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_IO_H 25 #define BLOCK_IO_H 26 27 #include "block-common.h" 28 29 /* 30 * I/O API functions. These functions are thread-safe, and therefore 31 * can run in any thread as long as the thread has called 32 * aio_context_acquire/release(). 33 * 34 * These functions can only call functions from I/O and Common categories, 35 * but can be invoked by GS, "I/O or GS" and I/O APIs. 36 * 37 * All functions in this category must use the macro 38 * IO_CODE(); 39 * to catch when they are accidentally called by the wrong API. 40 */ 41 42 int co_wrapper_mixed bdrv_pwrite_zeroes(BdrvChild *child, int64_t offset, 43 int64_t bytes, 44 BdrvRequestFlags flags); 45 int bdrv_make_zero(BdrvChild *child, BdrvRequestFlags flags); 46 int co_wrapper_mixed bdrv_pread(BdrvChild *child, int64_t offset, 47 int64_t bytes, void *buf, 48 BdrvRequestFlags flags); 49 int co_wrapper_mixed bdrv_pwrite(BdrvChild *child, int64_t offset, 50 int64_t bytes, const void *buf, 51 BdrvRequestFlags flags); 52 int co_wrapper_mixed bdrv_pwrite_sync(BdrvChild *child, int64_t offset, 53 int64_t bytes, const void *buf, 54 BdrvRequestFlags flags); 55 int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, 56 int64_t bytes, const void *buf, 57 BdrvRequestFlags flags); 58 /* 59 * Efficiently zero a region of the disk image. Note that this is a regular 60 * I/O request like read or write and should have a reasonable size. This 61 * function is not suitable for zeroing the entire image in a single request 62 * because it may allocate memory for the entire region. 63 */ 64 int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, 65 int64_t bytes, BdrvRequestFlags flags); 66 67 int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, 68 PreallocMode prealloc, BdrvRequestFlags flags, 69 Error **errp); 70 71 int64_t bdrv_nb_sectors(BlockDriverState *bs); 72 int64_t bdrv_getlength(BlockDriverState *bs); 73 int64_t bdrv_get_allocated_file_size(BlockDriverState *bs); 74 BlockMeasureInfo *bdrv_measure(BlockDriver *drv, QemuOpts *opts, 75 BlockDriverState *in_bs, Error **errp); 76 void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr); 77 int coroutine_fn bdrv_co_delete_file(BlockDriverState *bs, Error **errp); 78 void coroutine_fn bdrv_co_delete_file_noerr(BlockDriverState *bs); 79 80 81 /* async block I/O */ 82 void bdrv_aio_cancel(BlockAIOCB *acb); 83 void bdrv_aio_cancel_async(BlockAIOCB *acb); 84 85 /* sg packet commands */ 86 int coroutine_fn bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf); 87 88 /* Ensure contents are flushed to disk. */ 89 int coroutine_fn bdrv_co_flush(BlockDriverState *bs); 90 91 int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, 92 int64_t bytes); 93 bool bdrv_can_write_zeroes_with_unmap(BlockDriverState *bs); 94 int bdrv_block_status(BlockDriverState *bs, int64_t offset, 95 int64_t bytes, int64_t *pnum, int64_t *map, 96 BlockDriverState **file); 97 98 int coroutine_fn bdrv_co_block_status_above(BlockDriverState *bs, 99 BlockDriverState *base, 100 int64_t offset, int64_t bytes, 101 int64_t *pnum, int64_t *map, 102 BlockDriverState **file); 103 int bdrv_block_status_above(BlockDriverState *bs, BlockDriverState *base, 104 int64_t offset, int64_t bytes, int64_t *pnum, 105 int64_t *map, BlockDriverState **file); 106 107 int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t offset, 108 int64_t bytes, int64_t *pnum); 109 int bdrv_is_allocated(BlockDriverState *bs, int64_t offset, int64_t bytes, 110 int64_t *pnum); 111 112 int coroutine_fn bdrv_co_is_allocated_above(BlockDriverState *top, 113 BlockDriverState *base, 114 bool include_base, int64_t offset, 115 int64_t bytes, int64_t *pnum); 116 int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base, 117 bool include_base, int64_t offset, int64_t bytes, 118 int64_t *pnum); 119 120 int coroutine_fn bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, 121 int64_t bytes); 122 123 int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only, 124 bool ignore_allow_rdw, Error **errp); 125 int bdrv_apply_auto_read_only(BlockDriverState *bs, const char *errmsg, 126 Error **errp); 127 bool bdrv_is_read_only(BlockDriverState *bs); 128 bool bdrv_is_writable(BlockDriverState *bs); 129 bool bdrv_is_sg(BlockDriverState *bs); 130 int bdrv_get_flags(BlockDriverState *bs); 131 bool bdrv_is_inserted(BlockDriverState *bs); 132 void bdrv_lock_medium(BlockDriverState *bs, bool locked); 133 void bdrv_eject(BlockDriverState *bs, bool eject_flag); 134 const char *bdrv_get_format_name(BlockDriverState *bs); 135 136 bool bdrv_supports_compressed_writes(BlockDriverState *bs); 137 const char *bdrv_get_node_name(const BlockDriverState *bs); 138 const char *bdrv_get_device_name(const BlockDriverState *bs); 139 const char *bdrv_get_device_or_node_name(const BlockDriverState *bs); 140 int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi); 141 ImageInfoSpecific *bdrv_get_specific_info(BlockDriverState *bs, 142 Error **errp); 143 BlockStatsSpecific *bdrv_get_specific_stats(BlockDriverState *bs); 144 void bdrv_round_to_clusters(BlockDriverState *bs, 145 int64_t offset, int64_t bytes, 146 int64_t *cluster_offset, 147 int64_t *cluster_bytes); 148 149 void bdrv_get_backing_filename(BlockDriverState *bs, 150 char *filename, int filename_size); 151 152 int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf, 153 int64_t pos, int size); 154 155 int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf, 156 int64_t pos, int size); 157 158 /* 159 * Returns the alignment in bytes that is required so that no bounce buffer 160 * is required throughout the stack 161 */ 162 size_t bdrv_min_mem_align(BlockDriverState *bs); 163 /* Returns optimal alignment in bytes for bounce buffer */ 164 size_t bdrv_opt_mem_align(BlockDriverState *bs); 165 void *qemu_blockalign(BlockDriverState *bs, size_t size); 166 void *qemu_blockalign0(BlockDriverState *bs, size_t size); 167 void *qemu_try_blockalign(BlockDriverState *bs, size_t size); 168 void *qemu_try_blockalign0(BlockDriverState *bs, size_t size); 169 170 void bdrv_enable_copy_on_read(BlockDriverState *bs); 171 void bdrv_disable_copy_on_read(BlockDriverState *bs); 172 173 void bdrv_debug_event(BlockDriverState *bs, BlkdebugEvent event); 174 175 #define BLKDBG_EVENT(child, evt) \ 176 do { \ 177 if (child) { \ 178 bdrv_debug_event(child->bs, evt); \ 179 } \ 180 } while (0) 181 182 /** 183 * bdrv_get_aio_context: 184 * 185 * Returns: the currently bound #AioContext 186 */ 187 AioContext *bdrv_get_aio_context(BlockDriverState *bs); 188 189 AioContext *bdrv_child_get_parent_aio_context(BdrvChild *c); 190 191 /** 192 * Move the current coroutine to the AioContext of @bs and return the old 193 * AioContext of the coroutine. Increase bs->in_flight so that draining @bs 194 * will wait for the operation to proceed until the corresponding 195 * bdrv_co_leave(). 196 * 197 * Consequently, you can't call drain inside a bdrv_co_enter/leave() section as 198 * this will deadlock. 199 */ 200 AioContext *coroutine_fn bdrv_co_enter(BlockDriverState *bs); 201 202 /** 203 * Ends a section started by bdrv_co_enter(). Move the current coroutine back 204 * to old_ctx and decrease bs->in_flight again. 205 */ 206 void coroutine_fn bdrv_co_leave(BlockDriverState *bs, AioContext *old_ctx); 207 208 /** 209 * Transfer control to @co in the aio context of @bs 210 */ 211 void bdrv_coroutine_enter(BlockDriverState *bs, Coroutine *co); 212 213 AioContext *child_of_bds_get_parent_aio_context(BdrvChild *c); 214 215 void bdrv_io_plug(BlockDriverState *bs); 216 void bdrv_io_unplug(BlockDriverState *bs); 217 218 bool coroutine_fn bdrv_co_can_store_new_dirty_bitmap(BlockDriverState *bs, 219 const char *name, 220 uint32_t granularity, 221 Error **errp); 222 bool co_wrapper bdrv_can_store_new_dirty_bitmap(BlockDriverState *bs, 223 const char *name, 224 uint32_t granularity, 225 Error **errp); 226 227 /** 228 * 229 * bdrv_co_copy_range: 230 * 231 * Do offloaded copy between two children. If the operation is not implemented 232 * by the driver, or if the backend storage doesn't support it, a negative 233 * error code will be returned. 234 * 235 * Note: block layer doesn't emulate or fallback to a bounce buffer approach 236 * because usually the caller shouldn't attempt offloaded copy any more (e.g. 237 * calling copy_file_range(2)) after the first error, thus it should fall back 238 * to a read+write path in the caller level. 239 * 240 * @src: Source child to copy data from 241 * @src_offset: offset in @src image to read data 242 * @dst: Destination child to copy data to 243 * @dst_offset: offset in @dst image to write data 244 * @bytes: number of bytes to copy 245 * @flags: request flags. Supported flags: 246 * BDRV_REQ_ZERO_WRITE - treat the @src range as zero data and do zero 247 * write on @dst as if bdrv_co_pwrite_zeroes is 248 * called. Used to simplify caller code, or 249 * during BlockDriver.bdrv_co_copy_range_from() 250 * recursion. 251 * BDRV_REQ_NO_SERIALISING - do not serialize with other overlapping 252 * requests currently in flight. 253 * 254 * Returns: 0 if succeeded; negative error code if failed. 255 **/ 256 int coroutine_fn bdrv_co_copy_range(BdrvChild *src, int64_t src_offset, 257 BdrvChild *dst, int64_t dst_offset, 258 int64_t bytes, BdrvRequestFlags read_flags, 259 BdrvRequestFlags write_flags); 260 261 /* 262 * "I/O or GS" API functions. These functions can run without 263 * the BQL, but only in one specific iothread/main loop. 264 * 265 * More specifically, these functions use BDRV_POLL_WHILE(bs), which 266 * requires the caller to be either in the main thread and hold 267 * the BlockdriverState (bs) AioContext lock, or directly in the 268 * home thread that runs the bs AioContext. Calling them from 269 * another thread in another AioContext would cause deadlocks. 270 * 271 * Therefore, these functions are not proper I/O, because they 272 * can't run in *any* iothreads, but only in a specific one. 273 * 274 * These functions can call any function from I/O, Common and this 275 * categories, but must be invoked only by other "I/O or GS" and GS APIs. 276 * 277 * All functions in this category must use the macro 278 * IO_OR_GS_CODE(); 279 * to catch when they are accidentally called by the wrong API. 280 */ 281 282 #define BDRV_POLL_WHILE(bs, cond) ({ \ 283 BlockDriverState *bs_ = (bs); \ 284 IO_OR_GS_CODE(); \ 285 AIO_WAIT_WHILE(bdrv_get_aio_context(bs_), \ 286 cond); }) 287 288 void bdrv_drain(BlockDriverState *bs); 289 290 int co_wrapper_mixed 291 bdrv_truncate(BdrvChild *child, int64_t offset, bool exact, 292 PreallocMode prealloc, BdrvRequestFlags flags, Error **errp); 293 294 int co_wrapper_mixed bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, 295 BdrvCheckMode fix); 296 297 /* Invalidate any cached metadata used by image formats */ 298 int co_wrapper_mixed bdrv_invalidate_cache(BlockDriverState *bs, 299 Error **errp); 300 int co_wrapper_mixed bdrv_flush(BlockDriverState *bs); 301 int co_wrapper_mixed bdrv_pdiscard(BdrvChild *child, int64_t offset, 302 int64_t bytes); 303 int co_wrapper_mixed 304 bdrv_readv_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos); 305 int co_wrapper_mixed 306 bdrv_writev_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos); 307 308 /** 309 * bdrv_parent_drained_begin_single: 310 * 311 * Begin a quiesced section for the parent of @c. 312 */ 313 void bdrv_parent_drained_begin_single(BdrvChild *c); 314 315 /** 316 * bdrv_parent_drained_poll_single: 317 * 318 * Returns true if there is any pending activity to cease before @c can be 319 * called quiesced, false otherwise. 320 */ 321 bool bdrv_parent_drained_poll_single(BdrvChild *c); 322 323 /** 324 * bdrv_parent_drained_end_single: 325 * 326 * End a quiesced section for the parent of @c. 327 */ 328 void bdrv_parent_drained_end_single(BdrvChild *c); 329 330 /** 331 * bdrv_drain_poll: 332 * 333 * Poll for pending requests in @bs and its parents (except for @ignore_parent). 334 * 335 * If @ignore_bds_parents is true, parents that are BlockDriverStates must 336 * ignore the drain request because they will be drained separately (used for 337 * drain_all). 338 * 339 * This is part of bdrv_drained_begin. 340 */ 341 bool bdrv_drain_poll(BlockDriverState *bs, BdrvChild *ignore_parent, 342 bool ignore_bds_parents); 343 344 /** 345 * bdrv_drained_begin: 346 * 347 * Begin a quiesced section for exclusive access to the BDS, by disabling 348 * external request sources including NBD server, block jobs, and device model. 349 * 350 * This function can be recursive. 351 */ 352 void bdrv_drained_begin(BlockDriverState *bs); 353 354 /** 355 * bdrv_do_drained_begin_quiesce: 356 * 357 * Quiesces a BDS like bdrv_drained_begin(), but does not wait for already 358 * running requests to complete. 359 */ 360 void bdrv_do_drained_begin_quiesce(BlockDriverState *bs, BdrvChild *parent); 361 362 /** 363 * bdrv_drained_end: 364 * 365 * End a quiescent section started by bdrv_drained_begin(). 366 */ 367 void bdrv_drained_end(BlockDriverState *bs); 368 369 #endif /* BLOCK_IO_H */ 370