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 37 #define BLOCK_FLAG_ENCRYPT 1 38 #define BLOCK_FLAG_COMPAT6 4 39 #define BLOCK_FLAG_LAZY_REFCOUNTS 8 40 41 #define BLOCK_IO_LIMIT_READ 0 42 #define BLOCK_IO_LIMIT_WRITE 1 43 #define BLOCK_IO_LIMIT_TOTAL 2 44 45 #define BLOCK_IO_SLICE_TIME 100000000 46 #define NANOSECONDS_PER_SECOND 1000000000.0 47 48 #define BLOCK_OPT_SIZE "size" 49 #define BLOCK_OPT_ENCRYPT "encryption" 50 #define BLOCK_OPT_COMPAT6 "compat6" 51 #define BLOCK_OPT_BACKING_FILE "backing_file" 52 #define BLOCK_OPT_BACKING_FMT "backing_fmt" 53 #define BLOCK_OPT_CLUSTER_SIZE "cluster_size" 54 #define BLOCK_OPT_TABLE_SIZE "table_size" 55 #define BLOCK_OPT_PREALLOC "preallocation" 56 #define BLOCK_OPT_SUBFMT "subformat" 57 #define BLOCK_OPT_COMPAT_LEVEL "compat" 58 #define BLOCK_OPT_LAZY_REFCOUNTS "lazy_refcounts" 59 #define BLOCK_OPT_ADAPTER_TYPE "adapter_type" 60 61 typedef struct BdrvTrackedRequest BdrvTrackedRequest; 62 63 typedef struct BlockIOLimit { 64 int64_t bps[3]; 65 int64_t iops[3]; 66 } BlockIOLimit; 67 68 typedef struct BlockIOBaseValue { 69 uint64_t bytes[2]; 70 uint64_t ios[2]; 71 } BlockIOBaseValue; 72 73 struct BlockDriver { 74 const char *format_name; 75 int instance_size; 76 int (*bdrv_probe)(const uint8_t *buf, int buf_size, const char *filename); 77 int (*bdrv_probe_device)(const char *filename); 78 79 /* Any driver implementing this callback is expected to be able to handle 80 * NULL file names in its .bdrv_open() implementation */ 81 void (*bdrv_parse_filename)(const char *filename, QDict *options, Error **errp); 82 83 /* For handling image reopen for split or non-split files */ 84 int (*bdrv_reopen_prepare)(BDRVReopenState *reopen_state, 85 BlockReopenQueue *queue, Error **errp); 86 void (*bdrv_reopen_commit)(BDRVReopenState *reopen_state); 87 void (*bdrv_reopen_abort)(BDRVReopenState *reopen_state); 88 89 int (*bdrv_open)(BlockDriverState *bs, QDict *options, int flags); 90 int (*bdrv_file_open)(BlockDriverState *bs, const char *filename, 91 QDict *options, int flags); 92 int (*bdrv_read)(BlockDriverState *bs, int64_t sector_num, 93 uint8_t *buf, int nb_sectors); 94 int (*bdrv_write)(BlockDriverState *bs, int64_t sector_num, 95 const uint8_t *buf, int nb_sectors); 96 void (*bdrv_close)(BlockDriverState *bs); 97 void (*bdrv_rebind)(BlockDriverState *bs); 98 int (*bdrv_create)(const char *filename, QEMUOptionParameter *options); 99 int (*bdrv_set_key)(BlockDriverState *bs, const char *key); 100 int (*bdrv_make_empty)(BlockDriverState *bs); 101 /* aio */ 102 BlockDriverAIOCB *(*bdrv_aio_readv)(BlockDriverState *bs, 103 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors, 104 BlockDriverCompletionFunc *cb, void *opaque); 105 BlockDriverAIOCB *(*bdrv_aio_writev)(BlockDriverState *bs, 106 int64_t sector_num, QEMUIOVector *qiov, int nb_sectors, 107 BlockDriverCompletionFunc *cb, void *opaque); 108 BlockDriverAIOCB *(*bdrv_aio_flush)(BlockDriverState *bs, 109 BlockDriverCompletionFunc *cb, void *opaque); 110 BlockDriverAIOCB *(*bdrv_aio_discard)(BlockDriverState *bs, 111 int64_t sector_num, int nb_sectors, 112 BlockDriverCompletionFunc *cb, void *opaque); 113 114 int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs, 115 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov); 116 int coroutine_fn (*bdrv_co_writev)(BlockDriverState *bs, 117 int64_t sector_num, int nb_sectors, QEMUIOVector *qiov); 118 /* 119 * Efficiently zero a region of the disk image. Typically an image format 120 * would use a compact metadata representation to implement this. This 121 * function pointer may be NULL and .bdrv_co_writev() will be called 122 * instead. 123 */ 124 int coroutine_fn (*bdrv_co_write_zeroes)(BlockDriverState *bs, 125 int64_t sector_num, int nb_sectors); 126 int coroutine_fn (*bdrv_co_discard)(BlockDriverState *bs, 127 int64_t sector_num, int nb_sectors); 128 int coroutine_fn (*bdrv_co_is_allocated)(BlockDriverState *bs, 129 int64_t sector_num, int nb_sectors, int *pnum); 130 131 /* 132 * Invalidate any cached meta-data. 133 */ 134 void (*bdrv_invalidate_cache)(BlockDriverState *bs); 135 136 /* 137 * Flushes all data that was already written to the OS all the way down to 138 * the disk (for example raw-posix calls fsync()). 139 */ 140 int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs); 141 142 /* 143 * Flushes all internal caches to the OS. The data may still sit in a 144 * writeback cache of the host OS, but it will survive a crash of the qemu 145 * process. 146 */ 147 int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs); 148 149 const char *protocol_name; 150 int (*bdrv_truncate)(BlockDriverState *bs, int64_t offset); 151 int64_t (*bdrv_getlength)(BlockDriverState *bs); 152 int64_t (*bdrv_get_allocated_file_size)(BlockDriverState *bs); 153 int (*bdrv_write_compressed)(BlockDriverState *bs, int64_t sector_num, 154 const uint8_t *buf, int nb_sectors); 155 156 int (*bdrv_snapshot_create)(BlockDriverState *bs, 157 QEMUSnapshotInfo *sn_info); 158 int (*bdrv_snapshot_goto)(BlockDriverState *bs, 159 const char *snapshot_id); 160 int (*bdrv_snapshot_delete)(BlockDriverState *bs, const char *snapshot_id); 161 int (*bdrv_snapshot_list)(BlockDriverState *bs, 162 QEMUSnapshotInfo **psn_info); 163 int (*bdrv_snapshot_load_tmp)(BlockDriverState *bs, 164 const char *snapshot_name); 165 int (*bdrv_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi); 166 167 int (*bdrv_save_vmstate)(BlockDriverState *bs, QEMUIOVector *qiov, 168 int64_t pos); 169 int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf, 170 int64_t pos, int size); 171 172 int (*bdrv_change_backing_file)(BlockDriverState *bs, 173 const char *backing_file, const char *backing_fmt); 174 175 /* removable device specific */ 176 int (*bdrv_is_inserted)(BlockDriverState *bs); 177 int (*bdrv_media_changed)(BlockDriverState *bs); 178 void (*bdrv_eject)(BlockDriverState *bs, bool eject_flag); 179 void (*bdrv_lock_medium)(BlockDriverState *bs, bool locked); 180 181 /* to control generic scsi devices */ 182 int (*bdrv_ioctl)(BlockDriverState *bs, unsigned long int req, void *buf); 183 BlockDriverAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs, 184 unsigned long int req, void *buf, 185 BlockDriverCompletionFunc *cb, void *opaque); 186 187 /* List of options for creating images, terminated by name == NULL */ 188 QEMUOptionParameter *create_options; 189 190 191 /* 192 * Returns 0 for completed check, -errno for internal errors. 193 * The check results are stored in result. 194 */ 195 int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result, 196 BdrvCheckMode fix); 197 198 void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event); 199 200 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */ 201 int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event, 202 const char *tag); 203 int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag); 204 bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag); 205 206 /* 207 * Returns 1 if newly created images are guaranteed to contain only 208 * zeros, 0 otherwise. 209 */ 210 int (*bdrv_has_zero_init)(BlockDriverState *bs); 211 212 QLIST_ENTRY(BlockDriver) list; 213 }; 214 215 /* 216 * Note: the function bdrv_append() copies and swaps contents of 217 * BlockDriverStates, so if you add new fields to this struct, please 218 * inspect bdrv_append() to determine if the new fields need to be 219 * copied as well. 220 */ 221 struct BlockDriverState { 222 int64_t total_sectors; /* if we are reading a disk image, give its 223 size in sectors */ 224 int read_only; /* if true, the media is read only */ 225 int open_flags; /* flags used to open the file, re-used for re-open */ 226 int encrypted; /* if true, the media is encrypted */ 227 int valid_key; /* if true, a valid encryption key has been set */ 228 int sg; /* if true, the device is a /dev/sg* */ 229 int copy_on_read; /* if true, copy read backing sectors into image 230 note this is a reference count */ 231 232 BlockDriver *drv; /* NULL means no media */ 233 void *opaque; 234 235 void *dev; /* attached device model, if any */ 236 /* TODO change to DeviceState when all users are qdevified */ 237 const BlockDevOps *dev_ops; 238 void *dev_opaque; 239 240 char filename[1024]; 241 char backing_file[1024]; /* if non zero, the image is a diff of 242 this file image */ 243 char backing_format[16]; /* if non-zero and backing_file exists */ 244 int is_temporary; 245 246 BlockDriverState *backing_hd; 247 BlockDriverState *file; 248 249 NotifierList close_notifiers; 250 251 /* number of in-flight copy-on-read requests */ 252 unsigned int copy_on_read_in_flight; 253 254 /* the time for latest disk I/O */ 255 int64_t slice_start; 256 int64_t slice_end; 257 BlockIOLimit io_limits; 258 BlockIOBaseValue slice_submitted; 259 CoQueue throttled_reqs; 260 QEMUTimer *block_timer; 261 bool io_limits_enabled; 262 263 /* I/O stats (display with "info blockstats"). */ 264 uint64_t nr_bytes[BDRV_MAX_IOTYPE]; 265 uint64_t nr_ops[BDRV_MAX_IOTYPE]; 266 uint64_t total_time_ns[BDRV_MAX_IOTYPE]; 267 uint64_t wr_highest_sector; 268 269 /* Whether the disk can expand beyond total_sectors */ 270 int growable; 271 272 /* the memory alignment required for the buffers handled by this driver */ 273 int buffer_alignment; 274 275 /* do we need to tell the quest if we have a volatile write cache? */ 276 int enable_write_cache; 277 278 /* NOTE: the following infos are only hints for real hardware 279 drivers. They are not used by the block driver */ 280 BlockdevOnError on_read_error, on_write_error; 281 bool iostatus_enabled; 282 BlockDeviceIoStatus iostatus; 283 char device_name[32]; 284 HBitmap *dirty_bitmap; 285 int in_use; /* users other than guest access, eg. block migration */ 286 QTAILQ_ENTRY(BlockDriverState) list; 287 288 QLIST_HEAD(, BdrvTrackedRequest) tracked_requests; 289 290 /* long-running background operation */ 291 BlockJob *job; 292 293 QDict *options; 294 }; 295 296 int get_tmp_filename(char *filename, int size); 297 298 void bdrv_set_io_limits(BlockDriverState *bs, 299 BlockIOLimit *io_limits); 300 301 /** 302 * bdrv_get_aio_context: 303 * 304 * Returns: the currently bound #AioContext 305 */ 306 AioContext *bdrv_get_aio_context(BlockDriverState *bs); 307 308 #ifdef _WIN32 309 int is_windows_drive(const char *filename); 310 #endif 311 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv, 312 enum MonitorEvent ev, 313 BlockErrorAction action, bool is_read); 314 315 /** 316 * stream_start: 317 * @bs: Block device to operate on. 318 * @base: Block device that will become the new base, or %NULL to 319 * flatten the whole backing file chain onto @bs. 320 * @base_id: The file name that will be written to @bs as the new 321 * backing file if the job completes. Ignored if @base is %NULL. 322 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 323 * @on_error: The action to take upon error. 324 * @cb: Completion function for the job. 325 * @opaque: Opaque pointer value passed to @cb. 326 * @errp: Error object. 327 * 328 * Start a streaming operation on @bs. Clusters that are unallocated 329 * in @bs, but allocated in any image between @base and @bs (both 330 * exclusive) will be written to @bs. At the end of a successful 331 * streaming job, the backing file of @bs will be changed to 332 * @base_id in the written image and to @base in the live BlockDriverState. 333 */ 334 void stream_start(BlockDriverState *bs, BlockDriverState *base, 335 const char *base_id, int64_t speed, BlockdevOnError on_error, 336 BlockDriverCompletionFunc *cb, 337 void *opaque, Error **errp); 338 339 /** 340 * commit_start: 341 * @bs: Top Block device 342 * @base: Block device that will be written into, and become the new top 343 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 344 * @on_error: The action to take upon error. 345 * @cb: Completion function for the job. 346 * @opaque: Opaque pointer value passed to @cb. 347 * @errp: Error object. 348 * 349 */ 350 void commit_start(BlockDriverState *bs, BlockDriverState *base, 351 BlockDriverState *top, int64_t speed, 352 BlockdevOnError on_error, BlockDriverCompletionFunc *cb, 353 void *opaque, Error **errp); 354 355 /* 356 * mirror_start: 357 * @bs: Block device to operate on. 358 * @target: Block device to write to. 359 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 360 * @granularity: The chosen granularity for the dirty bitmap. 361 * @buf_size: The amount of data that can be in flight at one time. 362 * @mode: Whether to collapse all images in the chain to the target. 363 * @on_source_error: The action to take upon error reading from the source. 364 * @on_target_error: The action to take upon error writing to the target. 365 * @cb: Completion function for the job. 366 * @opaque: Opaque pointer value passed to @cb. 367 * @errp: Error object. 368 * 369 * Start a mirroring operation on @bs. Clusters that are allocated 370 * in @bs will be written to @bs until the job is cancelled or 371 * manually completed. At the end of a successful mirroring job, 372 * @bs will be switched to read from @target. 373 */ 374 void mirror_start(BlockDriverState *bs, BlockDriverState *target, 375 int64_t speed, int64_t granularity, int64_t buf_size, 376 MirrorSyncMode mode, BlockdevOnError on_source_error, 377 BlockdevOnError on_target_error, 378 BlockDriverCompletionFunc *cb, 379 void *opaque, Error **errp); 380 381 #endif /* BLOCK_INT_H */ 382