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 36 #define BLOCK_FLAG_ENCRYPT 1 37 #define BLOCK_FLAG_COMPAT6 4 38 #define BLOCK_FLAG_LAZY_REFCOUNTS 8 39 40 #define BLOCK_IO_LIMIT_READ 0 41 #define BLOCK_IO_LIMIT_WRITE 1 42 #define BLOCK_IO_LIMIT_TOTAL 2 43 44 #define BLOCK_IO_SLICE_TIME 100000000 45 #define NANOSECONDS_PER_SECOND 1000000000.0 46 47 #define BLOCK_OPT_SIZE "size" 48 #define BLOCK_OPT_ENCRYPT "encryption" 49 #define BLOCK_OPT_COMPAT6 "compat6" 50 #define BLOCK_OPT_BACKING_FILE "backing_file" 51 #define BLOCK_OPT_BACKING_FMT "backing_fmt" 52 #define BLOCK_OPT_CLUSTER_SIZE "cluster_size" 53 #define BLOCK_OPT_TABLE_SIZE "table_size" 54 #define BLOCK_OPT_PREALLOC "preallocation" 55 #define BLOCK_OPT_SUBFMT "subformat" 56 #define BLOCK_OPT_COMPAT_LEVEL "compat" 57 #define BLOCK_OPT_LAZY_REFCOUNTS "lazy_refcounts" 58 59 typedef struct BdrvTrackedRequest BdrvTrackedRequest; 60 61 typedef struct BlockIOLimit { 62 int64_t bps[3]; 63 int64_t iops[3]; 64 } BlockIOLimit; 65 66 typedef struct BlockIOBaseValue { 67 uint64_t bytes[2]; 68 uint64_t ios[2]; 69 } BlockIOBaseValue; 70 71 struct BlockDriver { 72 const char *format_name; 73 int instance_size; 74 int (*bdrv_probe)(const uint8_t *buf, int buf_size, const char *filename); 75 int (*bdrv_probe_device)(const char *filename); 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, int flags); 84 int (*bdrv_file_open)(BlockDriverState *bs, const char *filename, 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); 94 /* aio */ 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); 106 107 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); 111 /* 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 115 * instead. 116 */ 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 int coroutine_fn (*bdrv_co_is_allocated)(BlockDriverState *bs, 122 int64_t sector_num, int nb_sectors, int *pnum); 123 124 /* 125 * Invalidate any cached meta-data. 126 */ 127 void (*bdrv_invalidate_cache)(BlockDriverState *bs); 128 129 /* 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()). 132 */ 133 int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs); 134 135 /* 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 138 * process. 139 */ 140 int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs); 141 142 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); 148 149 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); 159 160 int (*bdrv_save_vmstate)(BlockDriverState *bs, const uint8_t *buf, 161 int64_t pos, int size); 162 int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf, 163 int64_t pos, int size); 164 165 int (*bdrv_change_backing_file)(BlockDriverState *bs, 166 const char *backing_file, const char *backing_fmt); 167 168 /* 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); 173 174 /* 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); 179 180 /* List of options for creating images, terminated by name == NULL */ 181 QEMUOptionParameter *create_options; 182 183 184 /* 185 * Returns 0 for completed check, -errno for internal errors. 186 * The check results are stored in result. 187 */ 188 int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result, 189 BdrvCheckMode fix); 190 191 void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event); 192 193 /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */ 194 int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event, 195 const char *tag); 196 int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag); 197 bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag); 198 199 /* 200 * Returns 1 if newly created images are guaranteed to contain only 201 * zeros, 0 otherwise. 202 */ 203 int (*bdrv_has_zero_init)(BlockDriverState *bs); 204 205 QLIST_ENTRY(BlockDriver) list; 206 }; 207 208 /* 209 * Note: the function bdrv_append() copies and swaps contents of 210 * BlockDriverStates, so if you add new fields to this struct, please 211 * inspect bdrv_append() to determine if the new fields need to be 212 * copied as well. 213 */ 214 struct BlockDriverState { 215 int64_t total_sectors; /* if we are reading a disk image, give its 216 size in sectors */ 217 int read_only; /* if true, the media is read only */ 218 int open_flags; /* flags used to open the file, re-used for re-open */ 219 int encrypted; /* if true, the media is encrypted */ 220 int valid_key; /* if true, a valid encryption key has been set */ 221 int sg; /* if true, the device is a /dev/sg* */ 222 int copy_on_read; /* if true, copy read backing sectors into image 223 note this is a reference count */ 224 225 BlockDriver *drv; /* NULL means no media */ 226 void *opaque; 227 228 void *dev; /* attached device model, if any */ 229 /* TODO change to DeviceState when all users are qdevified */ 230 const BlockDevOps *dev_ops; 231 void *dev_opaque; 232 233 char filename[1024]; 234 char backing_file[1024]; /* if non zero, the image is a diff of 235 this file image */ 236 char backing_format[16]; /* if non-zero and backing_file exists */ 237 int is_temporary; 238 239 BlockDriverState *backing_hd; 240 BlockDriverState *file; 241 242 NotifierList close_notifiers; 243 244 /* number of in-flight copy-on-read requests */ 245 unsigned int copy_on_read_in_flight; 246 247 /* the time for latest disk I/O */ 248 int64_t slice_time; 249 int64_t slice_start; 250 int64_t slice_end; 251 BlockIOLimit io_limits; 252 BlockIOBaseValue io_base; 253 CoQueue throttled_reqs; 254 QEMUTimer *block_timer; 255 bool io_limits_enabled; 256 257 /* I/O stats (display with "info blockstats"). */ 258 uint64_t nr_bytes[BDRV_MAX_IOTYPE]; 259 uint64_t nr_ops[BDRV_MAX_IOTYPE]; 260 uint64_t total_time_ns[BDRV_MAX_IOTYPE]; 261 uint64_t wr_highest_sector; 262 263 /* Whether the disk can expand beyond total_sectors */ 264 int growable; 265 266 /* the memory alignment required for the buffers handled by this driver */ 267 int buffer_alignment; 268 269 /* do we need to tell the quest if we have a volatile write cache? */ 270 int enable_write_cache; 271 272 /* NOTE: the following infos are only hints for real hardware 273 drivers. They are not used by the block driver */ 274 BlockdevOnError on_read_error, on_write_error; 275 bool iostatus_enabled; 276 BlockDeviceIoStatus iostatus; 277 char device_name[32]; 278 unsigned long *dirty_bitmap; 279 int64_t dirty_count; 280 int in_use; /* users other than guest access, eg. block migration */ 281 QTAILQ_ENTRY(BlockDriverState) list; 282 283 QLIST_HEAD(, BdrvTrackedRequest) tracked_requests; 284 285 /* long-running background operation */ 286 BlockJob *job; 287 288 }; 289 290 int get_tmp_filename(char *filename, int size); 291 292 void bdrv_set_io_limits(BlockDriverState *bs, 293 BlockIOLimit *io_limits); 294 295 #ifdef _WIN32 296 int is_windows_drive(const char *filename); 297 #endif 298 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv, 299 enum MonitorEvent ev, 300 BlockErrorAction action, bool is_read); 301 302 /** 303 * stream_start: 304 * @bs: Block device to operate on. 305 * @base: Block device that will become the new base, or %NULL to 306 * flatten the whole backing file chain onto @bs. 307 * @base_id: The file name that will be written to @bs as the new 308 * backing file if the job completes. Ignored if @base is %NULL. 309 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 310 * @on_error: The action to take upon error. 311 * @cb: Completion function for the job. 312 * @opaque: Opaque pointer value passed to @cb. 313 * @errp: Error object. 314 * 315 * Start a streaming operation on @bs. Clusters that are unallocated 316 * in @bs, but allocated in any image between @base and @bs (both 317 * exclusive) will be written to @bs. At the end of a successful 318 * streaming job, the backing file of @bs will be changed to 319 * @base_id in the written image and to @base in the live BlockDriverState. 320 */ 321 void stream_start(BlockDriverState *bs, BlockDriverState *base, 322 const char *base_id, int64_t speed, BlockdevOnError on_error, 323 BlockDriverCompletionFunc *cb, 324 void *opaque, Error **errp); 325 326 /** 327 * commit_start: 328 * @bs: Top Block device 329 * @base: Block device that will be written into, and become the new top 330 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 331 * @on_error: The action to take upon error. 332 * @cb: Completion function for the job. 333 * @opaque: Opaque pointer value passed to @cb. 334 * @errp: Error object. 335 * 336 */ 337 void commit_start(BlockDriverState *bs, BlockDriverState *base, 338 BlockDriverState *top, int64_t speed, 339 BlockdevOnError on_error, BlockDriverCompletionFunc *cb, 340 void *opaque, Error **errp); 341 342 /* 343 * mirror_start: 344 * @bs: Block device to operate on. 345 * @target: Block device to write to. 346 * @speed: The maximum speed, in bytes per second, or 0 for unlimited. 347 * @mode: Whether to collapse all images in the chain to the target. 348 * @on_source_error: The action to take upon error reading from the source. 349 * @on_target_error: The action to take upon error writing to the target. 350 * @cb: Completion function for the job. 351 * @opaque: Opaque pointer value passed to @cb. 352 * @errp: Error object. 353 * 354 * Start a mirroring operation on @bs. Clusters that are allocated 355 * in @bs will be written to @bs until the job is cancelled or 356 * manually completed. At the end of a successful mirroring job, 357 * @bs will be switched to read from @target. 358 */ 359 void mirror_start(BlockDriverState *bs, BlockDriverState *target, 360 int64_t speed, MirrorSyncMode mode, 361 BlockdevOnError on_source_error, 362 BlockdevOnError on_target_error, 363 BlockDriverCompletionFunc *cb, 364 void *opaque, Error **errp); 365 366 #endif /* BLOCK_INT_H */ 367