1 /* 2 * Image mirroring 3 * 4 * Copyright Red Hat, Inc. 2012 5 * 6 * Authors: 7 * Paolo Bonzini <pbonzini@redhat.com> 8 * 9 * This work is licensed under the terms of the GNU LGPL, version 2 or later. 10 * See the COPYING.LIB file in the top-level directory. 11 * 12 */ 13 14 #include "qemu/osdep.h" 15 #include "qemu/cutils.h" 16 #include "qemu/coroutine.h" 17 #include "qemu/range.h" 18 #include "trace.h" 19 #include "block/blockjob_int.h" 20 #include "block/block_int.h" 21 #include "block/dirty-bitmap.h" 22 #include "sysemu/block-backend.h" 23 #include "qapi/error.h" 24 #include "qemu/ratelimit.h" 25 #include "qemu/bitmap.h" 26 #include "qemu/memalign.h" 27 28 #define MAX_IN_FLIGHT 16 29 #define MAX_IO_BYTES (1 << 20) /* 1 Mb */ 30 #define DEFAULT_MIRROR_BUF_SIZE (MAX_IN_FLIGHT * MAX_IO_BYTES) 31 32 /* The mirroring buffer is a list of granularity-sized chunks. 33 * Free chunks are organized in a list. 34 */ 35 typedef struct MirrorBuffer { 36 QSIMPLEQ_ENTRY(MirrorBuffer) next; 37 } MirrorBuffer; 38 39 typedef struct MirrorOp MirrorOp; 40 41 typedef struct MirrorBlockJob { 42 BlockJob common; 43 BlockBackend *target; 44 BlockDriverState *mirror_top_bs; 45 BlockDriverState *base; 46 BlockDriverState *base_overlay; 47 48 /* The name of the graph node to replace */ 49 char *replaces; 50 /* The BDS to replace */ 51 BlockDriverState *to_replace; 52 /* Used to block operations on the drive-mirror-replace target */ 53 Error *replace_blocker; 54 bool is_none_mode; 55 BlockMirrorBackingMode backing_mode; 56 /* Whether the target image requires explicit zero-initialization */ 57 bool zero_target; 58 /* 59 * To be accesssed with atomics. Written only under the BQL (required by the 60 * current implementation of mirror_change()). 61 */ 62 MirrorCopyMode copy_mode; 63 BlockdevOnError on_source_error, on_target_error; 64 /* 65 * To be accessed with atomics. 66 * 67 * Set when the target is synced (dirty bitmap is clean, nothing in flight) 68 * and the job is running in active mode. 69 */ 70 bool actively_synced; 71 bool should_complete; 72 int64_t granularity; 73 size_t buf_size; 74 int64_t bdev_length; 75 unsigned long *cow_bitmap; 76 BdrvDirtyBitmap *dirty_bitmap; 77 BdrvDirtyBitmapIter *dbi; 78 uint8_t *buf; 79 QSIMPLEQ_HEAD(, MirrorBuffer) buf_free; 80 int buf_free_count; 81 82 uint64_t last_pause_ns; 83 unsigned long *in_flight_bitmap; 84 unsigned in_flight; 85 int64_t bytes_in_flight; 86 QTAILQ_HEAD(, MirrorOp) ops_in_flight; 87 int ret; 88 bool unmap; 89 int target_cluster_size; 90 int max_iov; 91 bool initial_zeroing_ongoing; 92 int in_active_write_counter; 93 int64_t active_write_bytes_in_flight; 94 bool prepared; 95 bool in_drain; 96 } MirrorBlockJob; 97 98 typedef struct MirrorBDSOpaque { 99 MirrorBlockJob *job; 100 bool stop; 101 bool is_commit; 102 } MirrorBDSOpaque; 103 104 struct MirrorOp { 105 MirrorBlockJob *s; 106 QEMUIOVector qiov; 107 int64_t offset; 108 uint64_t bytes; 109 110 /* The pointee is set by mirror_co_read(), mirror_co_zero(), and 111 * mirror_co_discard() before yielding for the first time */ 112 int64_t *bytes_handled; 113 114 bool is_pseudo_op; 115 bool is_active_write; 116 bool is_in_flight; 117 CoQueue waiting_requests; 118 Coroutine *co; 119 MirrorOp *waiting_for_op; 120 121 QTAILQ_ENTRY(MirrorOp) next; 122 }; 123 124 typedef enum MirrorMethod { 125 MIRROR_METHOD_COPY, 126 MIRROR_METHOD_ZERO, 127 MIRROR_METHOD_DISCARD, 128 } MirrorMethod; 129 130 static BlockErrorAction mirror_error_action(MirrorBlockJob *s, bool read, 131 int error) 132 { 133 qatomic_set(&s->actively_synced, false); 134 if (read) { 135 return block_job_error_action(&s->common, s->on_source_error, 136 true, error); 137 } else { 138 return block_job_error_action(&s->common, s->on_target_error, 139 false, error); 140 } 141 } 142 143 static void coroutine_fn mirror_wait_on_conflicts(MirrorOp *self, 144 MirrorBlockJob *s, 145 uint64_t offset, 146 uint64_t bytes) 147 { 148 uint64_t self_start_chunk = offset / s->granularity; 149 uint64_t self_end_chunk = DIV_ROUND_UP(offset + bytes, s->granularity); 150 uint64_t self_nb_chunks = self_end_chunk - self_start_chunk; 151 152 while (find_next_bit(s->in_flight_bitmap, self_end_chunk, 153 self_start_chunk) < self_end_chunk && 154 s->ret >= 0) 155 { 156 MirrorOp *op; 157 158 QTAILQ_FOREACH(op, &s->ops_in_flight, next) { 159 uint64_t op_start_chunk = op->offset / s->granularity; 160 uint64_t op_nb_chunks = DIV_ROUND_UP(op->offset + op->bytes, 161 s->granularity) - 162 op_start_chunk; 163 164 if (op == self) { 165 continue; 166 } 167 168 if (ranges_overlap(self_start_chunk, self_nb_chunks, 169 op_start_chunk, op_nb_chunks)) 170 { 171 if (self) { 172 /* 173 * If the operation is already (indirectly) waiting for us, 174 * or will wait for us as soon as it wakes up, then just go 175 * on (instead of producing a deadlock in the former case). 176 */ 177 if (op->waiting_for_op) { 178 continue; 179 } 180 181 self->waiting_for_op = op; 182 } 183 184 qemu_co_queue_wait(&op->waiting_requests, NULL); 185 186 if (self) { 187 self->waiting_for_op = NULL; 188 } 189 190 break; 191 } 192 } 193 } 194 } 195 196 static void coroutine_fn mirror_iteration_done(MirrorOp *op, int ret) 197 { 198 MirrorBlockJob *s = op->s; 199 struct iovec *iov; 200 int64_t chunk_num; 201 int i, nb_chunks; 202 203 trace_mirror_iteration_done(s, op->offset, op->bytes, ret); 204 205 s->in_flight--; 206 s->bytes_in_flight -= op->bytes; 207 iov = op->qiov.iov; 208 for (i = 0; i < op->qiov.niov; i++) { 209 MirrorBuffer *buf = (MirrorBuffer *) iov[i].iov_base; 210 QSIMPLEQ_INSERT_TAIL(&s->buf_free, buf, next); 211 s->buf_free_count++; 212 } 213 214 chunk_num = op->offset / s->granularity; 215 nb_chunks = DIV_ROUND_UP(op->bytes, s->granularity); 216 217 bitmap_clear(s->in_flight_bitmap, chunk_num, nb_chunks); 218 QTAILQ_REMOVE(&s->ops_in_flight, op, next); 219 if (ret >= 0) { 220 if (s->cow_bitmap) { 221 bitmap_set(s->cow_bitmap, chunk_num, nb_chunks); 222 } 223 if (!s->initial_zeroing_ongoing) { 224 job_progress_update(&s->common.job, op->bytes); 225 } 226 } 227 qemu_iovec_destroy(&op->qiov); 228 229 qemu_co_queue_restart_all(&op->waiting_requests); 230 g_free(op); 231 } 232 233 static void coroutine_fn mirror_write_complete(MirrorOp *op, int ret) 234 { 235 MirrorBlockJob *s = op->s; 236 237 if (ret < 0) { 238 BlockErrorAction action; 239 240 bdrv_set_dirty_bitmap(s->dirty_bitmap, op->offset, op->bytes); 241 action = mirror_error_action(s, false, -ret); 242 if (action == BLOCK_ERROR_ACTION_REPORT && s->ret >= 0) { 243 s->ret = ret; 244 } 245 } 246 247 mirror_iteration_done(op, ret); 248 } 249 250 static void coroutine_fn mirror_read_complete(MirrorOp *op, int ret) 251 { 252 MirrorBlockJob *s = op->s; 253 254 if (ret < 0) { 255 BlockErrorAction action; 256 257 bdrv_set_dirty_bitmap(s->dirty_bitmap, op->offset, op->bytes); 258 action = mirror_error_action(s, true, -ret); 259 if (action == BLOCK_ERROR_ACTION_REPORT && s->ret >= 0) { 260 s->ret = ret; 261 } 262 263 mirror_iteration_done(op, ret); 264 return; 265 } 266 267 ret = blk_co_pwritev(s->target, op->offset, op->qiov.size, &op->qiov, 0); 268 mirror_write_complete(op, ret); 269 } 270 271 /* Clip bytes relative to offset to not exceed end-of-file */ 272 static inline int64_t mirror_clip_bytes(MirrorBlockJob *s, 273 int64_t offset, 274 int64_t bytes) 275 { 276 return MIN(bytes, s->bdev_length - offset); 277 } 278 279 /* Round offset and/or bytes to target cluster if COW is needed, and 280 * return the offset of the adjusted tail against original. */ 281 static int coroutine_fn mirror_cow_align(MirrorBlockJob *s, int64_t *offset, 282 uint64_t *bytes) 283 { 284 bool need_cow; 285 int ret = 0; 286 int64_t align_offset = *offset; 287 int64_t align_bytes = *bytes; 288 int max_bytes = s->granularity * s->max_iov; 289 290 need_cow = !test_bit(*offset / s->granularity, s->cow_bitmap); 291 need_cow |= !test_bit((*offset + *bytes - 1) / s->granularity, 292 s->cow_bitmap); 293 if (need_cow) { 294 bdrv_round_to_subclusters(blk_bs(s->target), *offset, *bytes, 295 &align_offset, &align_bytes); 296 } 297 298 if (align_bytes > max_bytes) { 299 align_bytes = max_bytes; 300 if (need_cow) { 301 align_bytes = QEMU_ALIGN_DOWN(align_bytes, s->target_cluster_size); 302 } 303 } 304 /* Clipping may result in align_bytes unaligned to chunk boundary, but 305 * that doesn't matter because it's already the end of source image. */ 306 align_bytes = mirror_clip_bytes(s, align_offset, align_bytes); 307 308 ret = align_offset + align_bytes - (*offset + *bytes); 309 *offset = align_offset; 310 *bytes = align_bytes; 311 assert(ret >= 0); 312 return ret; 313 } 314 315 static inline void coroutine_fn 316 mirror_wait_for_free_in_flight_slot(MirrorBlockJob *s) 317 { 318 MirrorOp *op; 319 320 QTAILQ_FOREACH(op, &s->ops_in_flight, next) { 321 /* 322 * Do not wait on pseudo ops, because it may in turn wait on 323 * some other operation to start, which may in fact be the 324 * caller of this function. Since there is only one pseudo op 325 * at any given time, we will always find some real operation 326 * to wait on. 327 * Also, do not wait on active operations, because they do not 328 * use up in-flight slots. 329 */ 330 if (!op->is_pseudo_op && op->is_in_flight && !op->is_active_write) { 331 qemu_co_queue_wait(&op->waiting_requests, NULL); 332 return; 333 } 334 } 335 abort(); 336 } 337 338 /* Perform a mirror copy operation. 339 * 340 * *op->bytes_handled is set to the number of bytes copied after and 341 * including offset, excluding any bytes copied prior to offset due 342 * to alignment. This will be op->bytes if no alignment is necessary, 343 * or (new_end - op->offset) if the tail is rounded up or down due to 344 * alignment or buffer limit. 345 */ 346 static void coroutine_fn mirror_co_read(void *opaque) 347 { 348 MirrorOp *op = opaque; 349 MirrorBlockJob *s = op->s; 350 int nb_chunks; 351 uint64_t ret; 352 uint64_t max_bytes; 353 354 max_bytes = s->granularity * s->max_iov; 355 356 /* We can only handle as much as buf_size at a time. */ 357 op->bytes = MIN(s->buf_size, MIN(max_bytes, op->bytes)); 358 assert(op->bytes); 359 assert(op->bytes < BDRV_REQUEST_MAX_BYTES); 360 *op->bytes_handled = op->bytes; 361 362 if (s->cow_bitmap) { 363 *op->bytes_handled += mirror_cow_align(s, &op->offset, &op->bytes); 364 } 365 /* Cannot exceed BDRV_REQUEST_MAX_BYTES + INT_MAX */ 366 assert(*op->bytes_handled <= UINT_MAX); 367 assert(op->bytes <= s->buf_size); 368 /* The offset is granularity-aligned because: 369 * 1) Caller passes in aligned values; 370 * 2) mirror_cow_align is used only when target cluster is larger. */ 371 assert(QEMU_IS_ALIGNED(op->offset, s->granularity)); 372 /* The range is sector-aligned, since bdrv_getlength() rounds up. */ 373 assert(QEMU_IS_ALIGNED(op->bytes, BDRV_SECTOR_SIZE)); 374 nb_chunks = DIV_ROUND_UP(op->bytes, s->granularity); 375 376 while (s->buf_free_count < nb_chunks) { 377 trace_mirror_yield_in_flight(s, op->offset, s->in_flight); 378 mirror_wait_for_free_in_flight_slot(s); 379 } 380 381 /* Now make a QEMUIOVector taking enough granularity-sized chunks 382 * from s->buf_free. 383 */ 384 qemu_iovec_init(&op->qiov, nb_chunks); 385 while (nb_chunks-- > 0) { 386 MirrorBuffer *buf = QSIMPLEQ_FIRST(&s->buf_free); 387 size_t remaining = op->bytes - op->qiov.size; 388 389 QSIMPLEQ_REMOVE_HEAD(&s->buf_free, next); 390 s->buf_free_count--; 391 qemu_iovec_add(&op->qiov, buf, MIN(s->granularity, remaining)); 392 } 393 394 /* Copy the dirty cluster. */ 395 s->in_flight++; 396 s->bytes_in_flight += op->bytes; 397 op->is_in_flight = true; 398 trace_mirror_one_iteration(s, op->offset, op->bytes); 399 400 WITH_GRAPH_RDLOCK_GUARD() { 401 ret = bdrv_co_preadv(s->mirror_top_bs->backing, op->offset, op->bytes, 402 &op->qiov, 0); 403 } 404 mirror_read_complete(op, ret); 405 } 406 407 static void coroutine_fn mirror_co_zero(void *opaque) 408 { 409 MirrorOp *op = opaque; 410 int ret; 411 412 op->s->in_flight++; 413 op->s->bytes_in_flight += op->bytes; 414 *op->bytes_handled = op->bytes; 415 op->is_in_flight = true; 416 417 ret = blk_co_pwrite_zeroes(op->s->target, op->offset, op->bytes, 418 op->s->unmap ? BDRV_REQ_MAY_UNMAP : 0); 419 mirror_write_complete(op, ret); 420 } 421 422 static void coroutine_fn mirror_co_discard(void *opaque) 423 { 424 MirrorOp *op = opaque; 425 int ret; 426 427 op->s->in_flight++; 428 op->s->bytes_in_flight += op->bytes; 429 *op->bytes_handled = op->bytes; 430 op->is_in_flight = true; 431 432 ret = blk_co_pdiscard(op->s->target, op->offset, op->bytes); 433 mirror_write_complete(op, ret); 434 } 435 436 static unsigned mirror_perform(MirrorBlockJob *s, int64_t offset, 437 unsigned bytes, MirrorMethod mirror_method) 438 { 439 MirrorOp *op; 440 Coroutine *co; 441 int64_t bytes_handled = -1; 442 443 op = g_new(MirrorOp, 1); 444 *op = (MirrorOp){ 445 .s = s, 446 .offset = offset, 447 .bytes = bytes, 448 .bytes_handled = &bytes_handled, 449 }; 450 qemu_co_queue_init(&op->waiting_requests); 451 452 switch (mirror_method) { 453 case MIRROR_METHOD_COPY: 454 co = qemu_coroutine_create(mirror_co_read, op); 455 break; 456 case MIRROR_METHOD_ZERO: 457 co = qemu_coroutine_create(mirror_co_zero, op); 458 break; 459 case MIRROR_METHOD_DISCARD: 460 co = qemu_coroutine_create(mirror_co_discard, op); 461 break; 462 default: 463 abort(); 464 } 465 op->co = co; 466 467 QTAILQ_INSERT_TAIL(&s->ops_in_flight, op, next); 468 qemu_coroutine_enter(co); 469 /* At this point, ownership of op has been moved to the coroutine 470 * and the object may already be freed */ 471 472 /* Assert that this value has been set */ 473 assert(bytes_handled >= 0); 474 475 /* Same assertion as in mirror_co_read() (and for mirror_co_read() 476 * and mirror_co_discard(), bytes_handled == op->bytes, which 477 * is the @bytes parameter given to this function) */ 478 assert(bytes_handled <= UINT_MAX); 479 return bytes_handled; 480 } 481 482 static void coroutine_fn mirror_iteration(MirrorBlockJob *s) 483 { 484 BlockDriverState *source = s->mirror_top_bs->backing->bs; 485 MirrorOp *pseudo_op; 486 int64_t offset; 487 /* At least the first dirty chunk is mirrored in one iteration. */ 488 int nb_chunks = 1; 489 bool write_zeroes_ok = bdrv_can_write_zeroes_with_unmap(blk_bs(s->target)); 490 int max_io_bytes = MAX(s->buf_size / MAX_IN_FLIGHT, MAX_IO_BYTES); 491 492 bdrv_dirty_bitmap_lock(s->dirty_bitmap); 493 offset = bdrv_dirty_iter_next(s->dbi); 494 if (offset < 0) { 495 bdrv_set_dirty_iter(s->dbi, 0); 496 offset = bdrv_dirty_iter_next(s->dbi); 497 trace_mirror_restart_iter(s, bdrv_get_dirty_count(s->dirty_bitmap)); 498 assert(offset >= 0); 499 } 500 bdrv_dirty_bitmap_unlock(s->dirty_bitmap); 501 502 /* 503 * Wait for concurrent requests to @offset. The next loop will limit the 504 * copied area based on in_flight_bitmap so we only copy an area that does 505 * not overlap with concurrent in-flight requests. Still, we would like to 506 * copy something, so wait until there are at least no more requests to the 507 * very beginning of the area. 508 */ 509 mirror_wait_on_conflicts(NULL, s, offset, 1); 510 511 job_pause_point(&s->common.job); 512 513 /* Find the number of consecutive dirty chunks following the first dirty 514 * one, and wait for in flight requests in them. */ 515 bdrv_dirty_bitmap_lock(s->dirty_bitmap); 516 while (nb_chunks * s->granularity < s->buf_size) { 517 int64_t next_dirty; 518 int64_t next_offset = offset + nb_chunks * s->granularity; 519 int64_t next_chunk = next_offset / s->granularity; 520 if (next_offset >= s->bdev_length || 521 !bdrv_dirty_bitmap_get_locked(s->dirty_bitmap, next_offset)) { 522 break; 523 } 524 if (test_bit(next_chunk, s->in_flight_bitmap)) { 525 break; 526 } 527 528 next_dirty = bdrv_dirty_iter_next(s->dbi); 529 if (next_dirty > next_offset || next_dirty < 0) { 530 /* The bitmap iterator's cache is stale, refresh it */ 531 bdrv_set_dirty_iter(s->dbi, next_offset); 532 next_dirty = bdrv_dirty_iter_next(s->dbi); 533 } 534 assert(next_dirty == next_offset); 535 nb_chunks++; 536 } 537 538 /* Clear dirty bits before querying the block status, because 539 * calling bdrv_block_status_above could yield - if some blocks are 540 * marked dirty in this window, we need to know. 541 */ 542 bdrv_reset_dirty_bitmap_locked(s->dirty_bitmap, offset, 543 nb_chunks * s->granularity); 544 bdrv_dirty_bitmap_unlock(s->dirty_bitmap); 545 546 /* Before claiming an area in the in-flight bitmap, we have to 547 * create a MirrorOp for it so that conflicting requests can wait 548 * for it. mirror_perform() will create the real MirrorOps later, 549 * for now we just create a pseudo operation that will wake up all 550 * conflicting requests once all real operations have been 551 * launched. */ 552 pseudo_op = g_new(MirrorOp, 1); 553 *pseudo_op = (MirrorOp){ 554 .offset = offset, 555 .bytes = nb_chunks * s->granularity, 556 .is_pseudo_op = true, 557 }; 558 qemu_co_queue_init(&pseudo_op->waiting_requests); 559 QTAILQ_INSERT_TAIL(&s->ops_in_flight, pseudo_op, next); 560 561 bitmap_set(s->in_flight_bitmap, offset / s->granularity, nb_chunks); 562 while (nb_chunks > 0 && offset < s->bdev_length) { 563 int ret; 564 int64_t io_bytes; 565 int64_t io_bytes_acct; 566 MirrorMethod mirror_method = MIRROR_METHOD_COPY; 567 568 assert(!(offset % s->granularity)); 569 WITH_GRAPH_RDLOCK_GUARD() { 570 ret = bdrv_co_block_status_above(source, NULL, offset, 571 nb_chunks * s->granularity, 572 &io_bytes, NULL, NULL); 573 } 574 if (ret < 0) { 575 io_bytes = MIN(nb_chunks * s->granularity, max_io_bytes); 576 } else if (ret & BDRV_BLOCK_DATA) { 577 io_bytes = MIN(io_bytes, max_io_bytes); 578 } 579 580 io_bytes -= io_bytes % s->granularity; 581 if (io_bytes < s->granularity) { 582 io_bytes = s->granularity; 583 } else if (ret >= 0 && !(ret & BDRV_BLOCK_DATA)) { 584 int64_t target_offset; 585 int64_t target_bytes; 586 WITH_GRAPH_RDLOCK_GUARD() { 587 bdrv_round_to_subclusters(blk_bs(s->target), offset, io_bytes, 588 &target_offset, &target_bytes); 589 } 590 if (target_offset == offset && 591 target_bytes == io_bytes) { 592 mirror_method = ret & BDRV_BLOCK_ZERO ? 593 MIRROR_METHOD_ZERO : 594 MIRROR_METHOD_DISCARD; 595 } 596 } 597 598 while (s->in_flight >= MAX_IN_FLIGHT) { 599 trace_mirror_yield_in_flight(s, offset, s->in_flight); 600 mirror_wait_for_free_in_flight_slot(s); 601 } 602 603 if (s->ret < 0) { 604 ret = 0; 605 goto fail; 606 } 607 608 io_bytes = mirror_clip_bytes(s, offset, io_bytes); 609 io_bytes = mirror_perform(s, offset, io_bytes, mirror_method); 610 if (mirror_method != MIRROR_METHOD_COPY && write_zeroes_ok) { 611 io_bytes_acct = 0; 612 } else { 613 io_bytes_acct = io_bytes; 614 } 615 assert(io_bytes); 616 offset += io_bytes; 617 nb_chunks -= DIV_ROUND_UP(io_bytes, s->granularity); 618 block_job_ratelimit_processed_bytes(&s->common, io_bytes_acct); 619 } 620 621 fail: 622 QTAILQ_REMOVE(&s->ops_in_flight, pseudo_op, next); 623 qemu_co_queue_restart_all(&pseudo_op->waiting_requests); 624 g_free(pseudo_op); 625 } 626 627 static void mirror_free_init(MirrorBlockJob *s) 628 { 629 int granularity = s->granularity; 630 size_t buf_size = s->buf_size; 631 uint8_t *buf = s->buf; 632 633 assert(s->buf_free_count == 0); 634 QSIMPLEQ_INIT(&s->buf_free); 635 while (buf_size != 0) { 636 MirrorBuffer *cur = (MirrorBuffer *)buf; 637 QSIMPLEQ_INSERT_TAIL(&s->buf_free, cur, next); 638 s->buf_free_count++; 639 buf_size -= granularity; 640 buf += granularity; 641 } 642 } 643 644 /* This is also used for the .pause callback. There is no matching 645 * mirror_resume() because mirror_run() will begin iterating again 646 * when the job is resumed. 647 */ 648 static void coroutine_fn mirror_wait_for_all_io(MirrorBlockJob *s) 649 { 650 while (s->in_flight > 0) { 651 mirror_wait_for_free_in_flight_slot(s); 652 } 653 } 654 655 /** 656 * mirror_exit_common: handle both abort() and prepare() cases. 657 * for .prepare, returns 0 on success and -errno on failure. 658 * for .abort cases, denoted by abort = true, MUST return 0. 659 */ 660 static int mirror_exit_common(Job *job) 661 { 662 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); 663 BlockJob *bjob = &s->common; 664 MirrorBDSOpaque *bs_opaque; 665 AioContext *replace_aio_context = NULL; 666 BlockDriverState *src; 667 BlockDriverState *target_bs; 668 BlockDriverState *mirror_top_bs; 669 Error *local_err = NULL; 670 bool abort = job->ret < 0; 671 int ret = 0; 672 673 GLOBAL_STATE_CODE(); 674 675 if (s->prepared) { 676 return 0; 677 } 678 s->prepared = true; 679 680 aio_context_acquire(qemu_get_aio_context()); 681 682 mirror_top_bs = s->mirror_top_bs; 683 bs_opaque = mirror_top_bs->opaque; 684 src = mirror_top_bs->backing->bs; 685 target_bs = blk_bs(s->target); 686 687 if (bdrv_chain_contains(src, target_bs)) { 688 bdrv_unfreeze_backing_chain(mirror_top_bs, target_bs); 689 } 690 691 bdrv_release_dirty_bitmap(s->dirty_bitmap); 692 693 /* Make sure that the source BDS doesn't go away during bdrv_replace_node, 694 * before we can call bdrv_drained_end */ 695 bdrv_ref(src); 696 bdrv_ref(mirror_top_bs); 697 bdrv_ref(target_bs); 698 699 /* 700 * Remove target parent that still uses BLK_PERM_WRITE/RESIZE before 701 * inserting target_bs at s->to_replace, where we might not be able to get 702 * these permissions. 703 */ 704 blk_unref(s->target); 705 s->target = NULL; 706 707 /* We don't access the source any more. Dropping any WRITE/RESIZE is 708 * required before it could become a backing file of target_bs. Not having 709 * these permissions any more means that we can't allow any new requests on 710 * mirror_top_bs from now on, so keep it drained. */ 711 bdrv_drained_begin(mirror_top_bs); 712 bs_opaque->stop = true; 713 714 bdrv_graph_rdlock_main_loop(); 715 bdrv_child_refresh_perms(mirror_top_bs, mirror_top_bs->backing, 716 &error_abort); 717 bdrv_graph_rdunlock_main_loop(); 718 719 if (!abort && s->backing_mode == MIRROR_SOURCE_BACKING_CHAIN) { 720 BlockDriverState *backing = s->is_none_mode ? src : s->base; 721 BlockDriverState *unfiltered_target = bdrv_skip_filters(target_bs); 722 723 if (bdrv_cow_bs(unfiltered_target) != backing) { 724 bdrv_set_backing_hd(unfiltered_target, backing, &local_err); 725 if (local_err) { 726 error_report_err(local_err); 727 local_err = NULL; 728 ret = -EPERM; 729 } 730 } 731 } else if (!abort && s->backing_mode == MIRROR_OPEN_BACKING_CHAIN) { 732 assert(!bdrv_backing_chain_next(target_bs)); 733 ret = bdrv_open_backing_file(bdrv_skip_filters(target_bs), NULL, 734 "backing", &local_err); 735 if (ret < 0) { 736 error_report_err(local_err); 737 local_err = NULL; 738 } 739 } 740 741 if (s->to_replace) { 742 replace_aio_context = bdrv_get_aio_context(s->to_replace); 743 aio_context_acquire(replace_aio_context); 744 } 745 746 if (s->should_complete && !abort) { 747 BlockDriverState *to_replace = s->to_replace ?: src; 748 bool ro = bdrv_is_read_only(to_replace); 749 750 if (ro != bdrv_is_read_only(target_bs)) { 751 bdrv_reopen_set_read_only(target_bs, ro, NULL); 752 } 753 754 /* The mirror job has no requests in flight any more, but we need to 755 * drain potential other users of the BDS before changing the graph. */ 756 assert(s->in_drain); 757 bdrv_drained_begin(target_bs); 758 /* 759 * Cannot use check_to_replace_node() here, because that would 760 * check for an op blocker on @to_replace, and we have our own 761 * there. 762 * 763 * TODO Pull out the writer lock from bdrv_replace_node() to here 764 */ 765 bdrv_graph_rdlock_main_loop(); 766 if (bdrv_recurse_can_replace(src, to_replace)) { 767 bdrv_replace_node(to_replace, target_bs, &local_err); 768 } else { 769 error_setg(&local_err, "Can no longer replace '%s' by '%s', " 770 "because it can no longer be guaranteed that doing so " 771 "would not lead to an abrupt change of visible data", 772 to_replace->node_name, target_bs->node_name); 773 } 774 bdrv_graph_rdunlock_main_loop(); 775 bdrv_drained_end(target_bs); 776 if (local_err) { 777 error_report_err(local_err); 778 ret = -EPERM; 779 } 780 } 781 if (s->to_replace) { 782 bdrv_op_unblock_all(s->to_replace, s->replace_blocker); 783 error_free(s->replace_blocker); 784 bdrv_unref(s->to_replace); 785 } 786 if (replace_aio_context) { 787 aio_context_release(replace_aio_context); 788 } 789 g_free(s->replaces); 790 bdrv_unref(target_bs); 791 792 /* 793 * Remove the mirror filter driver from the graph. Before this, get rid of 794 * the blockers on the intermediate nodes so that the resulting state is 795 * valid. 796 */ 797 block_job_remove_all_bdrv(bjob); 798 bdrv_replace_node(mirror_top_bs, mirror_top_bs->backing->bs, &error_abort); 799 800 bs_opaque->job = NULL; 801 802 bdrv_drained_end(src); 803 bdrv_drained_end(mirror_top_bs); 804 s->in_drain = false; 805 bdrv_unref(mirror_top_bs); 806 bdrv_unref(src); 807 808 aio_context_release(qemu_get_aio_context()); 809 810 return ret; 811 } 812 813 static int mirror_prepare(Job *job) 814 { 815 return mirror_exit_common(job); 816 } 817 818 static void mirror_abort(Job *job) 819 { 820 int ret = mirror_exit_common(job); 821 assert(ret == 0); 822 } 823 824 static void coroutine_fn mirror_throttle(MirrorBlockJob *s) 825 { 826 int64_t now = qemu_clock_get_ns(QEMU_CLOCK_REALTIME); 827 828 if (now - s->last_pause_ns > BLOCK_JOB_SLICE_TIME) { 829 s->last_pause_ns = now; 830 job_sleep_ns(&s->common.job, 0); 831 } else { 832 job_pause_point(&s->common.job); 833 } 834 } 835 836 static int coroutine_fn mirror_dirty_init(MirrorBlockJob *s) 837 { 838 int64_t offset; 839 BlockDriverState *bs = s->mirror_top_bs->backing->bs; 840 BlockDriverState *target_bs = blk_bs(s->target); 841 int ret; 842 int64_t count; 843 844 if (s->zero_target) { 845 if (!bdrv_can_write_zeroes_with_unmap(target_bs)) { 846 bdrv_set_dirty_bitmap(s->dirty_bitmap, 0, s->bdev_length); 847 return 0; 848 } 849 850 s->initial_zeroing_ongoing = true; 851 for (offset = 0; offset < s->bdev_length; ) { 852 int bytes = MIN(s->bdev_length - offset, 853 QEMU_ALIGN_DOWN(INT_MAX, s->granularity)); 854 855 mirror_throttle(s); 856 857 if (job_is_cancelled(&s->common.job)) { 858 s->initial_zeroing_ongoing = false; 859 return 0; 860 } 861 862 if (s->in_flight >= MAX_IN_FLIGHT) { 863 trace_mirror_yield(s, UINT64_MAX, s->buf_free_count, 864 s->in_flight); 865 mirror_wait_for_free_in_flight_slot(s); 866 continue; 867 } 868 869 mirror_perform(s, offset, bytes, MIRROR_METHOD_ZERO); 870 offset += bytes; 871 } 872 873 mirror_wait_for_all_io(s); 874 s->initial_zeroing_ongoing = false; 875 } 876 877 /* First part, loop on the sectors and initialize the dirty bitmap. */ 878 for (offset = 0; offset < s->bdev_length; ) { 879 /* Just to make sure we are not exceeding int limit. */ 880 int bytes = MIN(s->bdev_length - offset, 881 QEMU_ALIGN_DOWN(INT_MAX, s->granularity)); 882 883 mirror_throttle(s); 884 885 if (job_is_cancelled(&s->common.job)) { 886 return 0; 887 } 888 889 WITH_GRAPH_RDLOCK_GUARD() { 890 ret = bdrv_co_is_allocated_above(bs, s->base_overlay, true, offset, 891 bytes, &count); 892 } 893 if (ret < 0) { 894 return ret; 895 } 896 897 assert(count); 898 if (ret > 0) { 899 bdrv_set_dirty_bitmap(s->dirty_bitmap, offset, count); 900 } 901 offset += count; 902 } 903 return 0; 904 } 905 906 /* Called when going out of the streaming phase to flush the bulk of the 907 * data to the medium, or just before completing. 908 */ 909 static int coroutine_fn mirror_flush(MirrorBlockJob *s) 910 { 911 int ret = blk_co_flush(s->target); 912 if (ret < 0) { 913 if (mirror_error_action(s, false, -ret) == BLOCK_ERROR_ACTION_REPORT) { 914 s->ret = ret; 915 } 916 } 917 return ret; 918 } 919 920 static int coroutine_fn mirror_run(Job *job, Error **errp) 921 { 922 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); 923 BlockDriverState *bs = s->mirror_top_bs->backing->bs; 924 MirrorBDSOpaque *mirror_top_opaque = s->mirror_top_bs->opaque; 925 BlockDriverState *target_bs = blk_bs(s->target); 926 bool need_drain = true; 927 BlockDeviceIoStatus iostatus; 928 int64_t length; 929 int64_t target_length; 930 BlockDriverInfo bdi; 931 char backing_filename[2]; /* we only need 2 characters because we are only 932 checking for a NULL string */ 933 int ret = 0; 934 935 if (job_is_cancelled(&s->common.job)) { 936 goto immediate_exit; 937 } 938 939 bdrv_graph_co_rdlock(); 940 s->bdev_length = bdrv_co_getlength(bs); 941 bdrv_graph_co_rdunlock(); 942 943 if (s->bdev_length < 0) { 944 ret = s->bdev_length; 945 goto immediate_exit; 946 } 947 948 target_length = blk_co_getlength(s->target); 949 if (target_length < 0) { 950 ret = target_length; 951 goto immediate_exit; 952 } 953 954 /* Active commit must resize the base image if its size differs from the 955 * active layer. */ 956 if (s->base == blk_bs(s->target)) { 957 if (s->bdev_length > target_length) { 958 ret = blk_co_truncate(s->target, s->bdev_length, false, 959 PREALLOC_MODE_OFF, 0, NULL); 960 if (ret < 0) { 961 goto immediate_exit; 962 } 963 } 964 } else if (s->bdev_length != target_length) { 965 error_setg(errp, "Source and target image have different sizes"); 966 ret = -EINVAL; 967 goto immediate_exit; 968 } 969 970 if (s->bdev_length == 0) { 971 /* Transition to the READY state and wait for complete. */ 972 job_transition_to_ready(&s->common.job); 973 qatomic_set(&s->actively_synced, true); 974 while (!job_cancel_requested(&s->common.job) && !s->should_complete) { 975 job_yield(&s->common.job); 976 } 977 goto immediate_exit; 978 } 979 980 length = DIV_ROUND_UP(s->bdev_length, s->granularity); 981 s->in_flight_bitmap = bitmap_new(length); 982 983 /* If we have no backing file yet in the destination, we cannot let 984 * the destination do COW. Instead, we copy sectors around the 985 * dirty data if needed. We need a bitmap to do that. 986 */ 987 bdrv_get_backing_filename(target_bs, backing_filename, 988 sizeof(backing_filename)); 989 bdrv_graph_co_rdlock(); 990 if (!bdrv_co_get_info(target_bs, &bdi) && bdi.cluster_size) { 991 s->target_cluster_size = bdi.cluster_size; 992 } else { 993 s->target_cluster_size = BDRV_SECTOR_SIZE; 994 } 995 bdrv_graph_co_rdunlock(); 996 if (backing_filename[0] && !bdrv_backing_chain_next(target_bs) && 997 s->granularity < s->target_cluster_size) { 998 s->buf_size = MAX(s->buf_size, s->target_cluster_size); 999 s->cow_bitmap = bitmap_new(length); 1000 } 1001 s->max_iov = MIN(bs->bl.max_iov, target_bs->bl.max_iov); 1002 1003 s->buf = qemu_try_blockalign(bs, s->buf_size); 1004 if (s->buf == NULL) { 1005 ret = -ENOMEM; 1006 goto immediate_exit; 1007 } 1008 1009 mirror_free_init(s); 1010 1011 s->last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME); 1012 if (!s->is_none_mode) { 1013 ret = mirror_dirty_init(s); 1014 if (ret < 0 || job_is_cancelled(&s->common.job)) { 1015 goto immediate_exit; 1016 } 1017 } 1018 1019 /* 1020 * Only now the job is fully initialised and mirror_top_bs should start 1021 * accessing it. 1022 */ 1023 mirror_top_opaque->job = s; 1024 1025 assert(!s->dbi); 1026 s->dbi = bdrv_dirty_iter_new(s->dirty_bitmap); 1027 for (;;) { 1028 int64_t cnt, delta; 1029 bool should_complete; 1030 1031 if (s->ret < 0) { 1032 ret = s->ret; 1033 goto immediate_exit; 1034 } 1035 1036 job_pause_point(&s->common.job); 1037 1038 if (job_is_cancelled(&s->common.job)) { 1039 ret = 0; 1040 goto immediate_exit; 1041 } 1042 1043 cnt = bdrv_get_dirty_count(s->dirty_bitmap); 1044 /* cnt is the number of dirty bytes remaining and s->bytes_in_flight is 1045 * the number of bytes currently being processed; together those are 1046 * the current remaining operation length */ 1047 job_progress_set_remaining(&s->common.job, 1048 s->bytes_in_flight + cnt + 1049 s->active_write_bytes_in_flight); 1050 1051 /* Note that even when no rate limit is applied we need to yield 1052 * periodically with no pending I/O so that bdrv_drain_all() returns. 1053 * We do so every BLKOCK_JOB_SLICE_TIME nanoseconds, or when there is 1054 * an error, or when the source is clean, whichever comes first. */ 1055 delta = qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - s->last_pause_ns; 1056 WITH_JOB_LOCK_GUARD() { 1057 iostatus = s->common.iostatus; 1058 } 1059 if (delta < BLOCK_JOB_SLICE_TIME && 1060 iostatus == BLOCK_DEVICE_IO_STATUS_OK) { 1061 if (s->in_flight >= MAX_IN_FLIGHT || s->buf_free_count == 0 || 1062 (cnt == 0 && s->in_flight > 0)) { 1063 trace_mirror_yield(s, cnt, s->buf_free_count, s->in_flight); 1064 mirror_wait_for_free_in_flight_slot(s); 1065 continue; 1066 } else if (cnt != 0) { 1067 mirror_iteration(s); 1068 } 1069 } 1070 1071 should_complete = false; 1072 if (s->in_flight == 0 && cnt == 0) { 1073 trace_mirror_before_flush(s); 1074 if (!job_is_ready(&s->common.job)) { 1075 if (mirror_flush(s) < 0) { 1076 /* Go check s->ret. */ 1077 continue; 1078 } 1079 /* We're out of the streaming phase. From now on, if the job 1080 * is cancelled we will actually complete all pending I/O and 1081 * report completion. This way, block-job-cancel will leave 1082 * the target in a consistent state. 1083 */ 1084 job_transition_to_ready(&s->common.job); 1085 } 1086 if (qatomic_read(&s->copy_mode) != MIRROR_COPY_MODE_BACKGROUND) { 1087 qatomic_set(&s->actively_synced, true); 1088 } 1089 1090 should_complete = s->should_complete || 1091 job_cancel_requested(&s->common.job); 1092 cnt = bdrv_get_dirty_count(s->dirty_bitmap); 1093 } 1094 1095 if (cnt == 0 && should_complete) { 1096 /* The dirty bitmap is not updated while operations are pending. 1097 * If we're about to exit, wait for pending operations before 1098 * calling bdrv_get_dirty_count(bs), or we may exit while the 1099 * source has dirty data to copy! 1100 * 1101 * Note that I/O can be submitted by the guest while 1102 * mirror_populate runs, so pause it now. Before deciding 1103 * whether to switch to target check one last time if I/O has 1104 * come in the meanwhile, and if not flush the data to disk. 1105 */ 1106 trace_mirror_before_drain(s, cnt); 1107 1108 s->in_drain = true; 1109 bdrv_drained_begin(bs); 1110 1111 /* Must be zero because we are drained */ 1112 assert(s->in_active_write_counter == 0); 1113 1114 cnt = bdrv_get_dirty_count(s->dirty_bitmap); 1115 if (cnt > 0 || mirror_flush(s) < 0) { 1116 bdrv_drained_end(bs); 1117 s->in_drain = false; 1118 continue; 1119 } 1120 1121 /* The two disks are in sync. Exit and report successful 1122 * completion. 1123 */ 1124 assert(QLIST_EMPTY(&bs->tracked_requests)); 1125 need_drain = false; 1126 break; 1127 } 1128 1129 if (job_is_ready(&s->common.job) && !should_complete) { 1130 if (s->in_flight == 0 && cnt == 0) { 1131 trace_mirror_before_sleep(s, cnt, job_is_ready(&s->common.job), 1132 BLOCK_JOB_SLICE_TIME); 1133 job_sleep_ns(&s->common.job, BLOCK_JOB_SLICE_TIME); 1134 } 1135 } else { 1136 block_job_ratelimit_sleep(&s->common); 1137 } 1138 s->last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME); 1139 } 1140 1141 immediate_exit: 1142 if (s->in_flight > 0) { 1143 /* We get here only if something went wrong. Either the job failed, 1144 * or it was cancelled prematurely so that we do not guarantee that 1145 * the target is a copy of the source. 1146 */ 1147 assert(ret < 0 || job_is_cancelled(&s->common.job)); 1148 assert(need_drain); 1149 mirror_wait_for_all_io(s); 1150 } 1151 1152 assert(s->in_flight == 0); 1153 qemu_vfree(s->buf); 1154 g_free(s->cow_bitmap); 1155 g_free(s->in_flight_bitmap); 1156 bdrv_dirty_iter_free(s->dbi); 1157 1158 if (need_drain) { 1159 s->in_drain = true; 1160 bdrv_drained_begin(bs); 1161 } 1162 1163 return ret; 1164 } 1165 1166 static void mirror_complete(Job *job, Error **errp) 1167 { 1168 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); 1169 1170 if (!job_is_ready(job)) { 1171 error_setg(errp, "The active block job '%s' cannot be completed", 1172 job->id); 1173 return; 1174 } 1175 1176 /* block all operations on to_replace bs */ 1177 if (s->replaces) { 1178 AioContext *replace_aio_context; 1179 1180 s->to_replace = bdrv_find_node(s->replaces); 1181 if (!s->to_replace) { 1182 error_setg(errp, "Node name '%s' not found", s->replaces); 1183 return; 1184 } 1185 1186 replace_aio_context = bdrv_get_aio_context(s->to_replace); 1187 aio_context_acquire(replace_aio_context); 1188 1189 /* TODO Translate this into child freeze system. */ 1190 error_setg(&s->replace_blocker, 1191 "block device is in use by block-job-complete"); 1192 bdrv_op_block_all(s->to_replace, s->replace_blocker); 1193 bdrv_ref(s->to_replace); 1194 1195 aio_context_release(replace_aio_context); 1196 } 1197 1198 s->should_complete = true; 1199 1200 /* If the job is paused, it will be re-entered when it is resumed */ 1201 WITH_JOB_LOCK_GUARD() { 1202 if (!job->paused) { 1203 job_enter_cond_locked(job, NULL); 1204 } 1205 } 1206 } 1207 1208 static void coroutine_fn mirror_pause(Job *job) 1209 { 1210 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); 1211 1212 mirror_wait_for_all_io(s); 1213 } 1214 1215 static bool mirror_drained_poll(BlockJob *job) 1216 { 1217 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common); 1218 1219 /* If the job isn't paused nor cancelled, we can't be sure that it won't 1220 * issue more requests. We make an exception if we've reached this point 1221 * from one of our own drain sections, to avoid a deadlock waiting for 1222 * ourselves. 1223 */ 1224 WITH_JOB_LOCK_GUARD() { 1225 if (!s->common.job.paused && !job_is_cancelled_locked(&job->job) 1226 && !s->in_drain) { 1227 return true; 1228 } 1229 } 1230 1231 return !!s->in_flight; 1232 } 1233 1234 static bool mirror_cancel(Job *job, bool force) 1235 { 1236 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); 1237 BlockDriverState *target = blk_bs(s->target); 1238 1239 /* 1240 * Before the job is READY, we treat any cancellation like a 1241 * force-cancellation. 1242 */ 1243 force = force || !job_is_ready(job); 1244 1245 if (force) { 1246 bdrv_cancel_in_flight(target); 1247 } 1248 return force; 1249 } 1250 1251 static bool commit_active_cancel(Job *job, bool force) 1252 { 1253 /* Same as above in mirror_cancel() */ 1254 return force || !job_is_ready(job); 1255 } 1256 1257 static void mirror_change(BlockJob *job, BlockJobChangeOptions *opts, 1258 Error **errp) 1259 { 1260 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common); 1261 BlockJobChangeOptionsMirror *change_opts = &opts->u.mirror; 1262 MirrorCopyMode current; 1263 1264 /* 1265 * The implementation relies on the fact that copy_mode is only written 1266 * under the BQL. Otherwise, further synchronization would be required. 1267 */ 1268 1269 GLOBAL_STATE_CODE(); 1270 1271 if (qatomic_read(&s->copy_mode) == change_opts->copy_mode) { 1272 return; 1273 } 1274 1275 if (change_opts->copy_mode != MIRROR_COPY_MODE_WRITE_BLOCKING) { 1276 error_setg(errp, "Change to copy mode '%s' is not implemented", 1277 MirrorCopyMode_str(change_opts->copy_mode)); 1278 return; 1279 } 1280 1281 current = qatomic_cmpxchg(&s->copy_mode, MIRROR_COPY_MODE_BACKGROUND, 1282 change_opts->copy_mode); 1283 if (current != MIRROR_COPY_MODE_BACKGROUND) { 1284 error_setg(errp, "Expected current copy mode '%s', got '%s'", 1285 MirrorCopyMode_str(MIRROR_COPY_MODE_BACKGROUND), 1286 MirrorCopyMode_str(current)); 1287 } 1288 } 1289 1290 static void mirror_query(BlockJob *job, BlockJobInfo *info) 1291 { 1292 MirrorBlockJob *s = container_of(job, MirrorBlockJob, common); 1293 1294 info->u.mirror = (BlockJobInfoMirror) { 1295 .actively_synced = qatomic_read(&s->actively_synced), 1296 }; 1297 } 1298 1299 static const BlockJobDriver mirror_job_driver = { 1300 .job_driver = { 1301 .instance_size = sizeof(MirrorBlockJob), 1302 .job_type = JOB_TYPE_MIRROR, 1303 .free = block_job_free, 1304 .user_resume = block_job_user_resume, 1305 .run = mirror_run, 1306 .prepare = mirror_prepare, 1307 .abort = mirror_abort, 1308 .pause = mirror_pause, 1309 .complete = mirror_complete, 1310 .cancel = mirror_cancel, 1311 }, 1312 .drained_poll = mirror_drained_poll, 1313 .change = mirror_change, 1314 .query = mirror_query, 1315 }; 1316 1317 static const BlockJobDriver commit_active_job_driver = { 1318 .job_driver = { 1319 .instance_size = sizeof(MirrorBlockJob), 1320 .job_type = JOB_TYPE_COMMIT, 1321 .free = block_job_free, 1322 .user_resume = block_job_user_resume, 1323 .run = mirror_run, 1324 .prepare = mirror_prepare, 1325 .abort = mirror_abort, 1326 .pause = mirror_pause, 1327 .complete = mirror_complete, 1328 .cancel = commit_active_cancel, 1329 }, 1330 .drained_poll = mirror_drained_poll, 1331 }; 1332 1333 static void coroutine_fn 1334 do_sync_target_write(MirrorBlockJob *job, MirrorMethod method, 1335 uint64_t offset, uint64_t bytes, 1336 QEMUIOVector *qiov, int flags) 1337 { 1338 int ret; 1339 size_t qiov_offset = 0; 1340 int64_t bitmap_offset, bitmap_end; 1341 1342 if (!QEMU_IS_ALIGNED(offset, job->granularity) && 1343 bdrv_dirty_bitmap_get(job->dirty_bitmap, offset)) 1344 { 1345 /* 1346 * Dirty unaligned padding: ignore it. 1347 * 1348 * Reasoning: 1349 * 1. If we copy it, we can't reset corresponding bit in 1350 * dirty_bitmap as there may be some "dirty" bytes still not 1351 * copied. 1352 * 2. It's already dirty, so skipping it we don't diverge mirror 1353 * progress. 1354 * 1355 * Note, that because of this, guest write may have no contribution 1356 * into mirror converge, but that's not bad, as we have background 1357 * process of mirroring. If under some bad circumstances (high guest 1358 * IO load) background process starve, we will not converge anyway, 1359 * even if each write will contribute, as guest is not guaranteed to 1360 * rewrite the whole disk. 1361 */ 1362 qiov_offset = QEMU_ALIGN_UP(offset, job->granularity) - offset; 1363 if (bytes <= qiov_offset) { 1364 /* nothing to do after shrink */ 1365 return; 1366 } 1367 offset += qiov_offset; 1368 bytes -= qiov_offset; 1369 } 1370 1371 if (!QEMU_IS_ALIGNED(offset + bytes, job->granularity) && 1372 bdrv_dirty_bitmap_get(job->dirty_bitmap, offset + bytes - 1)) 1373 { 1374 uint64_t tail = (offset + bytes) % job->granularity; 1375 1376 if (bytes <= tail) { 1377 /* nothing to do after shrink */ 1378 return; 1379 } 1380 bytes -= tail; 1381 } 1382 1383 /* 1384 * Tails are either clean or shrunk, so for bitmap resetting 1385 * we safely align the range down. 1386 */ 1387 bitmap_offset = QEMU_ALIGN_UP(offset, job->granularity); 1388 bitmap_end = QEMU_ALIGN_DOWN(offset + bytes, job->granularity); 1389 if (bitmap_offset < bitmap_end) { 1390 bdrv_reset_dirty_bitmap(job->dirty_bitmap, bitmap_offset, 1391 bitmap_end - bitmap_offset); 1392 } 1393 1394 job_progress_increase_remaining(&job->common.job, bytes); 1395 job->active_write_bytes_in_flight += bytes; 1396 1397 switch (method) { 1398 case MIRROR_METHOD_COPY: 1399 ret = blk_co_pwritev_part(job->target, offset, bytes, 1400 qiov, qiov_offset, flags); 1401 break; 1402 1403 case MIRROR_METHOD_ZERO: 1404 assert(!qiov); 1405 ret = blk_co_pwrite_zeroes(job->target, offset, bytes, flags); 1406 break; 1407 1408 case MIRROR_METHOD_DISCARD: 1409 assert(!qiov); 1410 ret = blk_co_pdiscard(job->target, offset, bytes); 1411 break; 1412 1413 default: 1414 abort(); 1415 } 1416 1417 job->active_write_bytes_in_flight -= bytes; 1418 if (ret >= 0) { 1419 job_progress_update(&job->common.job, bytes); 1420 } else { 1421 BlockErrorAction action; 1422 1423 /* 1424 * We failed, so we should mark dirty the whole area, aligned up. 1425 * Note that we don't care about shrunk tails if any: they were dirty 1426 * at function start, and they must be still dirty, as we've locked 1427 * the region for in-flight op. 1428 */ 1429 bitmap_offset = QEMU_ALIGN_DOWN(offset, job->granularity); 1430 bitmap_end = QEMU_ALIGN_UP(offset + bytes, job->granularity); 1431 bdrv_set_dirty_bitmap(job->dirty_bitmap, bitmap_offset, 1432 bitmap_end - bitmap_offset); 1433 qatomic_set(&job->actively_synced, false); 1434 1435 action = mirror_error_action(job, false, -ret); 1436 if (action == BLOCK_ERROR_ACTION_REPORT) { 1437 if (!job->ret) { 1438 job->ret = ret; 1439 } 1440 } 1441 } 1442 } 1443 1444 static MirrorOp *coroutine_fn active_write_prepare(MirrorBlockJob *s, 1445 uint64_t offset, 1446 uint64_t bytes) 1447 { 1448 MirrorOp *op; 1449 uint64_t start_chunk = offset / s->granularity; 1450 uint64_t end_chunk = DIV_ROUND_UP(offset + bytes, s->granularity); 1451 1452 op = g_new(MirrorOp, 1); 1453 *op = (MirrorOp){ 1454 .s = s, 1455 .offset = offset, 1456 .bytes = bytes, 1457 .is_active_write = true, 1458 .is_in_flight = true, 1459 .co = qemu_coroutine_self(), 1460 }; 1461 qemu_co_queue_init(&op->waiting_requests); 1462 QTAILQ_INSERT_TAIL(&s->ops_in_flight, op, next); 1463 1464 s->in_active_write_counter++; 1465 1466 /* 1467 * Wait for concurrent requests affecting the area. If there are already 1468 * running requests that are copying off now-to-be stale data in the area, 1469 * we must wait for them to finish before we begin writing fresh data to the 1470 * target so that the write operations appear in the correct order. 1471 * Note that background requests (see mirror_iteration()) in contrast only 1472 * wait for conflicting requests at the start of the dirty area, and then 1473 * (based on the in_flight_bitmap) truncate the area to copy so it will not 1474 * conflict with any requests beyond that. For active writes, however, we 1475 * cannot truncate that area. The request from our parent must be blocked 1476 * until the area is copied in full. Therefore, we must wait for the whole 1477 * area to become free of concurrent requests. 1478 */ 1479 mirror_wait_on_conflicts(op, s, offset, bytes); 1480 1481 bitmap_set(s->in_flight_bitmap, start_chunk, end_chunk - start_chunk); 1482 1483 return op; 1484 } 1485 1486 static void coroutine_fn GRAPH_RDLOCK active_write_settle(MirrorOp *op) 1487 { 1488 uint64_t start_chunk = op->offset / op->s->granularity; 1489 uint64_t end_chunk = DIV_ROUND_UP(op->offset + op->bytes, 1490 op->s->granularity); 1491 1492 if (!--op->s->in_active_write_counter && 1493 qatomic_read(&op->s->actively_synced)) { 1494 BdrvChild *source = op->s->mirror_top_bs->backing; 1495 1496 if (QLIST_FIRST(&source->bs->parents) == source && 1497 QLIST_NEXT(source, next_parent) == NULL) 1498 { 1499 /* Assert that we are back in sync once all active write 1500 * operations are settled. 1501 * Note that we can only assert this if the mirror node 1502 * is the source node's only parent. */ 1503 assert(!bdrv_get_dirty_count(op->s->dirty_bitmap)); 1504 } 1505 } 1506 bitmap_clear(op->s->in_flight_bitmap, start_chunk, end_chunk - start_chunk); 1507 QTAILQ_REMOVE(&op->s->ops_in_flight, op, next); 1508 qemu_co_queue_restart_all(&op->waiting_requests); 1509 g_free(op); 1510 } 1511 1512 static int coroutine_fn GRAPH_RDLOCK 1513 bdrv_mirror_top_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, 1514 QEMUIOVector *qiov, BdrvRequestFlags flags) 1515 { 1516 return bdrv_co_preadv(bs->backing, offset, bytes, qiov, flags); 1517 } 1518 1519 static bool should_copy_to_target(MirrorBDSOpaque *s) 1520 { 1521 return s->job && s->job->ret >= 0 && 1522 !job_is_cancelled(&s->job->common.job) && 1523 qatomic_read(&s->job->copy_mode) == MIRROR_COPY_MODE_WRITE_BLOCKING; 1524 } 1525 1526 static int coroutine_fn GRAPH_RDLOCK 1527 bdrv_mirror_top_do_write(BlockDriverState *bs, MirrorMethod method, 1528 bool copy_to_target, uint64_t offset, uint64_t bytes, 1529 QEMUIOVector *qiov, int flags) 1530 { 1531 MirrorOp *op = NULL; 1532 MirrorBDSOpaque *s = bs->opaque; 1533 int ret = 0; 1534 1535 if (copy_to_target) { 1536 op = active_write_prepare(s->job, offset, bytes); 1537 } 1538 1539 switch (method) { 1540 case MIRROR_METHOD_COPY: 1541 ret = bdrv_co_pwritev(bs->backing, offset, bytes, qiov, flags); 1542 break; 1543 1544 case MIRROR_METHOD_ZERO: 1545 ret = bdrv_co_pwrite_zeroes(bs->backing, offset, bytes, flags); 1546 break; 1547 1548 case MIRROR_METHOD_DISCARD: 1549 ret = bdrv_co_pdiscard(bs->backing, offset, bytes); 1550 break; 1551 1552 default: 1553 abort(); 1554 } 1555 1556 if (!copy_to_target && s->job && s->job->dirty_bitmap) { 1557 qatomic_set(&s->job->actively_synced, false); 1558 bdrv_set_dirty_bitmap(s->job->dirty_bitmap, offset, bytes); 1559 } 1560 1561 if (ret < 0) { 1562 goto out; 1563 } 1564 1565 if (copy_to_target) { 1566 do_sync_target_write(s->job, method, offset, bytes, qiov, flags); 1567 } 1568 1569 out: 1570 if (copy_to_target) { 1571 active_write_settle(op); 1572 } 1573 return ret; 1574 } 1575 1576 static int coroutine_fn GRAPH_RDLOCK 1577 bdrv_mirror_top_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, 1578 QEMUIOVector *qiov, BdrvRequestFlags flags) 1579 { 1580 QEMUIOVector bounce_qiov; 1581 void *bounce_buf; 1582 int ret = 0; 1583 bool copy_to_target = should_copy_to_target(bs->opaque); 1584 1585 if (copy_to_target) { 1586 /* The guest might concurrently modify the data to write; but 1587 * the data on source and destination must match, so we have 1588 * to use a bounce buffer if we are going to write to the 1589 * target now. */ 1590 bounce_buf = qemu_blockalign(bs, bytes); 1591 iov_to_buf_full(qiov->iov, qiov->niov, 0, bounce_buf, bytes); 1592 1593 qemu_iovec_init(&bounce_qiov, 1); 1594 qemu_iovec_add(&bounce_qiov, bounce_buf, bytes); 1595 qiov = &bounce_qiov; 1596 1597 flags &= ~BDRV_REQ_REGISTERED_BUF; 1598 } 1599 1600 ret = bdrv_mirror_top_do_write(bs, MIRROR_METHOD_COPY, copy_to_target, 1601 offset, bytes, qiov, flags); 1602 1603 if (copy_to_target) { 1604 qemu_iovec_destroy(&bounce_qiov); 1605 qemu_vfree(bounce_buf); 1606 } 1607 1608 return ret; 1609 } 1610 1611 static int coroutine_fn GRAPH_RDLOCK bdrv_mirror_top_flush(BlockDriverState *bs) 1612 { 1613 if (bs->backing == NULL) { 1614 /* we can be here after failed bdrv_append in mirror_start_job */ 1615 return 0; 1616 } 1617 return bdrv_co_flush(bs->backing->bs); 1618 } 1619 1620 static int coroutine_fn GRAPH_RDLOCK 1621 bdrv_mirror_top_pwrite_zeroes(BlockDriverState *bs, int64_t offset, 1622 int64_t bytes, BdrvRequestFlags flags) 1623 { 1624 bool copy_to_target = should_copy_to_target(bs->opaque); 1625 return bdrv_mirror_top_do_write(bs, MIRROR_METHOD_ZERO, copy_to_target, 1626 offset, bytes, NULL, flags); 1627 } 1628 1629 static int coroutine_fn GRAPH_RDLOCK 1630 bdrv_mirror_top_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) 1631 { 1632 bool copy_to_target = should_copy_to_target(bs->opaque); 1633 return bdrv_mirror_top_do_write(bs, MIRROR_METHOD_DISCARD, copy_to_target, 1634 offset, bytes, NULL, 0); 1635 } 1636 1637 static void bdrv_mirror_top_refresh_filename(BlockDriverState *bs) 1638 { 1639 if (bs->backing == NULL) { 1640 /* we can be here after failed bdrv_attach_child in 1641 * bdrv_set_backing_hd */ 1642 return; 1643 } 1644 pstrcpy(bs->exact_filename, sizeof(bs->exact_filename), 1645 bs->backing->bs->filename); 1646 } 1647 1648 static void bdrv_mirror_top_child_perm(BlockDriverState *bs, BdrvChild *c, 1649 BdrvChildRole role, 1650 BlockReopenQueue *reopen_queue, 1651 uint64_t perm, uint64_t shared, 1652 uint64_t *nperm, uint64_t *nshared) 1653 { 1654 MirrorBDSOpaque *s = bs->opaque; 1655 1656 if (s->stop) { 1657 /* 1658 * If the job is to be stopped, we do not need to forward 1659 * anything to the real image. 1660 */ 1661 *nperm = 0; 1662 *nshared = BLK_PERM_ALL; 1663 return; 1664 } 1665 1666 bdrv_default_perms(bs, c, role, reopen_queue, 1667 perm, shared, nperm, nshared); 1668 1669 if (s->is_commit) { 1670 /* 1671 * For commit jobs, we cannot take CONSISTENT_READ, because 1672 * that permission is unshared for everything above the base 1673 * node (except for filters on the base node). 1674 * We also have to force-share the WRITE permission, or 1675 * otherwise we would block ourselves at the base node (if 1676 * writes are blocked for a node, they are also blocked for 1677 * its backing file). 1678 * (We could also share RESIZE, because it may be needed for 1679 * the target if its size is less than the top node's; but 1680 * bdrv_default_perms_for_cow() automatically shares RESIZE 1681 * for backing nodes if WRITE is shared, so there is no need 1682 * to do it here.) 1683 */ 1684 *nperm &= ~BLK_PERM_CONSISTENT_READ; 1685 *nshared |= BLK_PERM_WRITE; 1686 } 1687 } 1688 1689 /* Dummy node that provides consistent read to its users without requiring it 1690 * from its backing file and that allows writes on the backing file chain. */ 1691 static BlockDriver bdrv_mirror_top = { 1692 .format_name = "mirror_top", 1693 .bdrv_co_preadv = bdrv_mirror_top_preadv, 1694 .bdrv_co_pwritev = bdrv_mirror_top_pwritev, 1695 .bdrv_co_pwrite_zeroes = bdrv_mirror_top_pwrite_zeroes, 1696 .bdrv_co_pdiscard = bdrv_mirror_top_pdiscard, 1697 .bdrv_co_flush = bdrv_mirror_top_flush, 1698 .bdrv_refresh_filename = bdrv_mirror_top_refresh_filename, 1699 .bdrv_child_perm = bdrv_mirror_top_child_perm, 1700 1701 .is_filter = true, 1702 .filtered_child_is_backing = true, 1703 }; 1704 1705 static BlockJob *mirror_start_job( 1706 const char *job_id, BlockDriverState *bs, 1707 int creation_flags, BlockDriverState *target, 1708 const char *replaces, int64_t speed, 1709 uint32_t granularity, int64_t buf_size, 1710 BlockMirrorBackingMode backing_mode, 1711 bool zero_target, 1712 BlockdevOnError on_source_error, 1713 BlockdevOnError on_target_error, 1714 bool unmap, 1715 BlockCompletionFunc *cb, 1716 void *opaque, 1717 const BlockJobDriver *driver, 1718 bool is_none_mode, BlockDriverState *base, 1719 bool auto_complete, const char *filter_node_name, 1720 bool is_mirror, MirrorCopyMode copy_mode, 1721 Error **errp) 1722 { 1723 MirrorBlockJob *s; 1724 MirrorBDSOpaque *bs_opaque; 1725 BlockDriverState *mirror_top_bs; 1726 bool target_is_backing; 1727 uint64_t target_perms, target_shared_perms; 1728 int ret; 1729 1730 GLOBAL_STATE_CODE(); 1731 1732 if (granularity == 0) { 1733 granularity = bdrv_get_default_bitmap_granularity(target); 1734 } 1735 1736 assert(is_power_of_2(granularity)); 1737 1738 if (buf_size < 0) { 1739 error_setg(errp, "Invalid parameter 'buf-size'"); 1740 return NULL; 1741 } 1742 1743 if (buf_size == 0) { 1744 buf_size = DEFAULT_MIRROR_BUF_SIZE; 1745 } 1746 1747 if (bdrv_skip_filters(bs) == bdrv_skip_filters(target)) { 1748 error_setg(errp, "Can't mirror node into itself"); 1749 return NULL; 1750 } 1751 1752 target_is_backing = bdrv_chain_contains(bs, target); 1753 1754 /* In the case of active commit, add dummy driver to provide consistent 1755 * reads on the top, while disabling it in the intermediate nodes, and make 1756 * the backing chain writable. */ 1757 mirror_top_bs = bdrv_new_open_driver(&bdrv_mirror_top, filter_node_name, 1758 BDRV_O_RDWR, errp); 1759 if (mirror_top_bs == NULL) { 1760 return NULL; 1761 } 1762 if (!filter_node_name) { 1763 mirror_top_bs->implicit = true; 1764 } 1765 1766 /* So that we can always drop this node */ 1767 mirror_top_bs->never_freeze = true; 1768 1769 mirror_top_bs->total_sectors = bs->total_sectors; 1770 mirror_top_bs->supported_write_flags = BDRV_REQ_WRITE_UNCHANGED; 1771 mirror_top_bs->supported_zero_flags = BDRV_REQ_WRITE_UNCHANGED | 1772 BDRV_REQ_NO_FALLBACK; 1773 bs_opaque = g_new0(MirrorBDSOpaque, 1); 1774 mirror_top_bs->opaque = bs_opaque; 1775 1776 bs_opaque->is_commit = target_is_backing; 1777 1778 bdrv_drained_begin(bs); 1779 ret = bdrv_append(mirror_top_bs, bs, errp); 1780 bdrv_drained_end(bs); 1781 1782 if (ret < 0) { 1783 bdrv_unref(mirror_top_bs); 1784 return NULL; 1785 } 1786 1787 /* Make sure that the source is not resized while the job is running */ 1788 s = block_job_create(job_id, driver, NULL, mirror_top_bs, 1789 BLK_PERM_CONSISTENT_READ, 1790 BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED | 1791 BLK_PERM_WRITE, speed, 1792 creation_flags, cb, opaque, errp); 1793 if (!s) { 1794 goto fail; 1795 } 1796 1797 /* The block job now has a reference to this node */ 1798 bdrv_unref(mirror_top_bs); 1799 1800 s->mirror_top_bs = mirror_top_bs; 1801 1802 /* No resize for the target either; while the mirror is still running, a 1803 * consistent read isn't necessarily possible. We could possibly allow 1804 * writes and graph modifications, though it would likely defeat the 1805 * purpose of a mirror, so leave them blocked for now. 1806 * 1807 * In the case of active commit, things look a bit different, though, 1808 * because the target is an already populated backing file in active use. 1809 * We can allow anything except resize there.*/ 1810 1811 target_perms = BLK_PERM_WRITE; 1812 target_shared_perms = BLK_PERM_WRITE_UNCHANGED; 1813 1814 if (target_is_backing) { 1815 int64_t bs_size, target_size; 1816 bs_size = bdrv_getlength(bs); 1817 if (bs_size < 0) { 1818 error_setg_errno(errp, -bs_size, 1819 "Could not inquire top image size"); 1820 goto fail; 1821 } 1822 1823 target_size = bdrv_getlength(target); 1824 if (target_size < 0) { 1825 error_setg_errno(errp, -target_size, 1826 "Could not inquire base image size"); 1827 goto fail; 1828 } 1829 1830 if (target_size < bs_size) { 1831 target_perms |= BLK_PERM_RESIZE; 1832 } 1833 1834 target_shared_perms |= BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE; 1835 } else if (bdrv_chain_contains(bs, bdrv_skip_filters(target))) { 1836 /* 1837 * We may want to allow this in the future, but it would 1838 * require taking some extra care. 1839 */ 1840 error_setg(errp, "Cannot mirror to a filter on top of a node in the " 1841 "source's backing chain"); 1842 goto fail; 1843 } 1844 1845 s->target = blk_new(s->common.job.aio_context, 1846 target_perms, target_shared_perms); 1847 ret = blk_insert_bs(s->target, target, errp); 1848 if (ret < 0) { 1849 goto fail; 1850 } 1851 if (is_mirror) { 1852 /* XXX: Mirror target could be a NBD server of target QEMU in the case 1853 * of non-shared block migration. To allow migration completion, we 1854 * have to allow "inactivate" of the target BB. When that happens, we 1855 * know the job is drained, and the vcpus are stopped, so no write 1856 * operation will be performed. Block layer already has assertions to 1857 * ensure that. */ 1858 blk_set_force_allow_inactivate(s->target); 1859 } 1860 blk_set_allow_aio_context_change(s->target, true); 1861 blk_set_disable_request_queuing(s->target, true); 1862 1863 s->replaces = g_strdup(replaces); 1864 s->on_source_error = on_source_error; 1865 s->on_target_error = on_target_error; 1866 s->is_none_mode = is_none_mode; 1867 s->backing_mode = backing_mode; 1868 s->zero_target = zero_target; 1869 qatomic_set(&s->copy_mode, copy_mode); 1870 s->base = base; 1871 s->base_overlay = bdrv_find_overlay(bs, base); 1872 s->granularity = granularity; 1873 s->buf_size = ROUND_UP(buf_size, granularity); 1874 s->unmap = unmap; 1875 if (auto_complete) { 1876 s->should_complete = true; 1877 } 1878 1879 s->dirty_bitmap = bdrv_create_dirty_bitmap(s->mirror_top_bs, granularity, 1880 NULL, errp); 1881 if (!s->dirty_bitmap) { 1882 goto fail; 1883 } 1884 1885 /* 1886 * The dirty bitmap is set by bdrv_mirror_top_do_write() when not in active 1887 * mode. 1888 */ 1889 bdrv_disable_dirty_bitmap(s->dirty_bitmap); 1890 1891 bdrv_graph_wrlock(bs); 1892 ret = block_job_add_bdrv(&s->common, "source", bs, 0, 1893 BLK_PERM_WRITE_UNCHANGED | BLK_PERM_WRITE | 1894 BLK_PERM_CONSISTENT_READ, 1895 errp); 1896 if (ret < 0) { 1897 bdrv_graph_wrunlock(); 1898 goto fail; 1899 } 1900 1901 /* Required permissions are already taken with blk_new() */ 1902 block_job_add_bdrv(&s->common, "target", target, 0, BLK_PERM_ALL, 1903 &error_abort); 1904 1905 /* In commit_active_start() all intermediate nodes disappear, so 1906 * any jobs in them must be blocked */ 1907 if (target_is_backing) { 1908 BlockDriverState *iter, *filtered_target; 1909 uint64_t iter_shared_perms; 1910 1911 /* 1912 * The topmost node with 1913 * bdrv_skip_filters(filtered_target) == bdrv_skip_filters(target) 1914 */ 1915 filtered_target = bdrv_cow_bs(bdrv_find_overlay(bs, target)); 1916 1917 assert(bdrv_skip_filters(filtered_target) == 1918 bdrv_skip_filters(target)); 1919 1920 /* 1921 * XXX BLK_PERM_WRITE needs to be allowed so we don't block 1922 * ourselves at s->base (if writes are blocked for a node, they are 1923 * also blocked for its backing file). The other options would be a 1924 * second filter driver above s->base (== target). 1925 */ 1926 iter_shared_perms = BLK_PERM_WRITE_UNCHANGED | BLK_PERM_WRITE; 1927 1928 for (iter = bdrv_filter_or_cow_bs(bs); iter != target; 1929 iter = bdrv_filter_or_cow_bs(iter)) 1930 { 1931 if (iter == filtered_target) { 1932 /* 1933 * From here on, all nodes are filters on the base. 1934 * This allows us to share BLK_PERM_CONSISTENT_READ. 1935 */ 1936 iter_shared_perms |= BLK_PERM_CONSISTENT_READ; 1937 } 1938 1939 ret = block_job_add_bdrv(&s->common, "intermediate node", iter, 0, 1940 iter_shared_perms, errp); 1941 if (ret < 0) { 1942 bdrv_graph_wrunlock(); 1943 goto fail; 1944 } 1945 } 1946 1947 if (bdrv_freeze_backing_chain(mirror_top_bs, target, errp) < 0) { 1948 bdrv_graph_wrunlock(); 1949 goto fail; 1950 } 1951 } 1952 bdrv_graph_wrunlock(); 1953 1954 QTAILQ_INIT(&s->ops_in_flight); 1955 1956 trace_mirror_start(bs, s, opaque); 1957 job_start(&s->common.job); 1958 1959 return &s->common; 1960 1961 fail: 1962 if (s) { 1963 /* Make sure this BDS does not go away until we have completed the graph 1964 * changes below */ 1965 bdrv_ref(mirror_top_bs); 1966 1967 g_free(s->replaces); 1968 blk_unref(s->target); 1969 bs_opaque->job = NULL; 1970 if (s->dirty_bitmap) { 1971 bdrv_release_dirty_bitmap(s->dirty_bitmap); 1972 } 1973 job_early_fail(&s->common.job); 1974 } 1975 1976 bs_opaque->stop = true; 1977 bdrv_graph_rdlock_main_loop(); 1978 bdrv_child_refresh_perms(mirror_top_bs, mirror_top_bs->backing, 1979 &error_abort); 1980 bdrv_graph_rdunlock_main_loop(); 1981 bdrv_replace_node(mirror_top_bs, mirror_top_bs->backing->bs, &error_abort); 1982 1983 bdrv_unref(mirror_top_bs); 1984 1985 return NULL; 1986 } 1987 1988 void mirror_start(const char *job_id, BlockDriverState *bs, 1989 BlockDriverState *target, const char *replaces, 1990 int creation_flags, int64_t speed, 1991 uint32_t granularity, int64_t buf_size, 1992 MirrorSyncMode mode, BlockMirrorBackingMode backing_mode, 1993 bool zero_target, 1994 BlockdevOnError on_source_error, 1995 BlockdevOnError on_target_error, 1996 bool unmap, const char *filter_node_name, 1997 MirrorCopyMode copy_mode, Error **errp) 1998 { 1999 bool is_none_mode; 2000 BlockDriverState *base; 2001 2002 GLOBAL_STATE_CODE(); 2003 2004 if ((mode == MIRROR_SYNC_MODE_INCREMENTAL) || 2005 (mode == MIRROR_SYNC_MODE_BITMAP)) { 2006 error_setg(errp, "Sync mode '%s' not supported", 2007 MirrorSyncMode_str(mode)); 2008 return; 2009 } 2010 is_none_mode = mode == MIRROR_SYNC_MODE_NONE; 2011 base = mode == MIRROR_SYNC_MODE_TOP ? bdrv_backing_chain_next(bs) : NULL; 2012 mirror_start_job(job_id, bs, creation_flags, target, replaces, 2013 speed, granularity, buf_size, backing_mode, zero_target, 2014 on_source_error, on_target_error, unmap, NULL, NULL, 2015 &mirror_job_driver, is_none_mode, base, false, 2016 filter_node_name, true, copy_mode, errp); 2017 } 2018 2019 BlockJob *commit_active_start(const char *job_id, BlockDriverState *bs, 2020 BlockDriverState *base, int creation_flags, 2021 int64_t speed, BlockdevOnError on_error, 2022 const char *filter_node_name, 2023 BlockCompletionFunc *cb, void *opaque, 2024 bool auto_complete, Error **errp) 2025 { 2026 bool base_read_only; 2027 BlockJob *job; 2028 2029 GLOBAL_STATE_CODE(); 2030 2031 base_read_only = bdrv_is_read_only(base); 2032 2033 if (base_read_only) { 2034 if (bdrv_reopen_set_read_only(base, false, errp) < 0) { 2035 return NULL; 2036 } 2037 } 2038 2039 job = mirror_start_job( 2040 job_id, bs, creation_flags, base, NULL, speed, 0, 0, 2041 MIRROR_LEAVE_BACKING_CHAIN, false, 2042 on_error, on_error, true, cb, opaque, 2043 &commit_active_job_driver, false, base, auto_complete, 2044 filter_node_name, false, MIRROR_COPY_MODE_BACKGROUND, 2045 errp); 2046 if (!job) { 2047 goto error_restore_flags; 2048 } 2049 2050 return job; 2051 2052 error_restore_flags: 2053 /* ignore error and errp for bdrv_reopen, because we want to propagate 2054 * the original error */ 2055 if (base_read_only) { 2056 bdrv_reopen_set_read_only(base, true, NULL); 2057 } 2058 return NULL; 2059 } 2060