| 1 | /* BlockDriver implementation for "raw" format driver |
| 2 | * |
| 3 | * Copyright (C) 2010-2016 Red Hat, Inc. |
| 4 | * Copyright (C) 2010, Blue Swirl <blauwirbel@gmail.com> |
| 5 | * Copyright (C) 2009, Anthony Liguori <aliguori@us.ibm.com> |
| 6 | * |
| 7 | * Author: |
| 8 | * Laszlo Ersek <lersek@redhat.com> |
| 9 | * |
| 10 | * Permission is hereby granted, free of charge, to any person obtaining a copy |
| 11 | * of this software and associated documentation files (the "Software"), to |
| 12 | * deal in the Software without restriction, including without limitation the |
| 13 | * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or |
| 14 | * sell copies of the Software, and to permit persons to whom the Software is |
| 15 | * furnished to do so, subject to the following conditions: |
| 16 | * |
| 17 | * The above copyright notice and this permission notice shall be included in |
| 18 | * all copies or substantial portions of the Software. |
| 19 | * |
| 20 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 21 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 22 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 23 | * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 24 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING |
| 25 | * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS |
| 26 | * IN THE SOFTWARE. |
| 27 | */ |
| 28 | |
| 29 | #include "qemu/osdep.h" |
| 30 | #include "block/block-io.h" |
| 31 | #include "block/block_int.h" |
| 32 | #include "qapi/error.h" |
| 33 | #include "qemu/module.h" |
| 34 | #include "qemu/option.h" |
| 35 | #include "qemu/memalign.h" |
| 36 | |
| 37 | typedef struct BDRVRawState { |
| 38 | uint64_t offset; |
| 39 | uint64_t size; |
| 40 | bool has_size; |
| 41 | } BDRVRawState; |
| 42 | |
| 43 | static const char *const mutable_opts[] = { "offset", "size", NULL }; |
| 44 | |
| 45 | static QemuOptsList raw_runtime_opts = { |
| 46 | .name = "raw", |
| 47 | .head = QTAILQ_HEAD_INITIALIZER(raw_runtime_opts.head), |
| 48 | .desc = { |
| 49 | { |
| 50 | .name = "offset", |
| 51 | .type = QEMU_OPT_SIZE, |
| 52 | .help = "offset in the disk where the image starts", |
| 53 | }, |
| 54 | { |
| 55 | .name = "size", |
| 56 | .type = QEMU_OPT_SIZE, |
| 57 | .help = "virtual disk size", |
| 58 | }, |
| 59 | { /* end of list */ } |
| 60 | }, |
| 61 | }; |
| 62 | |
| 63 | static QemuOptsList raw_create_opts = { |
| 64 | .name = "raw-create-opts", |
| 65 | .head = QTAILQ_HEAD_INITIALIZER(raw_create_opts.head), |
| 66 | .desc = { |
| 67 | { |
| 68 | .name = BLOCK_OPT_SIZE, |
| 69 | .type = QEMU_OPT_SIZE, |
| 70 | .help = "Virtual disk size" |
| 71 | }, |
| 72 | { /* end of list */ } |
| 73 | } |
| 74 | }; |
| 75 | |
| 76 | static int raw_read_options(QDict *options, uint64_t *offset, bool *has_size, |
| 77 | uint64_t *size, Error **errp) |
| 78 | { |
| 79 | QemuOpts *opts = NULL; |
| 80 | int ret; |
| 81 | |
| 82 | opts = qemu_opts_create(&raw_runtime_opts, NULL, 0, &error_abort); |
| 83 | if (!qemu_opts_absorb_qdict(opts, options, errp)) { |
| 84 | ret = -EINVAL; |
| 85 | goto end; |
| 86 | } |
| 87 | |
| 88 | *offset = qemu_opt_get_size(opts, "offset", 0); |
| 89 | *has_size = qemu_opt_find(opts, "size"); |
| 90 | *size = qemu_opt_get_size(opts, "size", 0); |
| 91 | |
| 92 | ret = 0; |
| 93 | end: |
| 94 | qemu_opts_del(opts); |
| 95 | return ret; |
| 96 | } |
| 97 | |
| 98 | static int GRAPH_RDLOCK |
| 99 | raw_apply_options(BlockDriverState *bs, BDRVRawState *s, uint64_t offset, |
| 100 | bool has_size, uint64_t size, Error **errp) |
| 101 | { |
| 102 | int64_t real_size = 0; |
| 103 | |
| 104 | real_size = bdrv_getlength(bs->file->bs); |
| 105 | if (real_size < 0) { |
| 106 | error_setg_errno(errp, -real_size, "Could not get image size"); |
| 107 | return real_size; |
| 108 | } |
| 109 | |
| 110 | /* Check size and offset */ |
| 111 | if (offset > real_size) { |
| 112 | error_setg(errp, "Offset (%" PRIu64 ") cannot be greater than " |
| 113 | "size of the containing file (%" PRId64 ")", |
| 114 | offset, real_size); |
| 115 | return -EINVAL; |
| 116 | } |
| 117 | |
| 118 | if (has_size && (real_size - offset) < size) { |
| 119 | error_setg(errp, "The sum of offset (%" PRIu64 ") and size " |
| 120 | "(%" PRIu64 ") has to be smaller or equal to the " |
| 121 | " actual size of the containing file (%" PRId64 ")", |
| 122 | offset, size, real_size); |
| 123 | return -EINVAL; |
| 124 | } |
| 125 | |
| 126 | /* Make sure size is multiple of BDRV_SECTOR_SIZE to prevent rounding |
| 127 | * up and leaking out of the specified area. */ |
| 128 | if (has_size && !QEMU_IS_ALIGNED(size, BDRV_SECTOR_SIZE)) { |
| 129 | error_setg(errp, "Specified size is not multiple of %llu", |
| 130 | BDRV_SECTOR_SIZE); |
| 131 | return -EINVAL; |
| 132 | } |
| 133 | |
| 134 | s->offset = offset; |
| 135 | s->has_size = has_size; |
| 136 | s->size = has_size ? size : real_size - offset; |
| 137 | |
| 138 | return 0; |
| 139 | } |
| 140 | |
| 141 | static int raw_reopen_prepare(BDRVReopenState *reopen_state, |
| 142 | BlockReopenQueue *queue, Error **errp) |
| 143 | { |
| 144 | bool has_size; |
| 145 | uint64_t offset, size; |
| 146 | int ret; |
| 147 | |
| 148 | GLOBAL_STATE_CODE(); |
| 149 | GRAPH_RDLOCK_GUARD_MAINLOOP(); |
| 150 | |
| 151 | assert(reopen_state != NULL); |
| 152 | assert(reopen_state->bs != NULL); |
| 153 | |
| 154 | reopen_state->opaque = g_new0(BDRVRawState, 1); |
| 155 | |
| 156 | ret = raw_read_options(reopen_state->options, &offset, &has_size, &size, |
| 157 | errp); |
| 158 | if (ret < 0) { |
| 159 | return ret; |
| 160 | } |
| 161 | |
| 162 | ret = raw_apply_options(reopen_state->bs, reopen_state->opaque, |
| 163 | offset, has_size, size, errp); |
| 164 | if (ret < 0) { |
| 165 | return ret; |
| 166 | } |
| 167 | |
| 168 | return 0; |
| 169 | } |
| 170 | |
| 171 | static void raw_reopen_commit(BDRVReopenState *state) |
| 172 | { |
| 173 | BDRVRawState *new_s = state->opaque; |
| 174 | BDRVRawState *s = state->bs->opaque; |
| 175 | |
| 176 | memcpy(s, new_s, sizeof(BDRVRawState)); |
| 177 | |
| 178 | g_free(state->opaque); |
| 179 | state->opaque = NULL; |
| 180 | } |
| 181 | |
| 182 | static void raw_reopen_abort(BDRVReopenState *state) |
| 183 | { |
| 184 | g_free(state->opaque); |
| 185 | state->opaque = NULL; |
| 186 | } |
| 187 | |
| 188 | /* Check and adjust the offset, against 'offset' and 'size' options. */ |
| 189 | static inline int raw_adjust_offset(BlockDriverState *bs, int64_t *offset, |
| 190 | int64_t bytes, bool is_write) |
| 191 | { |
| 192 | BDRVRawState *s = bs->opaque; |
| 193 | |
| 194 | if (s->has_size && (*offset > s->size || bytes > (s->size - *offset))) { |
| 195 | /* There's not enough space for the write, or the read request is |
| 196 | * out-of-range. Don't read/write anything to prevent leaking out of |
| 197 | * the size specified in options. */ |
| 198 | return is_write ? -ENOSPC : -EINVAL; |
| 199 | } |
| 200 | |
| 201 | if (*offset > INT64_MAX - s->offset) { |
| 202 | return -EINVAL; |
| 203 | } |
| 204 | *offset += s->offset; |
| 205 | |
| 206 | return 0; |
| 207 | } |
| 208 | |
| 209 | static int coroutine_fn GRAPH_RDLOCK |
| 210 | raw_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, |
| 211 | QEMUIOVector *qiov, BdrvRequestFlags flags) |
| 212 | { |
| 213 | int ret; |
| 214 | |
| 215 | ret = raw_adjust_offset(bs, &offset, bytes, false); |
| 216 | if (ret) { |
| 217 | return ret; |
| 218 | } |
| 219 | |
| 220 | BLKDBG_CO_EVENT(bs->file, BLKDBG_READ_AIO); |
| 221 | return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); |
| 222 | } |
| 223 | |
| 224 | static int coroutine_fn GRAPH_RDLOCK |
| 225 | raw_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, |
| 226 | QEMUIOVector *qiov, BdrvRequestFlags flags) |
| 227 | { |
| 228 | void *buf = NULL; |
| 229 | BlockDriver *drv; |
| 230 | QEMUIOVector local_qiov; |
| 231 | int ret; |
| 232 | |
| 233 | if (bs->probed && offset < BLOCK_PROBE_BUF_SIZE && bytes) { |
| 234 | /* Handling partial writes would be a pain - so we just |
| 235 | * require that guests have 512-byte request alignment if |
| 236 | * probing occurred */ |
| 237 | QEMU_BUILD_BUG_ON(BLOCK_PROBE_BUF_SIZE != 512); |
| 238 | QEMU_BUILD_BUG_ON(BDRV_SECTOR_SIZE != 512); |
| 239 | assert(offset == 0 && bytes >= BLOCK_PROBE_BUF_SIZE); |
| 240 | |
| 241 | buf = qemu_try_blockalign(bs->file->bs, 512); |
| 242 | if (!buf) { |
| 243 | ret = -ENOMEM; |
| 244 | goto fail; |
| 245 | } |
| 246 | |
| 247 | ret = qemu_iovec_to_buf(qiov, 0, buf, 512); |
| 248 | if (ret != 512) { |
| 249 | ret = -EINVAL; |
| 250 | goto fail; |
| 251 | } |
| 252 | |
| 253 | drv = bdrv_probe_all(buf, 512, NULL); |
| 254 | if (drv != bs->drv) { |
| 255 | ret = -EPERM; |
| 256 | goto fail; |
| 257 | } |
| 258 | |
| 259 | /* Use the checked buffer, a malicious guest might be overwriting its |
| 260 | * original buffer in the background. */ |
| 261 | qemu_iovec_init(&local_qiov, qiov->niov + 1); |
| 262 | qemu_iovec_add(&local_qiov, buf, 512); |
| 263 | qemu_iovec_concat(&local_qiov, qiov, 512, qiov->size - 512); |
| 264 | qiov = &local_qiov; |
| 265 | |
| 266 | flags &= ~BDRV_REQ_REGISTERED_BUF; |
| 267 | } |
| 268 | |
| 269 | ret = raw_adjust_offset(bs, &offset, bytes, true); |
| 270 | if (ret) { |
| 271 | goto fail; |
| 272 | } |
| 273 | |
| 274 | BLKDBG_CO_EVENT(bs->file, BLKDBG_WRITE_AIO); |
| 275 | ret = bdrv_co_pwritev(bs->file, offset, bytes, qiov, flags); |
| 276 | |
| 277 | fail: |
| 278 | if (qiov == &local_qiov) { |
| 279 | qemu_iovec_destroy(&local_qiov); |
| 280 | } |
| 281 | qemu_vfree(buf); |
| 282 | return ret; |
| 283 | } |
| 284 | |
| 285 | static int coroutine_fn GRAPH_RDLOCK |
| 286 | raw_co_block_status(BlockDriverState *bs, unsigned int mode, |
| 287 | int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, |
| 288 | BlockDriverState **file) |
| 289 | { |
| 290 | BDRVRawState *s = bs->opaque; |
| 291 | *pnum = bytes; |
| 292 | *file = bs->file->bs; |
| 293 | *map = offset + s->offset; |
| 294 | return BDRV_BLOCK_RAW | BDRV_BLOCK_OFFSET_VALID; |
| 295 | } |
| 296 | |
| 297 | static int coroutine_fn GRAPH_RDLOCK |
| 298 | raw_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, |
| 299 | BdrvRequestFlags flags) |
| 300 | { |
| 301 | int ret; |
| 302 | |
| 303 | ret = raw_adjust_offset(bs, &offset, bytes, true); |
| 304 | if (ret) { |
| 305 | return ret; |
| 306 | } |
| 307 | return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); |
| 308 | } |
| 309 | |
| 310 | static int coroutine_fn GRAPH_RDLOCK |
| 311 | raw_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) |
| 312 | { |
| 313 | int ret; |
| 314 | |
| 315 | ret = raw_adjust_offset(bs, &offset, bytes, true); |
| 316 | if (ret) { |
| 317 | return ret; |
| 318 | } |
| 319 | return bdrv_co_pdiscard(bs->file, offset, bytes); |
| 320 | } |
| 321 | |
| 322 | static int coroutine_fn GRAPH_RDLOCK |
| 323 | raw_co_zone_report(BlockDriverState *bs, int64_t offset, |
| 324 | unsigned int *nr_zones, |
| 325 | BlockZoneDescriptor *zones) |
| 326 | { |
| 327 | return bdrv_co_zone_report(bs->file->bs, offset, nr_zones, zones); |
| 328 | } |
| 329 | |
| 330 | static int coroutine_fn GRAPH_RDLOCK |
| 331 | raw_co_zone_mgmt(BlockDriverState *bs, BlockZoneOp op, |
| 332 | int64_t offset, int64_t len) |
| 333 | { |
| 334 | return bdrv_co_zone_mgmt(bs->file->bs, op, offset, len); |
| 335 | } |
| 336 | |
| 337 | static int coroutine_fn GRAPH_RDLOCK |
| 338 | raw_co_zone_append(BlockDriverState *bs,int64_t *offset, QEMUIOVector *qiov, |
| 339 | BdrvRequestFlags flags) |
| 340 | { |
| 341 | return bdrv_co_zone_append(bs->file->bs, offset, qiov, flags); |
| 342 | } |
| 343 | |
| 344 | static int64_t coroutine_fn GRAPH_RDLOCK |
| 345 | raw_co_getlength(BlockDriverState *bs) |
| 346 | { |
| 347 | int64_t len; |
| 348 | BDRVRawState *s = bs->opaque; |
| 349 | |
| 350 | /* Update size. It should not change unless the file was externally |
| 351 | * modified. */ |
| 352 | len = bdrv_co_getlength(bs->file->bs); |
| 353 | if (len < 0) { |
| 354 | return len; |
| 355 | } |
| 356 | |
| 357 | if (len < s->offset) { |
| 358 | s->size = 0; |
| 359 | } else { |
| 360 | if (s->has_size) { |
| 361 | /* Try to honour the size */ |
| 362 | s->size = MIN(s->size, len - s->offset); |
| 363 | } else { |
| 364 | s->size = len - s->offset; |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | return s->size; |
| 369 | } |
| 370 | |
| 371 | static BlockMeasureInfo *raw_measure(QemuOpts *opts, BlockDriverState *in_bs, |
| 372 | Error **errp) |
| 373 | { |
| 374 | BlockMeasureInfo *info; |
| 375 | int64_t required; |
| 376 | |
| 377 | if (in_bs) { |
| 378 | required = bdrv_getlength(in_bs); |
| 379 | if (required < 0) { |
| 380 | error_setg_errno(errp, -required, "Unable to get image size"); |
| 381 | return NULL; |
| 382 | } |
| 383 | } else { |
| 384 | required = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0), |
| 385 | BDRV_SECTOR_SIZE); |
| 386 | } |
| 387 | |
| 388 | info = g_new0(BlockMeasureInfo, 1); |
| 389 | info->required = required; |
| 390 | |
| 391 | /* Unallocated sectors count towards the file size in raw images */ |
| 392 | info->fully_allocated = info->required; |
| 393 | return info; |
| 394 | } |
| 395 | |
| 396 | static int coroutine_fn GRAPH_RDLOCK |
| 397 | raw_co_get_info(BlockDriverState *bs, BlockDriverInfo *bdi) |
| 398 | { |
| 399 | return bdrv_co_get_info(bs->file->bs, bdi); |
| 400 | } |
| 401 | |
| 402 | static void GRAPH_RDLOCK raw_refresh_limits(BlockDriverState *bs, Error **errp) |
| 403 | { |
| 404 | bs->bl.has_variable_length = bs->file->bs->bl.has_variable_length; |
| 405 | |
| 406 | if (bs->probed) { |
| 407 | /* To make it easier to protect the first sector, any probed |
| 408 | * image is restricted to read-modify-write on sub-sector |
| 409 | * operations. */ |
| 410 | bs->bl.request_alignment = BDRV_SECTOR_SIZE; |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | static int coroutine_fn GRAPH_RDLOCK |
| 415 | raw_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, |
| 416 | PreallocMode prealloc, BdrvRequestFlags flags, Error **errp) |
| 417 | { |
| 418 | BDRVRawState *s = bs->opaque; |
| 419 | |
| 420 | if (s->has_size) { |
| 421 | error_setg(errp, "Cannot resize fixed-size raw disks"); |
| 422 | return -ENOTSUP; |
| 423 | } |
| 424 | |
| 425 | if (INT64_MAX - offset < s->offset) { |
| 426 | error_setg(errp, "Disk size too large for the chosen offset"); |
| 427 | return -EINVAL; |
| 428 | } |
| 429 | |
| 430 | s->size = offset; |
| 431 | offset += s->offset; |
| 432 | return bdrv_co_truncate(bs->file, offset, exact, prealloc, flags, errp); |
| 433 | } |
| 434 | |
| 435 | static void coroutine_fn GRAPH_RDLOCK |
| 436 | raw_co_eject(BlockDriverState *bs, bool eject_flag) |
| 437 | { |
| 438 | bdrv_co_eject(bs->file->bs, eject_flag); |
| 439 | } |
| 440 | |
| 441 | static void coroutine_fn GRAPH_RDLOCK |
| 442 | raw_co_lock_medium(BlockDriverState *bs, bool locked) |
| 443 | { |
| 444 | bdrv_co_lock_medium(bs->file->bs, locked); |
| 445 | } |
| 446 | |
| 447 | static int coroutine_fn GRAPH_RDLOCK |
| 448 | raw_co_ioctl(BlockDriverState *bs, unsigned long int req, void *buf) |
| 449 | { |
| 450 | BDRVRawState *s = bs->opaque; |
| 451 | if (s->offset || s->has_size) { |
| 452 | return -ENOTSUP; |
| 453 | } |
| 454 | return bdrv_co_ioctl(bs->file->bs, req, buf); |
| 455 | } |
| 456 | |
| 457 | static int GRAPH_RDLOCK raw_has_zero_init(BlockDriverState *bs) |
| 458 | { |
| 459 | return bdrv_has_zero_init(bs->file->bs); |
| 460 | } |
| 461 | |
| 462 | static int coroutine_fn GRAPH_UNLOCKED |
| 463 | raw_co_create_opts(BlockDriver *drv, const char *filename, |
| 464 | QemuOpts *opts, Error **errp) |
| 465 | { |
| 466 | return bdrv_co_create_file(filename, opts, true, errp); |
| 467 | } |
| 468 | |
| 469 | static int raw_open(BlockDriverState *bs, QDict *options, int flags, |
| 470 | Error **errp) |
| 471 | { |
| 472 | BDRVRawState *s = bs->opaque; |
| 473 | bool has_size; |
| 474 | uint64_t offset, size; |
| 475 | BdrvChildRole file_role; |
| 476 | int ret; |
| 477 | |
| 478 | GLOBAL_STATE_CODE(); |
| 479 | |
| 480 | ret = raw_read_options(options, &offset, &has_size, &size, errp); |
| 481 | if (ret < 0) { |
| 482 | return ret; |
| 483 | } |
| 484 | |
| 485 | /* |
| 486 | * Without offset and a size limit, this driver behaves very much |
| 487 | * like a filter. With any such limit, it does not. |
| 488 | */ |
| 489 | if (offset || has_size) { |
| 490 | file_role = BDRV_CHILD_DATA | BDRV_CHILD_PRIMARY; |
| 491 | } else { |
| 492 | file_role = BDRV_CHILD_FILTERED | BDRV_CHILD_PRIMARY; |
| 493 | } |
| 494 | |
| 495 | bdrv_open_child(NULL, options, "file", bs, &child_of_bds, |
| 496 | file_role, false, errp); |
| 497 | |
| 498 | GRAPH_RDLOCK_GUARD_MAINLOOP(); |
| 499 | if (!bs->file) { |
| 500 | return -EINVAL; |
| 501 | } |
| 502 | |
| 503 | bs->sg = bdrv_is_sg(bs->file->bs); |
| 504 | bs->supported_write_flags = BDRV_REQ_WRITE_UNCHANGED | |
| 505 | (BDRV_REQ_FUA & bs->file->bs->supported_write_flags); |
| 506 | bs->supported_zero_flags = BDRV_REQ_WRITE_UNCHANGED | |
| 507 | ((BDRV_REQ_FUA | BDRV_REQ_MAY_UNMAP | BDRV_REQ_NO_FALLBACK) & |
| 508 | bs->file->bs->supported_zero_flags); |
| 509 | bs->supported_truncate_flags = bs->file->bs->supported_truncate_flags & |
| 510 | BDRV_REQ_ZERO_WRITE; |
| 511 | |
| 512 | if (bs->probed && !bdrv_is_read_only(bs)) { |
| 513 | bdrv_refresh_filename(bs->file->bs); |
| 514 | fprintf(stderr, |
| 515 | "WARNING: Image format was not specified for '%s' and probing " |
| 516 | "guessed raw.\n" |
| 517 | " Automatically detecting the format is dangerous for " |
| 518 | "raw images, write operations on block 0 will be restricted.\n" |
| 519 | " Specify the 'raw' format explicitly to remove the " |
| 520 | "restrictions.\n", |
| 521 | bs->file->bs->filename); |
| 522 | } |
| 523 | |
| 524 | ret = raw_apply_options(bs, s, offset, has_size, size, errp); |
| 525 | if (ret < 0) { |
| 526 | return ret; |
| 527 | } |
| 528 | |
| 529 | if (bdrv_is_sg(bs) && (s->offset || s->has_size)) { |
| 530 | error_setg(errp, "Cannot use offset/size with SCSI generic devices"); |
| 531 | return -EINVAL; |
| 532 | } |
| 533 | |
| 534 | return 0; |
| 535 | } |
| 536 | |
| 537 | static int raw_probe(const uint8_t *buf, int buf_size, const char *filename) |
| 538 | { |
| 539 | /* smallest possible positive score so that raw is used if and only if no |
| 540 | * other block driver works |
| 541 | */ |
| 542 | return 1; |
| 543 | } |
| 544 | |
| 545 | static int GRAPH_RDLOCK |
| 546 | raw_probe_blocksizes(BlockDriverState *bs, BlockSizes *bsz) |
| 547 | { |
| 548 | BDRVRawState *s = bs->opaque; |
| 549 | int ret; |
| 550 | |
| 551 | ret = bdrv_probe_blocksizes(bs->file->bs, bsz); |
| 552 | if (ret < 0) { |
| 553 | return ret; |
| 554 | } |
| 555 | |
| 556 | if (!QEMU_IS_ALIGNED(s->offset, MAX(bsz->log, bsz->phys))) { |
| 557 | return -ENOTSUP; |
| 558 | } |
| 559 | |
| 560 | return 0; |
| 561 | } |
| 562 | |
| 563 | static int GRAPH_RDLOCK |
| 564 | raw_probe_geometry(BlockDriverState *bs, HDGeometry *geo) |
| 565 | { |
| 566 | BDRVRawState *s = bs->opaque; |
| 567 | if (s->offset || s->has_size) { |
| 568 | return -ENOTSUP; |
| 569 | } |
| 570 | return bdrv_probe_geometry(bs->file->bs, geo); |
| 571 | } |
| 572 | |
| 573 | static int coroutine_fn GRAPH_RDLOCK |
| 574 | raw_co_copy_range_from(BlockDriverState *bs, |
| 575 | BdrvChild *src, int64_t src_offset, |
| 576 | BdrvChild *dst, int64_t dst_offset, |
| 577 | int64_t bytes, BdrvRequestFlags read_flags, |
| 578 | BdrvRequestFlags write_flags) |
| 579 | { |
| 580 | int ret; |
| 581 | |
| 582 | ret = raw_adjust_offset(bs, &src_offset, bytes, false); |
| 583 | if (ret) { |
| 584 | return ret; |
| 585 | } |
| 586 | return bdrv_co_copy_range_from(bs->file, src_offset, dst, dst_offset, |
| 587 | bytes, read_flags, write_flags); |
| 588 | } |
| 589 | |
| 590 | static int coroutine_fn GRAPH_RDLOCK |
| 591 | raw_co_copy_range_to(BlockDriverState *bs, |
| 592 | BdrvChild *src, int64_t src_offset, |
| 593 | BdrvChild *dst, int64_t dst_offset, |
| 594 | int64_t bytes, BdrvRequestFlags read_flags, |
| 595 | BdrvRequestFlags write_flags) |
| 596 | { |
| 597 | int ret; |
| 598 | |
| 599 | ret = raw_adjust_offset(bs, &dst_offset, bytes, true); |
| 600 | if (ret) { |
| 601 | return ret; |
| 602 | } |
| 603 | return bdrv_co_copy_range_to(src, src_offset, bs->file, dst_offset, bytes, |
| 604 | read_flags, write_flags); |
| 605 | } |
| 606 | |
| 607 | static const char *const raw_strong_runtime_opts[] = { |
| 608 | "offset", |
| 609 | "size", |
| 610 | |
| 611 | NULL |
| 612 | }; |
| 613 | |
| 614 | static void GRAPH_RDLOCK raw_cancel_in_flight(BlockDriverState *bs) |
| 615 | { |
| 616 | bdrv_cancel_in_flight(bs->file->bs); |
| 617 | } |
| 618 | |
| 619 | static void raw_child_perm(BlockDriverState *bs, BdrvChild *c, |
| 620 | BdrvChildRole role, |
| 621 | BlockReopenQueue *reopen_queue, |
| 622 | uint64_t parent_perm, uint64_t parent_shared, |
| 623 | uint64_t *nperm, uint64_t *nshared) |
| 624 | { |
| 625 | bdrv_default_perms(bs, c, role, reopen_queue, parent_perm, |
| 626 | parent_shared, nperm, nshared); |
| 627 | |
| 628 | /* |
| 629 | * bdrv_default_perms() may add WRITE and/or RESIZE (see comment in |
| 630 | * bdrv_default_perms_for_storage() for an explanation) but we only need |
| 631 | * them if they are in parent_perm. Drop WRITE and RESIZE whenever possible |
| 632 | * to avoid permission conflicts. |
| 633 | */ |
| 634 | *nperm &= ~(BLK_PERM_WRITE | BLK_PERM_RESIZE); |
| 635 | *nperm |= parent_perm & (BLK_PERM_WRITE | BLK_PERM_RESIZE); |
| 636 | } |
| 637 | |
| 638 | BlockDriver bdrv_raw = { |
| 639 | .format_name = "raw", |
| 640 | .instance_size = sizeof(BDRVRawState), |
| 641 | .supports_zoned_children = true, |
| 642 | .bdrv_probe = &raw_probe, |
| 643 | .bdrv_reopen_prepare = &raw_reopen_prepare, |
| 644 | .bdrv_reopen_commit = &raw_reopen_commit, |
| 645 | .bdrv_reopen_abort = &raw_reopen_abort, |
| 646 | .bdrv_open = &raw_open, |
| 647 | .bdrv_child_perm = raw_child_perm, |
| 648 | .bdrv_co_create_opts = &raw_co_create_opts, |
| 649 | .bdrv_co_preadv = &raw_co_preadv, |
| 650 | .bdrv_co_pwritev = &raw_co_pwritev, |
| 651 | .bdrv_co_pwrite_zeroes = &raw_co_pwrite_zeroes, |
| 652 | .bdrv_co_pdiscard = &raw_co_pdiscard, |
| 653 | .bdrv_co_zone_report = &raw_co_zone_report, |
| 654 | .bdrv_co_zone_mgmt = &raw_co_zone_mgmt, |
| 655 | .bdrv_co_zone_append = &raw_co_zone_append, |
| 656 | .bdrv_co_block_status = &raw_co_block_status, |
| 657 | .bdrv_co_copy_range_from = &raw_co_copy_range_from, |
| 658 | .bdrv_co_copy_range_to = &raw_co_copy_range_to, |
| 659 | .bdrv_co_truncate = &raw_co_truncate, |
| 660 | .bdrv_co_getlength = &raw_co_getlength, |
| 661 | .is_format = true, |
| 662 | .bdrv_measure = &raw_measure, |
| 663 | .bdrv_co_get_info = &raw_co_get_info, |
| 664 | .bdrv_refresh_limits = &raw_refresh_limits, |
| 665 | .bdrv_probe_blocksizes = &raw_probe_blocksizes, |
| 666 | .bdrv_probe_geometry = &raw_probe_geometry, |
| 667 | .bdrv_co_eject = &raw_co_eject, |
| 668 | .bdrv_co_lock_medium = &raw_co_lock_medium, |
| 669 | .bdrv_co_ioctl = &raw_co_ioctl, |
| 670 | .create_opts = &raw_create_opts, |
| 671 | .bdrv_has_zero_init = &raw_has_zero_init, |
| 672 | .strong_runtime_opts = raw_strong_runtime_opts, |
| 673 | .mutable_opts = mutable_opts, |
| 674 | .bdrv_cancel_in_flight = raw_cancel_in_flight, |
| 675 | }; |
| 676 | |
| 677 | static void bdrv_raw_init(void) |
| 678 | { |
| 679 | bdrv_register(&bdrv_raw); |
| 680 | } |
| 681 | |
| 682 | block_init(bdrv_raw_init); |