43 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 57 typedef struct BdrvTrackedRequest { 58 BlockDriverState *bs; 59 int64_t sector_num; 60 int nb_sectors; 61 bool is_write; 62 QLIST_ENTRY(BdrvTrackedRequest) list; 63 Coroutine *co; /* owner, used for deadlock detection */ 64 CoQueue wait_queue; /* coroutines blocked on this request */ 65 } BdrvTrackedRequest; 66 67 struct BlockDriver { 68 const char *format_name; 69 int instance_size; 70 int (*bdrv_probe)(const uint8_t *buf, int buf_size, const char *filename); 71 int (*bdrv_probe_device)(const char *filename); 72 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); 76 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); 82 83 int (*bdrv_open)(BlockDriverState *bs, QDict *options, int flags, 84 Error **errp); 85 int (*bdrv_file_open)(BlockDriverState *bs, QDict *options, int flags, 86 Error **errp); 87 int (*bdrv_read)(BlockDriverState *bs, int64_t sector_num, 88 uint8_t *buf, int nb_sectors); 89 int (*bdrv_write)(BlockDriverState *bs, int64_t sector_num, 90 const uint8_t *buf, int nb_sectors); 91 void (*bdrv_close)(BlockDriverState *bs); 92 void (*bdrv_rebind)(BlockDriverState *bs); 93 int (*bdrv_create)(const char *filename, QEMUOptionParameter *options, 94 Error **errp); 95 int (*bdrv_set_key)(BlockDriverState *bs, const char *key); 96 int (*bdrv_make_empty)(BlockDriverState *bs); 97 /* aio */ 98 BlockDriverAIOCB *(*bdrv_aio_readv)(BlockDriverState *bs, 99 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors, 100 BlockDriverCompletionFunc *cb, void *opaque); 101 BlockDriverAIOCB *(*bdrv_aio_writev)(BlockDriverState *bs, 102 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors, 103 BlockDriverCompletionFunc *cb, void *opaque); 104 BlockDriverAIOCB *(*bdrv_aio_flush)(BlockDriverState *bs, 105 BlockDriverCompletionFunc *cb, void *opaque); 106 BlockDriverAIOCB *(*bdrv_aio_discard)(BlockDriverState *bs, 107 int64_t sector_num, int nb_sectors, 108 BlockDriverCompletionFunc *cb, void *opaque); 109 110 int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs, 111 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov); 112 int coroutine_fn (*bdrv_co_writev)(BlockDriverState *bs, 113 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov); 114 /* 115 * Efficiently zero a region of the disk image. Typically an image format 116 * would use a compact metadata representation to implement this. This 117 * function pointer may be NULL and .bdrv_co_writev() will be called 118 * instead. 119 */ 120 int coroutine_fn (*bdrv_co_write_zeroes)(BlockDriverState *bs, 121 int64_t sector_num, int nb_sectors); 122 int coroutine_fn (*bdrv_co_discard)(BlockDriverState *bs, 123 int64_t sector_num, int nb_sectors); 124 int64_t coroutine_fn (*bdrv_co_get_block_status)(BlockDriverState *bs, 125 int64_t sector_num, int nb_sectors, int *pnum); 126 127 /* 128 * Invalidate any cached meta-data. 129 */ 130 void (*bdrv_invalidate_cache)(BlockDriverState *bs); 131 132 /* 133 * Flushes all data that was already written to the OS all the way down to 134 * the disk (for example raw-posix calls fsync()). 135 */ 136 int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs); 137 138 /* 139 * Flushes all internal caches to the OS. The data may still sit in a 140 * writeback cache of the host OS, but it will survive a crash of the qemu 141 * process. 142 */ 143 int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs); 144 145 const char *protocol_name; 146 int (*bdrv_truncate)(BlockDriverState *bs, int64_t offset); 147 int64_t (*bdrv_getlength)(BlockDriverState *bs); 148 int64_t (*bdrv_get_allocated_file_size)(BlockDriverState *bs); 149 int (*bdrv_write_compressed)(BlockDriverState *bs, int64_t sector_num, 150 const uint8_t *buf, int nb_sectors); 151 152 int (*bdrv_snapshot_create)(BlockDriverState *bs, 153 QEMUSnapshotInfo *sn_info); 154 int (*bdrv_snapshot_goto)(BlockDriverState *bs, 155 const char *snapshot_id); 156 int (*bdrv_snapshot_delete)(BlockDriverState *bs, 157 const char *snapshot_id, 158 const char *name, 159 Error **errp); 160 int (*bdrv_snapshot_list)(BlockDriverState *bs, 161 QEMUSnapshotInfo **psn_info); 162 int (*bdrv_snapshot_load_tmp)(BlockDriverState *bs, 163 const char *snapshot_name); 164 int (*bdrv_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi); 165 166 int (*bdrv_save_vmstate)(BlockDriverState *bs, QEMUIOVector *qiov, 167 int64_t pos); 168 int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf, 169 int64_t pos, int size); 170 171 int (*bdrv_change_backing_file)(BlockDriverState *bs, 172 const char *backing_file, const char *backing_fmt); 173 174 /* removable device specific */ 175 int (*bdrv_is_inserted)(BlockDriverState *bs); 176 int (*bdrv_media_changed)(BlockDriverState *bs); 177 void (*bdrv_eject)(BlockDriverState *bs, bool eject_flag); 178 void (*bdrv_lock_medium)(BlockDriverState *bs, bool locked); 179 180 /* to control generic scsi devices */ 181 int (*bdrv_ioctl)(BlockDriverState *bs, unsigned long int req, void *buf); 182 BlockDriverAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs, 183 unsigned long int req, void *buf, 184 BlockDriverCompletionFunc *cb, void *opaque); 185 186 /* List of options for creating images, terminated by name == NULL */ 187 QEMUOptionParameter *create_options; 188 189 190 /* 191 * Returns 0 for completed check, -errno for internal errors. 192 * The check results are stored in result. 193 */ 194 int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result, 195 BdrvCheckMode fix); 196 197 int (*bdrv_amend_options)(BlockDriverState *bs, 198 QEMUOptionParameter *options); 199 200 void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event); 201 202 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */ 203 int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event, 204 const char *tag); 205 int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag); 206 bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag); 207 208 /* 209 * Returns 1 if newly created images are guaranteed to contain only 210 * zeros, 0 otherwise. 211 */ 212 int (*bdrv_has_zero_init)(BlockDriverState *bs); 213 214 QLIST_ENTRY(BlockDriver) list; 215 }; 216 217 /* 218 * Note: the function bdrv_append() copies and swaps contents of 219 * BlockDriverStates, so if you add new fields to this struct, please 220 * inspect bdrv_append() to determine if the new fields need to be 221 * copied as well. 222 */ 223 struct BlockDriverState { 224 int64_t total_sectors; /* if we are reading a disk image, give its 225 size in sectors */ 226 int read_only; /* if true, the media is read only */ 227 int open_flags; /* flags used to open the file, re-used for re-open */ 228 int encrypted; /* if true, the media is encrypted */ 229 int valid_key; /* if true, a valid encryption key has been set */ 230 int sg; /* if true, the device is a /dev/sg* */ 231 int copy_on_read; /* if true, copy read backing sectors into image 232 note this is a reference count */ 233 234 BlockDriver *drv; /* NULL means no media */ 235 void *opaque; 236 237 void *dev; /* attached device model, if any */ 238 /* TODO change to DeviceState when all users are qdevified */ 239 const BlockDevOps *dev_ops; 240 void *dev_opaque; 241 242 char filename[1024]; 243 char backing_file[1024]; /* if non zero, the image is a diff of 244 this file image */ 245 char backing_format[16]; /* if non-zero and backing_file exists */ 246 int is_temporary; 247 248 BlockDriverState *backing_hd; 249 BlockDriverState *file; 250 251 NotifierList close_notifiers; 252 253 /* Callback before write request is processed */ 254 NotifierWithReturnList before_write_notifiers; 255 256 /* number of in-flight copy-on-read requests */ 257 unsigned int copy_on_read_in_flight; 258 259 /* I/O throttling */ 260 ThrottleState throttle_state; 261 CoQueue throttled_reqs[2]; 262 bool io_limits_enabled; 263 264 /* I/O stats (display with "info blockstats"). */ 265 uint64_t nr_bytes[BDRV_MAX_IOTYPE]; 266 uint64_t nr_ops[BDRV_MAX_IOTYPE]; 267 uint64_t total_time_ns[BDRV_MAX_IOTYPE]; 268 uint64_t wr_highest_sector; 269 270 /* Whether the disk can expand beyond total_sectors */ 271 int growable; 272 273 /* Whether produces zeros when read beyond eof */ 274 bool zero_beyond_eof; 275 276 /* the memory alignment required for the buffers handled by this driver */ 277 int buffer_alignment; 278 279 /* do we need to tell the quest if we have a volatile write cache? */ 280 int enable_write_cache; 281 282 /* NOTE: the following infos are only hints for real hardware 283 drivers. They are not used by the block driver */ 284 BlockdevOnError on_read_error, on_write_error; 285 bool iostatus_enabled; 286 BlockDeviceIoStatus iostatus; 287 char device_name[32]; 288 HBitmap *dirty_bitmap; 289 int refcnt; 290 int in_use; /* users other than guest access, eg. block migration */ 291 QTAILQ_ENTRY(BlockDriverState) list; 292 293 QLIST_HEAD(, BdrvTrackedRequest) tracked_requests; 294 295 /* long-running background operation */ 296 BlockJob *job; 297 298 QDict *options; 299 }; 300 301 int get_tmp_filename(char *filename, int size); 302 303 void bdrv_set_io_limits(BlockDriverState *bs, 304 ThrottleConfig *cfg); 305 306 307 /** 308 * bdrv_add_before_write_notifier: 309 * 310 * Register a callback that is invoked before write requests are processed but 311 * after any throttling or waiting for overlapping requests. 312 */ 313 void bdrv_add_before_write_notifier(BlockDriverState *bs, 314 NotifierWithReturn *notifier); 315 316 /** 317 * bdrv_get_aio_context: 318 * 319 * Returns: the currently bound #AioContext 320 */ 321 AioContext *bdrv_get_aio_context(BlockDriverState *bs); 322 323 #ifdef _WIN32 324 int is_windows_drive(const char *filename); 325 #endif 326 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv, 327 enum MonitorEvent ev, 328 BlockErrorAction action, bool is_read); 329 330 /** 331 * stream_start: 332 * @bs: Block device to operate on. 333 * @base: Block device that will become the new base, or %NULL to 334 * flatten the whole backing file chain onto @bs. 335 * @base_id: The file name that will be written to @bs as the new 336 * backing file if the job completes. Ignored if @base is %NULL. 337 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 338 * @on_error: The action to take upon error. 339 * @cb: Completion function for the job. 340 * @opaque: Opaque pointer value passed to @cb. 341 * @errp: Error object. 342 * 343 * Start a streaming operation on @bs. Clusters that are unallocated 344 * in @bs, but allocated in any image between @base and @bs (both 345 * exclusive) will be written to @bs. At the end of a successful 346 * streaming job, the backing file of @bs will be changed to 347 * @base_id in the written image and to @base in the live BlockDriverState. 348 */ 349 void stream_start(BlockDriverState *bs, BlockDriverState *base, 350 const char *base_id, int64_t speed, BlockdevOnError on_error, 351 BlockDriverCompletionFunc *cb, 352 void *opaque, Error **errp); 353 354 /** 355 * commit_start: 356 * @bs: Top Block device 357 * @base: Block device that will be written into, and become the new top 358 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 359 * @on_error: The action to take upon error. 360 * @cb: Completion function for the job. 361 * @opaque: Opaque pointer value passed to @cb. 362 * @errp: Error object. 363 * 364 */ 365 void commit_start(BlockDriverState *bs, BlockDriverState *base, 366 BlockDriverState *top, int64_t speed, 367 BlockdevOnError on_error, BlockDriverCompletionFunc *cb, 368 void *opaque, Error **errp); 369 370 /* 371 * mirror_start: 372 * @bs: Block device to operate on. 373 * @target: Block device to write to. 374 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 375 * @granularity: The chosen granularity for the dirty bitmap. 376 * @buf_size: The amount of data that can be in flight at one time. 377 * @mode: Whether to collapse all images in the chain to the target. 378 * @on_source_error: The action to take upon error reading from the source. 379 * @on_target_error: The action to take upon error writing to the target. 380 * @cb: Completion function for the job. 381 * @opaque: Opaque pointer value passed to @cb. 382 * @errp: Error object. 383 * 384 * Start a mirroring operation on @bs. Clusters that are allocated 385 * in @bs will be written to @bs until the job is cancelled or 386 * manually completed. At the end of a successful mirroring job, 387 * @bs will be switched to read from @target. 388 */ 389 void mirror_start(BlockDriverState *bs, BlockDriverState *target, 390 int64_t speed, int64_t granularity, int64_t buf_size, 391 MirrorSyncMode mode, BlockdevOnError on_source_error, 392 BlockdevOnError on_target_error, 393 BlockDriverCompletionFunc *cb, 394 void *opaque, Error **errp); 395 396 /* 397 * backup_start: 398 * @bs: Block device to operate on. 399 * @target: Block device to write to. 400 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 401 * @sync_mode: What parts of the disk image should be copied to the destination. 402 * @on_source_error: The action to take upon error reading from the source. 403 * @on_target_error: The action to take upon error writing to the target. 404 * @cb: Completion function for the job. 405 * @opaque: Opaque pointer value passed to @cb. 406 * 407 * Start a backup operation on @bs. Clusters in @bs are written to @target 408 * until the job is cancelled or manually completed. 409 */ 410 void backup_start(BlockDriverState *bs, BlockDriverState *target, 411 int64_t speed, MirrorSyncMode sync_mode, 412 BlockdevOnError on_source_error, 413 BlockdevOnError on_target_error, 414 BlockDriverCompletionFunc *cb, void *opaque, 415 Error **errp); 416 417 #endif /* BLOCK_INT_H */ 418