--- zzzz-none-000/linux-3.10.107/drivers/md/dm-thin.c 2017-06-27 09:49:32.000000000 +0000 +++ scorpion-7490-727/linux-3.10.107/drivers/md/dm-thin.c 2021-02-04 17:41:59.000000000 +0000 @@ -11,10 +11,16 @@ #include #include #include +#include +#include #include +#include #include #include #include +#include +#include +#include #define DM_MSG_PREFIX "thin" @@ -23,8 +29,10 @@ */ #define ENDIO_HOOK_POOL_SIZE 1024 #define MAPPING_POOL_SIZE 1024 -#define PRISON_CELLS 1024 #define COMMIT_PERIOD HZ +#define NO_SPACE_TIMEOUT_SECS 60 + +static unsigned no_space_timeout_secs = NO_SPACE_TIMEOUT_SECS; DECLARE_DM_KCOPYD_THROTTLE_WITH_MODULE_PARM(snapshot_copy_throttle, "A percentage of time allocated for copy on write"); @@ -104,20 +112,77 @@ /* * Key building. */ -static void build_data_key(struct dm_thin_device *td, - dm_block_t b, struct dm_cell_key *key) +enum lock_space { + VIRTUAL, + PHYSICAL +}; + +static void build_key(struct dm_thin_device *td, enum lock_space ls, + dm_block_t b, dm_block_t e, struct dm_cell_key *key) { - key->virtual = 0; + key->virtual = (ls == VIRTUAL); key->dev = dm_thin_dev_id(td); - key->block = b; + key->block_begin = b; + key->block_end = e; +} + +static void build_data_key(struct dm_thin_device *td, dm_block_t b, + struct dm_cell_key *key) +{ + build_key(td, PHYSICAL, b, b + 1llu, key); } static void build_virtual_key(struct dm_thin_device *td, dm_block_t b, struct dm_cell_key *key) { - key->virtual = 1; - key->dev = dm_thin_dev_id(td); - key->block = b; + build_key(td, VIRTUAL, b, b + 1llu, key); +} + +/*----------------------------------------------------------------*/ + +#define THROTTLE_THRESHOLD (1 * HZ) + +struct throttle { + struct rw_semaphore lock; + unsigned long threshold; + bool throttle_applied; +}; + +static void throttle_init(struct throttle *t) +{ + init_rwsem(&t->lock); + t->throttle_applied = false; +} + +static void throttle_work_start(struct throttle *t) +{ + t->threshold = jiffies + THROTTLE_THRESHOLD; +} + +static void throttle_work_update(struct throttle *t) +{ + if (!t->throttle_applied && jiffies > t->threshold) { + down_write(&t->lock); + t->throttle_applied = true; + } +} + +static void throttle_work_complete(struct throttle *t) +{ + if (t->throttle_applied) { + t->throttle_applied = false; + up_write(&t->lock); + } +} + +static void throttle_lock(struct throttle *t) +{ + down_read(&t->lock); +} + +static void throttle_unlock(struct throttle *t) +{ + up_read(&t->lock); } /*----------------------------------------------------------------*/ @@ -130,10 +195,11 @@ struct dm_thin_new_mapping; /* - * The pool runs in 3 modes. Ordered in degraded order for comparisons. + * The pool runs in 4 modes. Ordered in degraded order for comparisons. */ enum pool_mode { PM_WRITE, /* metadata may be changed */ + PM_OUT_OF_DATA_SPACE, /* metadata may be changed, though data may not be allocated */ PM_READ_ONLY, /* metadata may not be changed */ PM_FAIL, /* all I/O fails */ }; @@ -144,12 +210,16 @@ bool zero_new_blocks:1; bool discard_enabled:1; bool discard_passdown:1; + bool error_if_no_space:1; }; struct thin_c; typedef void (*process_bio_fn)(struct thin_c *tc, struct bio *bio); +typedef void (*process_cell_fn)(struct thin_c *tc, struct dm_bio_prison_cell *cell); typedef void (*process_mapping_fn)(struct dm_thin_new_mapping *m); +#define CELL_SORT_ARRAY_SIZE 8192 + struct pool { struct list_head list; struct dm_target *ti; /* Only set if a pool target is bound */ @@ -163,26 +233,26 @@ int sectors_per_block_shift; struct pool_features pf; - unsigned low_water_triggered:1; /* A dm event has been sent */ - unsigned no_free_space:1; /* A -ENOSPC warning has been issued */ + bool low_water_triggered:1; /* A dm event has been sent */ + bool suspended:1; struct dm_bio_prison *prison; struct dm_kcopyd_client *copier; struct workqueue_struct *wq; + struct throttle throttle; struct work_struct worker; struct delayed_work waker; + struct delayed_work no_space_timeout; unsigned long last_commit_jiffies; unsigned ref_count; spinlock_t lock; - struct bio_list deferred_bios; struct bio_list deferred_flush_bios; struct list_head prepared_mappings; struct list_head prepared_discards; - - struct bio_list retry_on_resume_list; + struct list_head active_thins; struct dm_deferred_set *shared_read_ds; struct dm_deferred_set *all_io_ds; @@ -193,12 +263,17 @@ process_bio_fn process_bio; process_bio_fn process_discard; + process_cell_fn process_cell; + process_cell_fn process_discard_cell; + process_mapping_fn process_prepared_mapping; process_mapping_fn process_prepared_discard; + + struct dm_bio_prison_cell **cell_sort_array; }; static enum pool_mode get_pool_mode(struct pool *pool); -static void set_pool_mode(struct pool *pool, enum pool_mode mode); +static void metadata_operation_failed(struct pool *pool, const char *op, int r); /* * Target context for a pool. @@ -219,16 +294,107 @@ * Target context for a thin. */ struct thin_c { + struct list_head list; struct dm_dev *pool_dev; struct dm_dev *origin_dev; + sector_t origin_size; dm_thin_id dev_id; struct pool *pool; struct dm_thin_device *td; + struct mapped_device *thin_md; + + bool requeue_mode:1; + spinlock_t lock; + struct list_head deferred_cells; + struct bio_list deferred_bio_list; + struct bio_list retry_on_resume_list; + struct rb_root sort_bio_list; /* sorted list of deferred bios */ + + /* + * Ensures the thin is not destroyed until the worker has finished + * iterating the active_thins list. + */ + atomic_t refcount; + struct completion can_destroy; }; /*----------------------------------------------------------------*/ +/** + * __blkdev_issue_discard_async - queue a discard with async completion + * @bdev: blockdev to issue discard for + * @sector: start sector + * @nr_sects: number of sectors to discard + * @gfp_mask: memory allocation flags (for bio_alloc) + * @flags: BLKDEV_IFL_* flags to control behaviour + * @parent_bio: parent discard bio that all sub discards get chained to + * + * Description: + * Asynchronously issue a discard request for the sectors in question. + */ +static int __blkdev_issue_discard_async(struct block_device *bdev, sector_t sector, + sector_t nr_sects, gfp_t gfp_mask, unsigned long flags, + struct bio *parent_bio) +{ + struct request_queue *q = bdev_get_queue(bdev); + int type = REQ_WRITE | REQ_DISCARD; + struct bio *bio; + + if (!q || !nr_sects) + return -ENXIO; + + if (!blk_queue_discard(q)) + return -EOPNOTSUPP; + + if (flags & BLKDEV_DISCARD_SECURE) { + if (!blk_queue_secdiscard(q)) + return -EOPNOTSUPP; + type |= REQ_SECURE; + } + + /* + * Required bio_put occurs in bio_endio thanks to bio_chain below + */ + bio = bio_alloc(gfp_mask, 1); + if (!bio) + return -ENOMEM; + + bio_chain(bio, parent_bio); + + bio->bi_iter.bi_sector = sector; + bio->bi_bdev = bdev; + bio->bi_iter.bi_size = nr_sects << 9; + + submit_bio(type, bio); + + return 0; +} + +static bool block_size_is_power_of_two(struct pool *pool) +{ + return pool->sectors_per_block_shift >= 0; +} + +static sector_t block_to_sectors(struct pool *pool, dm_block_t b) +{ + return block_size_is_power_of_two(pool) ? + (b << pool->sectors_per_block_shift) : + (b * pool->sectors_per_block); +} + +static int issue_discard(struct thin_c *tc, dm_block_t data_b, dm_block_t data_e, + struct bio *parent_bio) +{ + sector_t s = block_to_sectors(tc->pool, data_b); + sector_t len = block_to_sectors(tc->pool, data_e - data_b); + + return __blkdev_issue_discard_async(tc->pool_dev->bdev, s, len, + GFP_NOWAIT, 0, parent_bio); +} + +/*----------------------------------------------------------------*/ + /* * wake_worker() is used when new work is queued and when pool_resume is * ready to continue deferred IO processing. @@ -271,6 +437,15 @@ dm_bio_prison_free_cell(pool->prison, cell); } +static void cell_visit_release(struct pool *pool, + void (*fn)(void *, struct dm_bio_prison_cell *), + void *context, + struct dm_bio_prison_cell *cell) +{ + dm_cell_visit_release(pool->prison, fn, context, cell); + dm_bio_prison_free_cell(pool->prison, cell); +} + static void cell_release_no_holder(struct pool *pool, struct dm_bio_prison_cell *cell, struct bio_list *bios) @@ -279,24 +454,26 @@ dm_bio_prison_free_cell(pool->prison, cell); } -static void cell_defer_no_holder_no_free(struct thin_c *tc, - struct dm_bio_prison_cell *cell) +static void cell_error_with_code(struct pool *pool, + struct dm_bio_prison_cell *cell, int error_code) { - struct pool *pool = tc->pool; - unsigned long flags; + dm_cell_error(pool->prison, cell, error_code); + dm_bio_prison_free_cell(pool->prison, cell); +} - spin_lock_irqsave(&pool->lock, flags); - dm_cell_release_no_holder(pool->prison, cell, &pool->deferred_bios); - spin_unlock_irqrestore(&pool->lock, flags); +static void cell_error(struct pool *pool, struct dm_bio_prison_cell *cell) +{ + cell_error_with_code(pool, cell, -EIO); +} - wake_worker(pool); +static void cell_success(struct pool *pool, struct dm_bio_prison_cell *cell) +{ + cell_error_with_code(pool, cell, 0); } -static void cell_error(struct pool *pool, - struct dm_bio_prison_cell *cell) +static void cell_requeue(struct pool *pool, struct dm_bio_prison_cell *cell) { - dm_cell_error(pool->prison, cell); - dm_bio_prison_free_cell(pool->prison, cell); + cell_error_with_code(pool, cell, DM_ENDIO_REQUEUE); } /*----------------------------------------------------------------*/ @@ -366,36 +543,86 @@ struct dm_deferred_entry *shared_read_entry; struct dm_deferred_entry *all_io_entry; struct dm_thin_new_mapping *overwrite_mapping; + struct rb_node rb_node; + struct dm_bio_prison_cell *cell; }; -static void __requeue_bio_list(struct thin_c *tc, struct bio_list *master) +static void __merge_bio_list(struct bio_list *bios, struct bio_list *master) +{ + bio_list_merge(bios, master); + bio_list_init(master); +} + +static void error_bio_list(struct bio_list *bios, int error) { struct bio *bio; + + while ((bio = bio_list_pop(bios))) { + bio->bi_error = error; + bio_endio(bio); + } +} + +static void error_thin_bio_list(struct thin_c *tc, struct bio_list *master, int error) +{ struct bio_list bios; + unsigned long flags; bio_list_init(&bios); - bio_list_merge(&bios, master); - bio_list_init(master); - while ((bio = bio_list_pop(&bios))) { - struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); + spin_lock_irqsave(&tc->lock, flags); + __merge_bio_list(&bios, master); + spin_unlock_irqrestore(&tc->lock, flags); - if (h->tc == tc) - bio_endio(bio, DM_ENDIO_REQUEUE); - else - bio_list_add(master, bio); - } + error_bio_list(&bios, error); } -static void requeue_io(struct thin_c *tc) +static void requeue_deferred_cells(struct thin_c *tc) { struct pool *pool = tc->pool; unsigned long flags; + struct list_head cells; + struct dm_bio_prison_cell *cell, *tmp; - spin_lock_irqsave(&pool->lock, flags); - __requeue_bio_list(tc, &pool->deferred_bios); - __requeue_bio_list(tc, &pool->retry_on_resume_list); - spin_unlock_irqrestore(&pool->lock, flags); + INIT_LIST_HEAD(&cells); + + spin_lock_irqsave(&tc->lock, flags); + list_splice_init(&tc->deferred_cells, &cells); + spin_unlock_irqrestore(&tc->lock, flags); + + list_for_each_entry_safe(cell, tmp, &cells, user_list) + cell_requeue(pool, cell); +} + +static void requeue_io(struct thin_c *tc) +{ + struct bio_list bios; + unsigned long flags; + + bio_list_init(&bios); + + spin_lock_irqsave(&tc->lock, flags); + __merge_bio_list(&bios, &tc->deferred_bio_list); + __merge_bio_list(&bios, &tc->retry_on_resume_list); + spin_unlock_irqrestore(&tc->lock, flags); + + error_bio_list(&bios, DM_ENDIO_REQUEUE); + requeue_deferred_cells(tc); +} + +static void error_retry_list_with_code(struct pool *pool, int error) +{ + struct thin_c *tc; + + rcu_read_lock(); + list_for_each_entry_rcu(tc, &pool->active_thins, list) + error_thin_bio_list(tc, &tc->retry_on_resume_list, error); + rcu_read_unlock(); +} + +static void error_retry_list(struct pool *pool) +{ + return error_retry_list_with_code(pool, -EIO); } /* @@ -405,15 +632,10 @@ * target. */ -static bool block_size_is_power_of_two(struct pool *pool) -{ - return pool->sectors_per_block_shift >= 0; -} - static dm_block_t get_bio_block(struct thin_c *tc, struct bio *bio) { struct pool *pool = tc->pool; - sector_t block_nr = bio->bi_sector; + sector_t block_nr = bio->bi_iter.bi_sector; if (block_size_is_power_of_two(pool)) block_nr >>= pool->sectors_per_block_shift; @@ -423,17 +645,46 @@ return block_nr; } +/* + * Returns the _complete_ blocks that this bio covers. + */ +static void get_bio_block_range(struct thin_c *tc, struct bio *bio, + dm_block_t *begin, dm_block_t *end) +{ + struct pool *pool = tc->pool; + sector_t b = bio->bi_iter.bi_sector; + sector_t e = b + (bio->bi_iter.bi_size >> SECTOR_SHIFT); + + b += pool->sectors_per_block - 1ull; /* so we round up */ + + if (block_size_is_power_of_two(pool)) { + b >>= pool->sectors_per_block_shift; + e >>= pool->sectors_per_block_shift; + } else { + (void) sector_div(b, pool->sectors_per_block); + (void) sector_div(e, pool->sectors_per_block); + } + + if (e < b) + /* Can happen if the bio is within a single block. */ + e = b; + + *begin = b; + *end = e; +} + static void remap(struct thin_c *tc, struct bio *bio, dm_block_t block) { struct pool *pool = tc->pool; - sector_t bi_sector = bio->bi_sector; + sector_t bi_sector = bio->bi_iter.bi_sector; bio->bi_bdev = tc->pool_dev->bdev; if (block_size_is_power_of_two(pool)) - bio->bi_sector = (block << pool->sectors_per_block_shift) | - (bi_sector & (pool->sectors_per_block - 1)); + bio->bi_iter.bi_sector = + (block << pool->sectors_per_block_shift) | + (bi_sector & (pool->sectors_per_block - 1)); else - bio->bi_sector = (block * pool->sectors_per_block) + + bio->bi_iter.bi_sector = (block * pool->sectors_per_block) + sector_div(bi_sector, pool->sectors_per_block); } @@ -509,16 +760,21 @@ struct dm_thin_new_mapping { struct list_head list; - unsigned quiesced:1; - unsigned prepared:1; - unsigned pass_discard:1; - unsigned definitely_not_shared:1; + bool pass_discard:1; + bool maybe_shared:1; + /* + * Track quiescing, copying and zeroing preparation actions. When this + * counter hits zero the block is prepared and can be inserted into the + * btree. + */ + atomic_t prepare_actions; + + int err; struct thin_c *tc; - dm_block_t virt_block; + dm_block_t virt_begin, virt_end; dm_block_t data_block; - struct dm_bio_prison_cell *cell, *cell2; - int err; + struct dm_bio_prison_cell *cell; /* * If the bio covers the whole area of a block then we can avoid @@ -530,43 +786,43 @@ bio_end_io_t *saved_bi_end_io; }; -static void __maybe_add_mapping(struct dm_thin_new_mapping *m) +static void __complete_mapping_preparation(struct dm_thin_new_mapping *m) { struct pool *pool = m->tc->pool; - if (m->quiesced && m->prepared) { - list_add(&m->list, &pool->prepared_mappings); + if (atomic_dec_and_test(&m->prepare_actions)) { + list_add_tail(&m->list, &pool->prepared_mappings); wake_worker(pool); } } -static void copy_complete(int read_err, unsigned long write_err, void *context) +static void complete_mapping_preparation(struct dm_thin_new_mapping *m) { unsigned long flags; - struct dm_thin_new_mapping *m = context; struct pool *pool = m->tc->pool; - m->err = read_err || write_err ? -EIO : 0; - spin_lock_irqsave(&pool->lock, flags); - m->prepared = 1; - __maybe_add_mapping(m); + __complete_mapping_preparation(m); spin_unlock_irqrestore(&pool->lock, flags); } -static void overwrite_endio(struct bio *bio, int err) +static void copy_complete(int read_err, unsigned long write_err, void *context) +{ + struct dm_thin_new_mapping *m = context; + + m->err = read_err || write_err ? -EIO : 0; + complete_mapping_preparation(m); +} + +static void overwrite_endio(struct bio *bio) { - unsigned long flags; struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); struct dm_thin_new_mapping *m = h->overwrite_mapping; - struct pool *pool = m->tc->pool; - m->err = err; + bio->bi_end_io = m->saved_bi_end_io; - spin_lock_irqsave(&pool->lock, flags); - m->prepared = 1; - __maybe_add_mapping(m); - spin_unlock_irqrestore(&pool->lock, flags); + m->err = bio->bi_error; + complete_mapping_preparation(m); } /*----------------------------------------------------------------*/ @@ -580,39 +836,79 @@ */ /* - * This sends the bios in the cell back to the deferred_bios list. + * This sends the bios in the cell, except the original holder, back + * to the deferred_bios list. */ -static void cell_defer(struct thin_c *tc, struct dm_bio_prison_cell *cell) +static void cell_defer_no_holder(struct thin_c *tc, struct dm_bio_prison_cell *cell) { struct pool *pool = tc->pool; unsigned long flags; - spin_lock_irqsave(&pool->lock, flags); - cell_release(pool, cell, &pool->deferred_bios); - spin_unlock_irqrestore(&tc->pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); + cell_release_no_holder(pool, cell, &tc->deferred_bio_list); + spin_unlock_irqrestore(&tc->lock, flags); wake_worker(pool); } -/* - * Same as cell_defer above, except it omits the original holder of the cell. - */ -static void cell_defer_no_holder(struct thin_c *tc, struct dm_bio_prison_cell *cell) +static void thin_defer_bio(struct thin_c *tc, struct bio *bio); + +struct remap_info { + struct thin_c *tc; + struct bio_list defer_bios; + struct bio_list issue_bios; +}; + +static void __inc_remap_and_issue_cell(void *context, + struct dm_bio_prison_cell *cell) { - struct pool *pool = tc->pool; - unsigned long flags; + struct remap_info *info = context; + struct bio *bio; - spin_lock_irqsave(&pool->lock, flags); - cell_release_no_holder(pool, cell, &pool->deferred_bios); - spin_unlock_irqrestore(&pool->lock, flags); + while ((bio = bio_list_pop(&cell->bios))) { + if (bio->bi_rw & (REQ_DISCARD | REQ_FLUSH | REQ_FUA)) + bio_list_add(&info->defer_bios, bio); + else { + inc_all_io_entry(info->tc->pool, bio); - wake_worker(pool); + /* + * We can't issue the bios with the bio prison lock + * held, so we add them to a list to issue on + * return from this function. + */ + bio_list_add(&info->issue_bios, bio); + } + } +} + +static void inc_remap_and_issue_cell(struct thin_c *tc, + struct dm_bio_prison_cell *cell, + dm_block_t block) +{ + struct bio *bio; + struct remap_info info; + + info.tc = tc; + bio_list_init(&info.defer_bios); + bio_list_init(&info.issue_bios); + + /* + * We have to be careful to inc any bios we're about to issue + * before the cell is released, and avoid a race with new bios + * being added to the cell. + */ + cell_visit_release(tc->pool, __inc_remap_and_issue_cell, + &info, cell); + + while ((bio = bio_list_pop(&info.defer_bios))) + thin_defer_bio(tc, bio); + + while ((bio = bio_list_pop(&info.issue_bios))) + remap_and_issue(info.tc, bio, block); } static void process_prepared_mapping_fail(struct dm_thin_new_mapping *m) { - if (m->bio) - m->bio->bi_end_io = m->saved_bi_end_io; cell_error(m->tc->pool, m->cell); list_del(&m->list); mempool_free(m, m->tc->pool->mapping_pool); @@ -622,13 +918,9 @@ { struct thin_c *tc = m->tc; struct pool *pool = tc->pool; - struct bio *bio; + struct bio *bio = m->bio; int r; - bio = m->bio; - if (bio) - bio->bi_end_io = m->saved_bi_end_io; - if (m->err) { cell_error(pool, m->cell); goto out; @@ -639,11 +931,9 @@ * Any I/O for this block arriving after this point will get * remapped to it directly. */ - r = dm_thin_insert_block(tc->td, m->virt_block, m->data_block); + r = dm_thin_insert_block(tc->td, m->virt_begin, m->data_block); if (r) { - DMERR_LIMIT("%s: dm_thin_insert_block() failed: error = %d", - dm_device_name(pool->pool_md), r); - set_pool_mode(pool, PM_READ_ONLY); + metadata_operation_failed(pool, "dm_thin_insert_block", r); cell_error(pool, m->cell); goto out; } @@ -655,60 +945,126 @@ * the bios in the cell. */ if (bio) { - cell_defer_no_holder(tc, m->cell); - bio_endio(bio, 0); - } else - cell_defer(tc, m->cell); + inc_remap_and_issue_cell(tc, m->cell, m->data_block); + bio_endio(bio); + } else { + inc_all_io_entry(tc->pool, m->cell->holder); + remap_and_issue(tc, m->cell->holder, m->data_block); + inc_remap_and_issue_cell(tc, m->cell, m->data_block); + } out: list_del(&m->list); mempool_free(m, pool->mapping_pool); } -static void process_prepared_discard_fail(struct dm_thin_new_mapping *m) +/*----------------------------------------------------------------*/ + +static void free_discard_mapping(struct dm_thin_new_mapping *m) { struct thin_c *tc = m->tc; + if (m->cell) + cell_defer_no_holder(tc, m->cell); + mempool_free(m, tc->pool->mapping_pool); +} +static void process_prepared_discard_fail(struct dm_thin_new_mapping *m) +{ bio_io_error(m->bio); + free_discard_mapping(m); +} + +static void process_prepared_discard_success(struct dm_thin_new_mapping *m) +{ + bio_endio(m->bio); + free_discard_mapping(m); +} + +static void process_prepared_discard_no_passdown(struct dm_thin_new_mapping *m) +{ + int r; + struct thin_c *tc = m->tc; + + r = dm_thin_remove_range(tc->td, m->cell->key.block_begin, m->cell->key.block_end); + if (r) { + metadata_operation_failed(tc->pool, "dm_thin_remove_range", r); + bio_io_error(m->bio); + } else + bio_endio(m->bio); + cell_defer_no_holder(tc, m->cell); - cell_defer_no_holder(tc, m->cell2); mempool_free(m, tc->pool->mapping_pool); } -static void process_prepared_discard_passdown(struct dm_thin_new_mapping *m) +static int passdown_double_checking_shared_status(struct dm_thin_new_mapping *m) { + /* + * We've already unmapped this range of blocks, but before we + * passdown we have to check that these blocks are now unused. + */ + int r; + bool used = true; struct thin_c *tc = m->tc; + struct pool *pool = tc->pool; + dm_block_t b = m->data_block, e, end = m->data_block + m->virt_end - m->virt_begin; - inc_all_io_entry(tc->pool, m->bio); - cell_defer_no_holder(tc, m->cell); - cell_defer_no_holder(tc, m->cell2); + while (b != end) { + /* find start of unmapped run */ + for (; b < end; b++) { + r = dm_pool_block_is_used(pool->pmd, b, &used); + if (r) + return r; - if (m->pass_discard) - if (m->definitely_not_shared) - remap_and_issue(tc, m->bio, m->data_block); - else { - bool used = false; - if (dm_pool_block_is_used(tc->pool->pmd, m->data_block, &used) || used) - bio_endio(m->bio, 0); - else - remap_and_issue(tc, m->bio, m->data_block); + if (!used) + break; } - else - bio_endio(m->bio, 0); - mempool_free(m, tc->pool->mapping_pool); + if (b == end) + break; + + /* find end of run */ + for (e = b + 1; e != end; e++) { + r = dm_pool_block_is_used(pool->pmd, e, &used); + if (r) + return r; + + if (used) + break; + } + + r = issue_discard(tc, b, e, m->bio); + if (r) + return r; + + b = e; + } + + return 0; } -static void process_prepared_discard(struct dm_thin_new_mapping *m) +static void process_prepared_discard_passdown(struct dm_thin_new_mapping *m) { int r; struct thin_c *tc = m->tc; + struct pool *pool = tc->pool; - r = dm_thin_remove_block(tc->td, m->virt_block); + r = dm_thin_remove_range(tc->td, m->virt_begin, m->virt_end); if (r) - DMERR_LIMIT("dm_thin_remove_block() failed"); + metadata_operation_failed(pool, "dm_thin_remove_range", r); - process_prepared_discard_passdown(m); + else if (m->maybe_shared) + r = passdown_double_checking_shared_status(m); + else + r = issue_discard(tc, m->data_block, m->data_block + (m->virt_end - m->virt_begin), m->bio); + + /* + * Even if r is set, there could be sub discards in flight that we + * need to wait for. + */ + m->bio->bi_error = r; + bio_endio(m->bio); + cell_defer_no_holder(tc, m->cell); + mempool_free(m, pool->mapping_pool); } static void process_prepared(struct pool *pool, struct list_head *head, @@ -732,7 +1088,8 @@ */ static int io_overlaps_block(struct pool *pool, struct bio *bio) { - return bio->bi_size == (pool->sectors_per_block << SECTOR_SHIFT); + return bio->bi_iter.bi_size == + (pool->sectors_per_block << SECTOR_SHIFT); } static int io_overwrites_block(struct pool *pool, struct bio *bio) @@ -773,22 +1130,65 @@ return m; } +static void ll_zero(struct thin_c *tc, struct dm_thin_new_mapping *m, + sector_t begin, sector_t end) +{ + int r; + struct dm_io_region to; + + to.bdev = tc->pool_dev->bdev; + to.sector = begin; + to.count = end - begin; + + r = dm_kcopyd_zero(tc->pool->copier, 1, &to, 0, copy_complete, m); + if (r < 0) { + DMERR_LIMIT("dm_kcopyd_zero() failed"); + copy_complete(1, 1, m); + } +} + +static void remap_and_issue_overwrite(struct thin_c *tc, struct bio *bio, + dm_block_t data_begin, + struct dm_thin_new_mapping *m) +{ + struct pool *pool = tc->pool; + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); + + h->overwrite_mapping = m; + m->bio = bio; + save_and_set_endio(bio, &m->saved_bi_end_io, overwrite_endio); + inc_all_io_entry(pool, bio); + remap_and_issue(tc, bio, data_begin); +} + +/* + * A partial copy also needs to zero the uncopied region. + */ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block, struct dm_dev *origin, dm_block_t data_origin, dm_block_t data_dest, - struct dm_bio_prison_cell *cell, struct bio *bio) + struct dm_bio_prison_cell *cell, struct bio *bio, + sector_t len) { int r; struct pool *pool = tc->pool; struct dm_thin_new_mapping *m = get_next_mapping(pool); m->tc = tc; - m->virt_block = virt_block; + m->virt_begin = virt_block; + m->virt_end = virt_block + 1u; m->data_block = data_dest; m->cell = cell; + /* + * quiesce action + copy action + an extra reference held for the + * duration of this function (we may need to inc later for a + * partial zero). + */ + atomic_set(&m->prepare_actions, 3); + if (!dm_deferred_set_add_work(pool->shared_read_ds, &m->list)) - m->quiesced = 1; + complete_mapping_preparation(m); /* already quiesced */ /* * IO to pool_dev remaps to the pool target's data_dev. @@ -796,33 +1196,45 @@ * If the whole block of data is being overwritten, we can issue the * bio immediately. Otherwise we use kcopyd to clone the data first. */ - if (io_overwrites_block(pool, bio)) { - struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); - - h->overwrite_mapping = m; - m->bio = bio; - save_and_set_endio(bio, &m->saved_bi_end_io, overwrite_endio); - inc_all_io_entry(pool, bio); - remap_and_issue(tc, bio, data_dest); - } else { + if (io_overwrites_block(pool, bio)) + remap_and_issue_overwrite(tc, bio, data_dest, m); + else { struct dm_io_region from, to; from.bdev = origin->bdev; from.sector = data_origin * pool->sectors_per_block; - from.count = pool->sectors_per_block; + from.count = len; to.bdev = tc->pool_dev->bdev; to.sector = data_dest * pool->sectors_per_block; - to.count = pool->sectors_per_block; + to.count = len; r = dm_kcopyd_copy(pool->copier, &from, 1, &to, 0, copy_complete, m); if (r < 0) { - mempool_free(m, pool->mapping_pool); DMERR_LIMIT("dm_kcopyd_copy() failed"); - cell_error(pool, cell); + copy_complete(1, 1, m); + + /* + * We allow the zero to be issued, to simplify the + * error path. Otherwise we'd need to start + * worrying about decrementing the prepare_actions + * counter. + */ + } + + /* + * Do we need to zero a tail region? + */ + if (len < pool->sectors_per_block && pool->pf.zero_new_blocks) { + atomic_inc(&m->prepare_actions); + ll_zero(tc, m, + data_dest * pool->sectors_per_block + len, + (data_dest + 1) * pool->sectors_per_block); } } + + complete_mapping_preparation(m); /* drop our ref */ } static void schedule_internal_copy(struct thin_c *tc, dm_block_t virt_block, @@ -830,15 +1242,8 @@ struct dm_bio_prison_cell *cell, struct bio *bio) { schedule_copy(tc, virt_block, tc->pool_dev, - data_origin, data_dest, cell, bio); -} - -static void schedule_external_copy(struct thin_c *tc, dm_block_t virt_block, - dm_block_t data_dest, - struct dm_bio_prison_cell *cell, struct bio *bio) -{ - schedule_copy(tc, virt_block, tc->origin_dev, - virt_block, data_dest, cell, bio); + data_origin, data_dest, cell, bio, + tc->pool->sectors_per_block); } static void schedule_zero(struct thin_c *tc, dm_block_t virt_block, @@ -848,10 +1253,10 @@ struct pool *pool = tc->pool; struct dm_thin_new_mapping *m = get_next_mapping(pool); - m->quiesced = 1; - m->prepared = 0; + atomic_set(&m->prepare_actions, 1); /* no need to quiesce */ m->tc = tc; - m->virt_block = virt_block; + m->virt_begin = virt_block; + m->virt_end = virt_block + 1u; m->data_block = data_block; m->cell = cell; @@ -860,115 +1265,133 @@ * zeroing pre-existing data, we can issue the bio immediately. * Otherwise we use kcopyd to zero the data first. */ - if (!pool->pf.zero_new_blocks) + if (pool->pf.zero_new_blocks) { + if (io_overwrites_block(pool, bio)) + remap_and_issue_overwrite(tc, bio, data_block, m); + else + ll_zero(tc, m, data_block * pool->sectors_per_block, + (data_block + 1) * pool->sectors_per_block); + } else process_prepared_mapping(m); +} - else if (io_overwrites_block(pool, bio)) { - struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); - - h->overwrite_mapping = m; - m->bio = bio; - save_and_set_endio(bio, &m->saved_bi_end_io, overwrite_endio); - inc_all_io_entry(pool, bio); - remap_and_issue(tc, bio, data_block); - } else { - int r; - struct dm_io_region to; +static void schedule_external_copy(struct thin_c *tc, dm_block_t virt_block, + dm_block_t data_dest, + struct dm_bio_prison_cell *cell, struct bio *bio) +{ + struct pool *pool = tc->pool; + sector_t virt_block_begin = virt_block * pool->sectors_per_block; + sector_t virt_block_end = (virt_block + 1) * pool->sectors_per_block; - to.bdev = tc->pool_dev->bdev; - to.sector = data_block * pool->sectors_per_block; - to.count = pool->sectors_per_block; + if (virt_block_end <= tc->origin_size) + schedule_copy(tc, virt_block, tc->origin_dev, + virt_block, data_dest, cell, bio, + pool->sectors_per_block); + + else if (virt_block_begin < tc->origin_size) + schedule_copy(tc, virt_block, tc->origin_dev, + virt_block, data_dest, cell, bio, + tc->origin_size - virt_block_begin); - r = dm_kcopyd_zero(pool->copier, 1, &to, 0, copy_complete, m); - if (r < 0) { - mempool_free(m, pool->mapping_pool); - DMERR_LIMIT("dm_kcopyd_zero() failed"); - cell_error(pool, cell); - } - } + else + schedule_zero(tc, virt_block, data_dest, cell, bio); } -static int commit(struct pool *pool) +static void set_pool_mode(struct pool *pool, enum pool_mode new_mode); + +static void check_for_space(struct pool *pool) { int r; + dm_block_t nr_free; - r = dm_pool_commit_metadata(pool->pmd); + if (get_pool_mode(pool) != PM_OUT_OF_DATA_SPACE) + return; + + r = dm_pool_get_free_block_count(pool->pmd, &nr_free); if (r) - DMERR_LIMIT("commit failed: error = %d", r); + return; - return r; + if (nr_free) + set_pool_mode(pool, PM_WRITE); } /* * A non-zero return indicates read_only or fail_io mode. * Many callers don't care about the return value. */ -static int commit_or_fallback(struct pool *pool) +static int commit(struct pool *pool) { int r; - if (get_pool_mode(pool) != PM_WRITE) + if (get_pool_mode(pool) >= PM_READ_ONLY) return -EINVAL; - r = commit(pool); + r = dm_pool_commit_metadata(pool->pmd); if (r) - set_pool_mode(pool, PM_READ_ONLY); + metadata_operation_failed(pool, "dm_pool_commit_metadata", r); + else + check_for_space(pool); return r; } -static int alloc_data_block(struct thin_c *tc, dm_block_t *result) +static void check_low_water_mark(struct pool *pool, dm_block_t free_blocks) { - int r; - dm_block_t free_blocks; unsigned long flags; - struct pool *pool = tc->pool; - - r = dm_pool_get_free_block_count(pool->pmd, &free_blocks); - if (r) - return r; if (free_blocks <= pool->low_water_blocks && !pool->low_water_triggered) { DMWARN("%s: reached low water mark for data device: sending event.", dm_device_name(pool->pool_md)); spin_lock_irqsave(&pool->lock, flags); - pool->low_water_triggered = 1; + pool->low_water_triggered = true; spin_unlock_irqrestore(&pool->lock, flags); dm_table_event(pool->ti->table); } +} + +static int alloc_data_block(struct thin_c *tc, dm_block_t *result) +{ + int r; + dm_block_t free_blocks; + struct pool *pool = tc->pool; + + if (WARN_ON(get_pool_mode(pool) != PM_WRITE)) + return -EINVAL; + + r = dm_pool_get_free_block_count(pool->pmd, &free_blocks); + if (r) { + metadata_operation_failed(pool, "dm_pool_get_free_block_count", r); + return r; + } + + check_low_water_mark(pool, free_blocks); if (!free_blocks) { - if (pool->no_free_space) - return -ENOSPC; - else { - /* - * Try to commit to see if that will free up some - * more space. - */ - (void) commit_or_fallback(pool); + /* + * Try to commit to see if that will free up some + * more space. + */ + r = commit(pool); + if (r) + return r; - r = dm_pool_get_free_block_count(pool->pmd, &free_blocks); - if (r) - return r; + r = dm_pool_get_free_block_count(pool->pmd, &free_blocks); + if (r) { + metadata_operation_failed(pool, "dm_pool_get_free_block_count", r); + return r; + } - /* - * If we still have no space we set a flag to avoid - * doing all this checking and return -ENOSPC. - */ - if (!free_blocks) { - DMWARN("%s: no free space available.", - dm_device_name(pool->pool_md)); - spin_lock_irqsave(&pool->lock, flags); - pool->no_free_space = 1; - spin_unlock_irqrestore(&pool->lock, flags); - return -ENOSPC; - } + if (!free_blocks) { + set_pool_mode(pool, PM_OUT_OF_DATA_SPACE); + return -ENOSPC; } } r = dm_pool_alloc_data_block(pool->pmd, result); - if (r) + if (r) { + metadata_operation_failed(pool, "dm_pool_alloc_data_block", r); return r; + } return 0; } @@ -981,18 +1404,58 @@ { struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); struct thin_c *tc = h->tc; - struct pool *pool = tc->pool; unsigned long flags; - spin_lock_irqsave(&pool->lock, flags); - bio_list_add(&pool->retry_on_resume_list, bio); - spin_unlock_irqrestore(&pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); + bio_list_add(&tc->retry_on_resume_list, bio); + spin_unlock_irqrestore(&tc->lock, flags); +} + +static int should_error_unserviceable_bio(struct pool *pool) +{ + enum pool_mode m = get_pool_mode(pool); + + switch (m) { + case PM_WRITE: + /* Shouldn't get here */ + DMERR_LIMIT("bio unserviceable, yet pool is in PM_WRITE mode"); + return -EIO; + + case PM_OUT_OF_DATA_SPACE: + return pool->pf.error_if_no_space ? -ENOSPC : 0; + + case PM_READ_ONLY: + case PM_FAIL: + return -EIO; + default: + /* Shouldn't get here */ + DMERR_LIMIT("bio unserviceable, yet pool has an unknown mode"); + return -EIO; + } +} + +static void handle_unserviceable_bio(struct pool *pool, struct bio *bio) +{ + int error = should_error_unserviceable_bio(pool); + + if (error) { + bio->bi_error = error; + bio_endio(bio); + } else + retry_on_resume(bio); } -static void no_space(struct pool *pool, struct dm_bio_prison_cell *cell) +static void retry_bios_on_resume(struct pool *pool, struct dm_bio_prison_cell *cell) { struct bio *bio; struct bio_list bios; + int error; + + error = should_error_unserviceable_bio(pool); + if (error) { + cell_error_with_code(pool, cell, error); + return; + } bio_list_init(&bios); cell_release(pool, cell, &bios); @@ -1001,90 +1464,150 @@ retry_on_resume(bio); } -static void process_discard(struct thin_c *tc, struct bio *bio) +static void process_discard_cell_no_passdown(struct thin_c *tc, + struct dm_bio_prison_cell *virt_cell) { - int r; - unsigned long flags; struct pool *pool = tc->pool; - struct dm_bio_prison_cell *cell, *cell2; - struct dm_cell_key key, key2; - dm_block_t block = get_bio_block(tc, bio); - struct dm_thin_lookup_result lookup_result; - struct dm_thin_new_mapping *m; + struct dm_thin_new_mapping *m = get_next_mapping(pool); - build_virtual_key(tc->td, block, &key); - if (bio_detain(tc->pool, &key, bio, &cell)) - return; + /* + * We don't need to lock the data blocks, since there's no + * passdown. We only lock data blocks for allocation and breaking sharing. + */ + m->tc = tc; + m->virt_begin = virt_cell->key.block_begin; + m->virt_end = virt_cell->key.block_end; + m->cell = virt_cell; + m->bio = virt_cell->holder; - r = dm_thin_find_block(tc->td, block, 1, &lookup_result); - switch (r) { - case 0: - /* - * Check nobody is fiddling with this pool block. This can - * happen if someone's in the process of breaking sharing - * on this block. - */ - build_data_key(tc->td, lookup_result.block, &key2); - if (bio_detain(tc->pool, &key2, bio, &cell2)) { - cell_defer_no_holder(tc, cell); - break; - } + if (!dm_deferred_set_add_work(pool->all_io_ds, &m->list)) + pool->process_prepared_discard(m); +} - if (io_overlaps_block(pool, bio)) { - /* - * IO may still be going to the destination block. We must - * quiesce before we can do the removal. - */ - m = get_next_mapping(pool); - m->tc = tc; - m->pass_discard = pool->pf.discard_passdown; - m->definitely_not_shared = !lookup_result.shared; - m->virt_block = block; - m->data_block = lookup_result.block; - m->cell = cell; - m->cell2 = cell2; - m->bio = bio; - - if (!dm_deferred_set_add_work(pool->all_io_ds, &m->list)) { - spin_lock_irqsave(&pool->lock, flags); - list_add(&m->list, &pool->prepared_discards); - spin_unlock_irqrestore(&pool->lock, flags); - wake_worker(pool); - } - } else { - inc_all_io_entry(pool, bio); - cell_defer_no_holder(tc, cell); - cell_defer_no_holder(tc, cell2); +/* + * __bio_inc_remaining() is used to defer parent bios's end_io until + * we _know_ all chained sub range discard bios have completed. + */ +static inline void __bio_inc_remaining(struct bio *bio) +{ + bio->bi_flags |= (1 << BIO_CHAIN); + smp_mb__before_atomic(); + atomic_inc(&bio->__bi_remaining); +} +static void break_up_discard_bio(struct thin_c *tc, dm_block_t begin, dm_block_t end, + struct bio *bio) +{ + struct pool *pool = tc->pool; + + int r; + bool maybe_shared; + struct dm_cell_key data_key; + struct dm_bio_prison_cell *data_cell; + struct dm_thin_new_mapping *m; + dm_block_t virt_begin, virt_end, data_begin; + + while (begin != end) { + r = ensure_next_mapping(pool); + if (r) + /* we did our best */ + return; + + r = dm_thin_find_mapped_range(tc->td, begin, end, &virt_begin, &virt_end, + &data_begin, &maybe_shared); + if (r) /* - * The DM core makes sure that the discard doesn't span - * a block boundary. So we submit the discard of a - * partial block appropriately. + * Silently fail, letting any mappings we've + * created complete. */ - if ((!lookup_result.shared) && pool->pf.discard_passdown) - remap_and_issue(tc, bio, lookup_result.block); - else - bio_endio(bio, 0); + break; + + build_key(tc->td, PHYSICAL, data_begin, data_begin + (virt_end - virt_begin), &data_key); + if (bio_detain(tc->pool, &data_key, NULL, &data_cell)) { + /* contention, we'll give up with this range */ + begin = virt_end; + continue; } - break; - case -ENODATA: /* - * It isn't provisioned, just forget it. + * IO may still be going to the destination block. We must + * quiesce before we can do the removal. */ - cell_defer_no_holder(tc, cell); - bio_endio(bio, 0); - break; + m = get_next_mapping(pool); + m->tc = tc; + m->maybe_shared = maybe_shared; + m->virt_begin = virt_begin; + m->virt_end = virt_end; + m->data_block = data_begin; + m->cell = data_cell; + m->bio = bio; - default: - DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d", - __func__, r); - cell_defer_no_holder(tc, cell); - bio_io_error(bio); - break; + /* + * The parent bio must not complete before sub discard bios are + * chained to it (see __blkdev_issue_discard_async's bio_chain)! + * + * This per-mapping bi_remaining increment is paired with + * the implicit decrement that occurs via bio_endio() in + * process_prepared_discard_{passdown,no_passdown}. + */ + __bio_inc_remaining(bio); + if (!dm_deferred_set_add_work(pool->all_io_ds, &m->list)) + pool->process_prepared_discard(m); + + begin = virt_end; } } +static void process_discard_cell_passdown(struct thin_c *tc, struct dm_bio_prison_cell *virt_cell) +{ + struct bio *bio = virt_cell->holder; + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); + + /* + * The virt_cell will only get freed once the origin bio completes. + * This means it will remain locked while all the individual + * passdown bios are in flight. + */ + h->cell = virt_cell; + break_up_discard_bio(tc, virt_cell->key.block_begin, virt_cell->key.block_end, bio); + + /* + * We complete the bio now, knowing that the bi_remaining field + * will prevent completion until the sub range discards have + * completed. + */ + bio_endio(bio); +} + +static void process_discard_bio(struct thin_c *tc, struct bio *bio) +{ + dm_block_t begin, end; + struct dm_cell_key virt_key; + struct dm_bio_prison_cell *virt_cell; + + get_bio_block_range(tc, bio, &begin, &end); + if (begin == end) { + /* + * The discard covers less than a block. + */ + bio_endio(bio); + return; + } + + build_key(tc->td, VIRTUAL, begin, end, &virt_key); + if (bio_detain(tc->pool, &virt_key, bio, &virt_cell)) + /* + * Potential starvation issue: We're relying on the + * fs/application being well behaved, and not trying to + * send IO to a region at the same time as discarding it. + * If they do this persistently then it's possible this + * cell will never be granted. + */ + return; + + tc->pool->process_discard_cell(tc, virt_cell); +} + static void break_sharing(struct thin_c *tc, struct bio *bio, dm_block_t block, struct dm_cell_key *key, struct dm_thin_lookup_result *lookup_result, @@ -1092,6 +1615,7 @@ { int r; dm_block_t data_block; + struct pool *pool = tc->pool; r = alloc_data_block(tc, &data_block); switch (r) { @@ -1101,22 +1625,64 @@ break; case -ENOSPC: - no_space(tc->pool, cell); + retry_bios_on_resume(pool, cell); break; default: DMERR_LIMIT("%s: alloc_data_block() failed: error = %d", __func__, r); - cell_error(tc->pool, cell); + cell_error(pool, cell); break; } } +static void __remap_and_issue_shared_cell(void *context, + struct dm_bio_prison_cell *cell) +{ + struct remap_info *info = context; + struct bio *bio; + + while ((bio = bio_list_pop(&cell->bios))) { + if ((bio_data_dir(bio) == WRITE) || + (bio->bi_rw & (REQ_DISCARD | REQ_FLUSH | REQ_FUA))) + bio_list_add(&info->defer_bios, bio); + else { + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));; + + h->shared_read_entry = dm_deferred_entry_inc(info->tc->pool->shared_read_ds); + inc_all_io_entry(info->tc->pool, bio); + bio_list_add(&info->issue_bios, bio); + } + } +} + +static void remap_and_issue_shared_cell(struct thin_c *tc, + struct dm_bio_prison_cell *cell, + dm_block_t block) +{ + struct bio *bio; + struct remap_info info; + + info.tc = tc; + bio_list_init(&info.defer_bios); + bio_list_init(&info.issue_bios); + + cell_visit_release(tc->pool, __remap_and_issue_shared_cell, + &info, cell); + + while ((bio = bio_list_pop(&info.defer_bios))) + thin_defer_bio(tc, bio); + + while ((bio = bio_list_pop(&info.issue_bios))) + remap_and_issue(tc, bio, block); +} + static void process_shared_bio(struct thin_c *tc, struct bio *bio, dm_block_t block, - struct dm_thin_lookup_result *lookup_result) + struct dm_thin_lookup_result *lookup_result, + struct dm_bio_prison_cell *virt_cell) { - struct dm_bio_prison_cell *cell; + struct dm_bio_prison_cell *data_cell; struct pool *pool = tc->pool; struct dm_cell_key key; @@ -1125,19 +1691,23 @@ * of being broken so we have nothing further to do here. */ build_data_key(tc->td, lookup_result->block, &key); - if (bio_detain(pool, &key, bio, &cell)) + if (bio_detain(pool, &key, bio, &data_cell)) { + cell_defer_no_holder(tc, virt_cell); return; + } - if (bio_data_dir(bio) == WRITE && bio->bi_size) - break_sharing(tc, bio, block, &key, lookup_result, cell); - else { + if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) { + break_sharing(tc, bio, block, &key, lookup_result, data_cell); + cell_defer_no_holder(tc, virt_cell); + } else { struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); h->shared_read_entry = dm_deferred_entry_inc(pool->shared_read_ds); inc_all_io_entry(pool, bio); - cell_defer_no_holder(tc, cell); - remap_and_issue(tc, bio, lookup_result->block); + + remap_and_issue_shared_cell(tc, data_cell, lookup_result->block); + remap_and_issue_shared_cell(tc, virt_cell, lookup_result->block); } } @@ -1151,7 +1721,7 @@ /* * Remap empty bios (flushes) immediately, without provisioning. */ - if (!bio->bi_size) { + if (!bio->bi_iter.bi_size) { inc_all_io_entry(pool, bio); cell_defer_no_holder(tc, cell); @@ -1165,7 +1735,7 @@ if (bio_data_dir(bio) == READ) { zero_fill_bio(bio); cell_defer_no_holder(tc, cell); - bio_endio(bio, 0); + bio_endio(bio); return; } @@ -1179,46 +1749,39 @@ break; case -ENOSPC: - no_space(pool, cell); + retry_bios_on_resume(pool, cell); break; default: DMERR_LIMIT("%s: alloc_data_block() failed: error = %d", __func__, r); - set_pool_mode(pool, PM_READ_ONLY); cell_error(pool, cell); break; } } -static void process_bio(struct thin_c *tc, struct bio *bio) +static void process_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell) { int r; struct pool *pool = tc->pool; + struct bio *bio = cell->holder; dm_block_t block = get_bio_block(tc, bio); - struct dm_bio_prison_cell *cell; - struct dm_cell_key key; struct dm_thin_lookup_result lookup_result; - /* - * If cell is already occupied, then the block is already - * being provisioned so we have nothing further to do here. - */ - build_virtual_key(tc->td, block, &key); - if (bio_detain(pool, &key, bio, &cell)) + if (tc->requeue_mode) { + cell_requeue(pool, cell); return; + } r = dm_thin_find_block(tc->td, block, 1, &lookup_result); switch (r) { case 0: - if (lookup_result.shared) { - process_shared_bio(tc, bio, block, &lookup_result); - cell_defer_no_holder(tc, cell); /* FIXME: pass this cell into process_shared? */ - } else { + if (lookup_result.shared) + process_shared_bio(tc, bio, block, &lookup_result, cell); + else { inc_all_io_entry(pool, bio); - cell_defer_no_holder(tc, cell); - remap_and_issue(tc, bio, lookup_result.block); + inc_remap_and_issue_cell(tc, cell, lookup_result.block); } break; @@ -1227,7 +1790,18 @@ inc_all_io_entry(pool, bio); cell_defer_no_holder(tc, cell); - remap_to_origin_and_issue(tc, bio); + if (bio_end_sector(bio) <= tc->origin_size) + remap_to_origin_and_issue(tc, bio); + + else if (bio->bi_iter.bi_sector < tc->origin_size) { + zero_fill_bio(bio); + bio->bi_iter.bi_size = (tc->origin_size - bio->bi_iter.bi_sector) << SECTOR_SHIFT; + remap_to_origin_and_issue(tc, bio); + + } else { + zero_fill_bio(bio); + bio_endio(bio); + } } else provision_block(tc, bio, block, cell); break; @@ -1241,7 +1815,26 @@ } } -static void process_bio_read_only(struct thin_c *tc, struct bio *bio) +static void process_bio(struct thin_c *tc, struct bio *bio) +{ + struct pool *pool = tc->pool; + dm_block_t block = get_bio_block(tc, bio); + struct dm_bio_prison_cell *cell; + struct dm_cell_key key; + + /* + * If cell is already occupied, then the block is already + * being provisioned so we have nothing further to do here. + */ + build_virtual_key(tc->td, block, &key); + if (bio_detain(pool, &key, bio, &cell)) + return; + + process_cell(tc, cell); +} + +static void __process_bio_read_only(struct thin_c *tc, struct bio *bio, + struct dm_bio_prison_cell *cell) { int r; int rw = bio_data_dir(bio); @@ -1251,17 +1844,23 @@ r = dm_thin_find_block(tc->td, block, 1, &lookup_result); switch (r) { case 0: - if (lookup_result.shared && (rw == WRITE) && bio->bi_size) - bio_io_error(bio); - else { + if (lookup_result.shared && (rw == WRITE) && bio->bi_iter.bi_size) { + handle_unserviceable_bio(tc->pool, bio); + if (cell) + cell_defer_no_holder(tc, cell); + } else { inc_all_io_entry(tc->pool, bio); remap_and_issue(tc, bio, lookup_result.block); + if (cell) + inc_remap_and_issue_cell(tc, cell, lookup_result.block); } break; case -ENODATA: + if (cell) + cell_defer_no_holder(tc, cell); if (rw != READ) { - bio_io_error(bio); + handle_unserviceable_bio(tc->pool, bio); break; } @@ -1272,59 +1871,165 @@ } zero_fill_bio(bio); - bio_endio(bio, 0); + bio_endio(bio); break; default: DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d", __func__, r); + if (cell) + cell_defer_no_holder(tc, cell); bio_io_error(bio); break; } } +static void process_bio_read_only(struct thin_c *tc, struct bio *bio) +{ + __process_bio_read_only(tc, bio, NULL); +} + +static void process_cell_read_only(struct thin_c *tc, struct dm_bio_prison_cell *cell) +{ + __process_bio_read_only(tc, cell->holder, cell); +} + +static void process_bio_success(struct thin_c *tc, struct bio *bio) +{ + bio_endio(bio); +} + static void process_bio_fail(struct thin_c *tc, struct bio *bio) { bio_io_error(bio); } +static void process_cell_success(struct thin_c *tc, struct dm_bio_prison_cell *cell) +{ + cell_success(tc->pool, cell); +} + +static void process_cell_fail(struct thin_c *tc, struct dm_bio_prison_cell *cell) +{ + cell_error(tc->pool, cell); +} + /* * FIXME: should we also commit due to size of transaction, measured in * metadata blocks? */ static int need_commit_due_to_time(struct pool *pool) { - return jiffies < pool->last_commit_jiffies || - jiffies > pool->last_commit_jiffies + COMMIT_PERIOD; + return !time_in_range(jiffies, pool->last_commit_jiffies, + pool->last_commit_jiffies + COMMIT_PERIOD); } -static void process_deferred_bios(struct pool *pool) +#define thin_pbd(node) rb_entry((node), struct dm_thin_endio_hook, rb_node) +#define thin_bio(pbd) dm_bio_from_per_bio_data((pbd), sizeof(struct dm_thin_endio_hook)) + +static void __thin_bio_rb_add(struct thin_c *tc, struct bio *bio) { + struct rb_node **rbp, *parent; + struct dm_thin_endio_hook *pbd; + sector_t bi_sector = bio->bi_iter.bi_sector; + + rbp = &tc->sort_bio_list.rb_node; + parent = NULL; + while (*rbp) { + parent = *rbp; + pbd = thin_pbd(parent); + + if (bi_sector < thin_bio(pbd)->bi_iter.bi_sector) + rbp = &(*rbp)->rb_left; + else + rbp = &(*rbp)->rb_right; + } + + pbd = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); + rb_link_node(&pbd->rb_node, parent, rbp); + rb_insert_color(&pbd->rb_node, &tc->sort_bio_list); +} + +static void __extract_sorted_bios(struct thin_c *tc) +{ + struct rb_node *node; + struct dm_thin_endio_hook *pbd; + struct bio *bio; + + for (node = rb_first(&tc->sort_bio_list); node; node = rb_next(node)) { + pbd = thin_pbd(node); + bio = thin_bio(pbd); + + bio_list_add(&tc->deferred_bio_list, bio); + rb_erase(&pbd->rb_node, &tc->sort_bio_list); + } + + WARN_ON(!RB_EMPTY_ROOT(&tc->sort_bio_list)); +} + +static void __sort_thin_deferred_bios(struct thin_c *tc) +{ + struct bio *bio; + struct bio_list bios; + + bio_list_init(&bios); + bio_list_merge(&bios, &tc->deferred_bio_list); + bio_list_init(&tc->deferred_bio_list); + + /* Sort deferred_bio_list using rb-tree */ + while ((bio = bio_list_pop(&bios))) + __thin_bio_rb_add(tc, bio); + + /* + * Transfer the sorted bios in sort_bio_list back to + * deferred_bio_list to allow lockless submission of + * all bios. + */ + __extract_sorted_bios(tc); +} + +static void process_thin_deferred_bios(struct thin_c *tc) +{ + struct pool *pool = tc->pool; unsigned long flags; struct bio *bio; struct bio_list bios; + struct blk_plug plug; + unsigned count = 0; + + if (tc->requeue_mode) { + error_thin_bio_list(tc, &tc->deferred_bio_list, DM_ENDIO_REQUEUE); + return; + } bio_list_init(&bios); - spin_lock_irqsave(&pool->lock, flags); - bio_list_merge(&bios, &pool->deferred_bios); - bio_list_init(&pool->deferred_bios); - spin_unlock_irqrestore(&pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); - while ((bio = bio_list_pop(&bios))) { - struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); - struct thin_c *tc = h->tc; + if (bio_list_empty(&tc->deferred_bio_list)) { + spin_unlock_irqrestore(&tc->lock, flags); + return; + } + + __sort_thin_deferred_bios(tc); + + bio_list_merge(&bios, &tc->deferred_bio_list); + bio_list_init(&tc->deferred_bio_list); + + spin_unlock_irqrestore(&tc->lock, flags); + blk_start_plug(&plug); + while ((bio = bio_list_pop(&bios))) { /* * If we've got no free new_mapping structs, and processing * this bio might require one, we pause until there are some * prepared mappings to process. */ if (ensure_next_mapping(pool)) { - spin_lock_irqsave(&pool->lock, flags); - bio_list_add(&pool->deferred_bios, bio); - bio_list_merge(&pool->deferred_bios, &bios); - spin_unlock_irqrestore(&pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); + bio_list_add(&tc->deferred_bio_list, bio); + bio_list_merge(&tc->deferred_bio_list, &bios); + spin_unlock_irqrestore(&tc->lock, flags); break; } @@ -1332,6 +2037,148 @@ pool->process_discard(tc, bio); else pool->process_bio(tc, bio); + + if ((count++ & 127) == 0) { + throttle_work_update(&pool->throttle); + dm_pool_issue_prefetches(pool->pmd); + } + } + blk_finish_plug(&plug); +} + +static int cmp_cells(const void *lhs, const void *rhs) +{ + struct dm_bio_prison_cell *lhs_cell = *((struct dm_bio_prison_cell **) lhs); + struct dm_bio_prison_cell *rhs_cell = *((struct dm_bio_prison_cell **) rhs); + + BUG_ON(!lhs_cell->holder); + BUG_ON(!rhs_cell->holder); + + if (lhs_cell->holder->bi_iter.bi_sector < rhs_cell->holder->bi_iter.bi_sector) + return -1; + + if (lhs_cell->holder->bi_iter.bi_sector > rhs_cell->holder->bi_iter.bi_sector) + return 1; + + return 0; +} + +static unsigned sort_cells(struct pool *pool, struct list_head *cells) +{ + unsigned count = 0; + struct dm_bio_prison_cell *cell, *tmp; + + list_for_each_entry_safe(cell, tmp, cells, user_list) { + if (count >= CELL_SORT_ARRAY_SIZE) + break; + + pool->cell_sort_array[count++] = cell; + list_del(&cell->user_list); + } + + sort(pool->cell_sort_array, count, sizeof(cell), cmp_cells, NULL); + + return count; +} + +static void process_thin_deferred_cells(struct thin_c *tc) +{ + struct pool *pool = tc->pool; + unsigned long flags; + struct list_head cells; + struct dm_bio_prison_cell *cell; + unsigned i, j, count; + + INIT_LIST_HEAD(&cells); + + spin_lock_irqsave(&tc->lock, flags); + list_splice_init(&tc->deferred_cells, &cells); + spin_unlock_irqrestore(&tc->lock, flags); + + if (list_empty(&cells)) + return; + + do { + count = sort_cells(tc->pool, &cells); + + for (i = 0; i < count; i++) { + cell = pool->cell_sort_array[i]; + BUG_ON(!cell->holder); + + /* + * If we've got no free new_mapping structs, and processing + * this bio might require one, we pause until there are some + * prepared mappings to process. + */ + if (ensure_next_mapping(pool)) { + for (j = i; j < count; j++) + list_add(&pool->cell_sort_array[j]->user_list, &cells); + + spin_lock_irqsave(&tc->lock, flags); + list_splice(&cells, &tc->deferred_cells); + spin_unlock_irqrestore(&tc->lock, flags); + return; + } + + if (cell->holder->bi_rw & REQ_DISCARD) + pool->process_discard_cell(tc, cell); + else + pool->process_cell(tc, cell); + } + } while (!list_empty(&cells)); +} + +static void thin_get(struct thin_c *tc); +static void thin_put(struct thin_c *tc); + +/* + * We can't hold rcu_read_lock() around code that can block. So we + * find a thin with the rcu lock held; bump a refcount; then drop + * the lock. + */ +static struct thin_c *get_first_thin(struct pool *pool) +{ + struct thin_c *tc = NULL; + + rcu_read_lock(); + if (!list_empty(&pool->active_thins)) { + tc = list_entry_rcu(pool->active_thins.next, struct thin_c, list); + thin_get(tc); + } + rcu_read_unlock(); + + return tc; +} + +static struct thin_c *get_next_thin(struct pool *pool, struct thin_c *tc) +{ + struct thin_c *old_tc = tc; + + rcu_read_lock(); + list_for_each_entry_continue_rcu(tc, &pool->active_thins, list) { + thin_get(tc); + thin_put(old_tc); + rcu_read_unlock(); + return tc; + } + thin_put(old_tc); + rcu_read_unlock(); + + return NULL; +} + +static void process_deferred_bios(struct pool *pool) +{ + unsigned long flags; + struct bio *bio; + struct bio_list bios; + struct thin_c *tc; + + tc = get_first_thin(pool); + while (tc) { + process_thin_deferred_cells(tc); + process_thin_deferred_bios(tc); + tc = get_next_thin(pool, tc); } /* @@ -1348,7 +2195,7 @@ !(dm_pool_changed_this_transaction(pool->pmd) && need_commit_due_to_time(pool))) return; - if (commit_or_fallback(pool)) { + if (commit(pool)) { while ((bio = bio_list_pop(&bios))) bio_io_error(bio); return; @@ -1363,9 +2210,15 @@ { struct pool *pool = container_of(ws, struct pool, worker); + throttle_work_start(&pool->throttle); + dm_pool_issue_prefetches(pool->pmd); + throttle_work_update(&pool->throttle); process_prepared(pool, &pool->prepared_mappings, &pool->process_prepared_mapping); + throttle_work_update(&pool->throttle); process_prepared(pool, &pool->prepared_discards, &pool->process_prepared_discard); + throttle_work_update(&pool->throttle); process_deferred_bios(pool); + throttle_work_complete(&pool->throttle); } /* @@ -1379,6 +2232,86 @@ queue_delayed_work(pool->wq, &pool->waker, COMMIT_PERIOD); } +static void notify_of_pool_mode_change_to_oods(struct pool *pool); + +/* + * We're holding onto IO to allow userland time to react. After the + * timeout either the pool will have been resized (and thus back in + * PM_WRITE mode), or we degrade to PM_OUT_OF_DATA_SPACE w/ error_if_no_space. + */ +static void do_no_space_timeout(struct work_struct *ws) +{ + struct pool *pool = container_of(to_delayed_work(ws), struct pool, + no_space_timeout); + + if (get_pool_mode(pool) == PM_OUT_OF_DATA_SPACE && !pool->pf.error_if_no_space) { + pool->pf.error_if_no_space = true; + notify_of_pool_mode_change_to_oods(pool); + error_retry_list_with_code(pool, -ENOSPC); + } +} + +/*----------------------------------------------------------------*/ + +struct pool_work { + struct work_struct worker; + struct completion complete; +}; + +static struct pool_work *to_pool_work(struct work_struct *ws) +{ + return container_of(ws, struct pool_work, worker); +} + +static void pool_work_complete(struct pool_work *pw) +{ + complete(&pw->complete); +} + +static void pool_work_wait(struct pool_work *pw, struct pool *pool, + void (*fn)(struct work_struct *)) +{ + INIT_WORK_ONSTACK(&pw->worker, fn); + init_completion(&pw->complete); + queue_work(pool->wq, &pw->worker); + wait_for_completion(&pw->complete); +} + +/*----------------------------------------------------------------*/ + +struct noflush_work { + struct pool_work pw; + struct thin_c *tc; +}; + +static struct noflush_work *to_noflush(struct work_struct *ws) +{ + return container_of(to_pool_work(ws), struct noflush_work, pw); +} + +static void do_noflush_start(struct work_struct *ws) +{ + struct noflush_work *w = to_noflush(ws); + w->tc->requeue_mode = true; + requeue_io(w->tc); + pool_work_complete(&w->pw); +} + +static void do_noflush_stop(struct work_struct *ws) +{ + struct noflush_work *w = to_noflush(ws); + w->tc->requeue_mode = false; + pool_work_complete(&w->pw); +} + +static void noflush_work(struct thin_c *tc, void (*fn)(struct work_struct *)) +{ + struct noflush_work w; + + w.tc = tc; + pool_work_wait(&w.pw, tc->pool, fn); +} + /*----------------------------------------------------------------*/ static enum pool_mode get_pool_mode(struct pool *pool) @@ -1386,43 +2319,160 @@ return pool->pf.mode; } -static void set_pool_mode(struct pool *pool, enum pool_mode mode) +static void notify_of_pool_mode_change(struct pool *pool, const char *new_mode) { - int r; + dm_table_event(pool->ti->table); + DMINFO("%s: switching pool to %s mode", + dm_device_name(pool->pool_md), new_mode); +} + +static void notify_of_pool_mode_change_to_oods(struct pool *pool) +{ + if (!pool->pf.error_if_no_space) + notify_of_pool_mode_change(pool, "out-of-data-space (queue IO)"); + else + notify_of_pool_mode_change(pool, "out-of-data-space (error IO)"); +} - pool->pf.mode = mode; +static bool passdown_enabled(struct pool_c *pt) +{ + return pt->adjusted_pf.discard_passdown; +} - switch (mode) { +static void set_discard_callbacks(struct pool *pool) +{ + struct pool_c *pt = pool->ti->private; + + if (passdown_enabled(pt)) { + pool->process_discard_cell = process_discard_cell_passdown; + pool->process_prepared_discard = process_prepared_discard_passdown; + } else { + pool->process_discard_cell = process_discard_cell_no_passdown; + pool->process_prepared_discard = process_prepared_discard_no_passdown; + } +} + +static void set_pool_mode(struct pool *pool, enum pool_mode new_mode) +{ + struct pool_c *pt = pool->ti->private; + bool needs_check = dm_pool_metadata_needs_check(pool->pmd); + enum pool_mode old_mode = get_pool_mode(pool); + unsigned long no_space_timeout = ACCESS_ONCE(no_space_timeout_secs) * HZ; + + /* + * Never allow the pool to transition to PM_WRITE mode if user + * intervention is required to verify metadata and data consistency. + */ + if (new_mode == PM_WRITE && needs_check) { + DMERR("%s: unable to switch pool to write mode until repaired.", + dm_device_name(pool->pool_md)); + if (old_mode != new_mode) + new_mode = old_mode; + else + new_mode = PM_READ_ONLY; + } + /* + * If we were in PM_FAIL mode, rollback of metadata failed. We're + * not going to recover without a thin_repair. So we never let the + * pool move out of the old mode. + */ + if (old_mode == PM_FAIL) + new_mode = old_mode; + + switch (new_mode) { case PM_FAIL: - DMERR("switching pool to failure mode"); + if (old_mode != new_mode) + notify_of_pool_mode_change(pool, "failure"); + dm_pool_metadata_read_only(pool->pmd); pool->process_bio = process_bio_fail; pool->process_discard = process_bio_fail; + pool->process_cell = process_cell_fail; + pool->process_discard_cell = process_cell_fail; pool->process_prepared_mapping = process_prepared_mapping_fail; pool->process_prepared_discard = process_prepared_discard_fail; + + error_retry_list(pool); break; case PM_READ_ONLY: - DMERR("switching pool to read-only mode"); - r = dm_pool_abort_metadata(pool->pmd); - if (r) { - DMERR("aborting transaction failed"); - set_pool_mode(pool, PM_FAIL); - } else { - dm_pool_metadata_read_only(pool->pmd); - pool->process_bio = process_bio_read_only; - pool->process_discard = process_discard; - pool->process_prepared_mapping = process_prepared_mapping_fail; - pool->process_prepared_discard = process_prepared_discard_passdown; - } + if (old_mode != new_mode) + notify_of_pool_mode_change(pool, "read-only"); + dm_pool_metadata_read_only(pool->pmd); + pool->process_bio = process_bio_read_only; + pool->process_discard = process_bio_success; + pool->process_cell = process_cell_read_only; + pool->process_discard_cell = process_cell_success; + pool->process_prepared_mapping = process_prepared_mapping_fail; + pool->process_prepared_discard = process_prepared_discard_success; + + error_retry_list(pool); + break; + + case PM_OUT_OF_DATA_SPACE: + /* + * Ideally we'd never hit this state; the low water mark + * would trigger userland to extend the pool before we + * completely run out of data space. However, many small + * IOs to unprovisioned space can consume data space at an + * alarming rate. Adjust your low water mark if you're + * frequently seeing this mode. + */ + if (old_mode != new_mode) + notify_of_pool_mode_change_to_oods(pool); + pool->process_bio = process_bio_read_only; + pool->process_discard = process_discard_bio; + pool->process_cell = process_cell_read_only; + pool->process_prepared_mapping = process_prepared_mapping; + set_discard_callbacks(pool); + + if (!pool->pf.error_if_no_space && no_space_timeout) + queue_delayed_work(pool->wq, &pool->no_space_timeout, no_space_timeout); break; case PM_WRITE: + if (old_mode != new_mode) + notify_of_pool_mode_change(pool, "write"); + pool->pf.error_if_no_space = pt->requested_pf.error_if_no_space; + dm_pool_metadata_read_write(pool->pmd); pool->process_bio = process_bio; - pool->process_discard = process_discard; + pool->process_discard = process_discard_bio; + pool->process_cell = process_cell; pool->process_prepared_mapping = process_prepared_mapping; - pool->process_prepared_discard = process_prepared_discard; + set_discard_callbacks(pool); break; } + + pool->pf.mode = new_mode; + /* + * The pool mode may have changed, sync it so bind_control_target() + * doesn't cause an unexpected mode transition on resume. + */ + pt->adjusted_pf.mode = new_mode; +} + +static void abort_transaction(struct pool *pool) +{ + const char *dev_name = dm_device_name(pool->pool_md); + + DMERR_LIMIT("%s: aborting current metadata transaction", dev_name); + if (dm_pool_abort_metadata(pool->pmd)) { + DMERR("%s: failed to abort metadata transaction", dev_name); + set_pool_mode(pool, PM_FAIL); + } + + if (dm_pool_metadata_set_needs_check(pool->pmd)) { + DMERR("%s: failed to set 'needs_check' flag in metadata", dev_name); + set_pool_mode(pool, PM_FAIL); + } +} + +static void metadata_operation_failed(struct pool *pool, const char *op, int r) +{ + DMERR_LIMIT("%s: metadata operation '%s' failed: error = %d", + dm_device_name(pool->pool_md), op, r); + + abort_transaction(pool); + set_pool_mode(pool, PM_READ_ONLY); } /*----------------------------------------------------------------*/ @@ -1439,9 +2489,32 @@ unsigned long flags; struct pool *pool = tc->pool; - spin_lock_irqsave(&pool->lock, flags); - bio_list_add(&pool->deferred_bios, bio); - spin_unlock_irqrestore(&pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); + bio_list_add(&tc->deferred_bio_list, bio); + spin_unlock_irqrestore(&tc->lock, flags); + + wake_worker(pool); +} + +static void thin_defer_bio_with_throttle(struct thin_c *tc, struct bio *bio) +{ + struct pool *pool = tc->pool; + + throttle_lock(&pool->throttle); + thin_defer_bio(tc, bio); + throttle_unlock(&pool->throttle); +} + +static void thin_defer_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell) +{ + unsigned long flags; + struct pool *pool = tc->pool; + + throttle_lock(&pool->throttle); + spin_lock_irqsave(&tc->lock, flags); + list_add_tail(&cell->user_list, &tc->deferred_cells); + spin_unlock_irqrestore(&tc->lock, flags); + throttle_unlock(&pool->throttle); wake_worker(pool); } @@ -1454,6 +2527,7 @@ h->shared_read_entry = NULL; h->all_io_entry = NULL; h->overwrite_mapping = NULL; + h->cell = NULL; } /* @@ -1466,22 +2540,35 @@ dm_block_t block = get_bio_block(tc, bio); struct dm_thin_device *td = tc->td; struct dm_thin_lookup_result result; - struct dm_bio_prison_cell cell1, cell2; - struct dm_bio_prison_cell *cell_result; + struct dm_bio_prison_cell *virt_cell, *data_cell; struct dm_cell_key key; thin_hook_bio(tc, bio); + if (tc->requeue_mode) { + bio->bi_error = DM_ENDIO_REQUEUE; + bio_endio(bio); + return DM_MAPIO_SUBMITTED; + } + if (get_pool_mode(tc->pool) == PM_FAIL) { bio_io_error(bio); return DM_MAPIO_SUBMITTED; } if (bio->bi_rw & (REQ_DISCARD | REQ_FLUSH | REQ_FUA)) { - thin_defer_bio(tc, bio); + thin_defer_bio_with_throttle(tc, bio); return DM_MAPIO_SUBMITTED; } + /* + * We must hold the virtual cell before doing the lookup, otherwise + * there's a race with discard. + */ + build_virtual_key(tc->td, block, &key); + if (bio_detain(tc->pool, &key, bio, &virt_cell)) + return DM_MAPIO_SUBMITTED; + r = dm_thin_find_block(td, block, 0, &result); /* @@ -1504,44 +2591,26 @@ * More distant ancestors are irrelevant. The * shared flag will be set in their case. */ - thin_defer_bio(tc, bio); + thin_defer_cell(tc, virt_cell); return DM_MAPIO_SUBMITTED; } - build_virtual_key(tc->td, block, &key); - if (dm_bio_detain(tc->pool->prison, &key, bio, &cell1, &cell_result)) - return DM_MAPIO_SUBMITTED; - build_data_key(tc->td, result.block, &key); - if (dm_bio_detain(tc->pool->prison, &key, bio, &cell2, &cell_result)) { - cell_defer_no_holder_no_free(tc, &cell1); + if (bio_detain(tc->pool, &key, bio, &data_cell)) { + cell_defer_no_holder(tc, virt_cell); return DM_MAPIO_SUBMITTED; } inc_all_io_entry(tc->pool, bio); - cell_defer_no_holder_no_free(tc, &cell2); - cell_defer_no_holder_no_free(tc, &cell1); + cell_defer_no_holder(tc, data_cell); + cell_defer_no_holder(tc, virt_cell); remap(tc, bio, result.block); return DM_MAPIO_REMAPPED; case -ENODATA: - if (get_pool_mode(tc->pool) == PM_READ_ONLY) { - /* - * This block isn't provisioned, and we have no way - * of doing so. Just error it. - */ - bio_io_error(bio); - return DM_MAPIO_SUBMITTED; - } - /* fall through */ - case -EWOULDBLOCK: - /* - * In future, the failed dm_thin_find_block above could - * provide the hint to load the metadata into cache. - */ - thin_defer_bio(tc, bio); + thin_defer_cell(tc, virt_cell); return DM_MAPIO_SUBMITTED; default: @@ -1551,32 +2620,36 @@ * pool is switched to fail-io mode. */ bio_io_error(bio); + cell_defer_no_holder(tc, virt_cell); return DM_MAPIO_SUBMITTED; } } static int pool_is_congested(struct dm_target_callbacks *cb, int bdi_bits) { - int r; - unsigned long flags; struct pool_c *pt = container_of(cb, struct pool_c, callbacks); + struct request_queue *q; - spin_lock_irqsave(&pt->pool->lock, flags); - r = !bio_list_empty(&pt->pool->retry_on_resume_list); - spin_unlock_irqrestore(&pt->pool->lock, flags); - - if (!r) { - struct request_queue *q = bdev_get_queue(pt->data_dev->bdev); - r = bdi_congested(&q->backing_dev_info, bdi_bits); - } + if (get_pool_mode(pt->pool) == PM_OUT_OF_DATA_SPACE) + return 1; - return r; + q = bdev_get_queue(pt->data_dev->bdev); + return bdi_congested(&q->backing_dev_info, bdi_bits); } -static void __requeue_bios(struct pool *pool) +static void requeue_bios(struct pool *pool) { - bio_list_merge(&pool->deferred_bios, &pool->retry_on_resume_list); - bio_list_init(&pool->retry_on_resume_list); + unsigned long flags; + struct thin_c *tc; + + rcu_read_lock(); + list_for_each_entry_rcu(tc, &pool->active_thins, list) { + spin_lock_irqsave(&tc->lock, flags); + bio_list_merge(&tc->deferred_bio_list, &tc->retry_on_resume_list); + bio_list_init(&tc->retry_on_resume_list); + spin_unlock_irqrestore(&tc->lock, flags); + } + rcu_read_unlock(); } /*---------------------------------------------------------------- @@ -1603,7 +2676,6 @@ struct pool *pool = pt->pool; struct block_device *data_bdev = pt->data_dev->bdev; struct queue_limits *data_limits = &bdev_get_queue(data_bdev)->limits; - sector_t block_size = pool->sectors_per_block << SECTOR_SHIFT; const char *reason = NULL; char buf[BDEVNAME_SIZE]; @@ -1616,12 +2688,6 @@ else if (data_limits->max_discard_sectors < pool->sectors_per_block) reason = "max discard sectors smaller than a block"; - else if (data_limits->discard_granularity > block_size) - reason = "discard granularity larger than a block"; - - else if (!is_factor(block_size, data_limits->discard_granularity)) - reason = "discard granularity not a factor of block size"; - if (reason) { DMWARN("Data device (%s) %s: Disabling discard passdown.", bdevname(data_bdev, buf), reason); pt->adjusted_pf.discard_passdown = false; @@ -1633,17 +2699,21 @@ struct pool_c *pt = ti->private; /* - * We want to make sure that degraded pools are never upgraded. + * We want to make sure that a pool in PM_FAIL mode is never upgraded. */ - enum pool_mode old_mode = pool->pf.mode; + enum pool_mode old_mode = get_pool_mode(pool); enum pool_mode new_mode = pt->adjusted_pf.mode; - if (old_mode > new_mode) - new_mode = old_mode; + /* + * Don't change the pool's mode until set_pool_mode() below. + * Otherwise the pool's process_* function pointers may + * not match the desired pool mode. + */ + pt->adjusted_pf.mode = old_mode; pool->ti = ti; - pool->low_water_blocks = pt->low_water_blocks; pool->pf = pt->adjusted_pf; + pool->low_water_blocks = pt->low_water_blocks; set_pool_mode(pool, new_mode); @@ -1666,12 +2736,14 @@ pf->zero_new_blocks = true; pf->discard_enabled = true; pf->discard_passdown = true; + pf->error_if_no_space = false; } static void __pool_destroy(struct pool *pool) { __pool_table_remove(pool); + vfree(pool->cell_sort_array); if (dm_pool_metadata_close(pool->pmd) < 0) DMWARN("%s: dm_pool_metadata_close() failed.", __func__); @@ -1723,7 +2795,7 @@ pool->sectors_per_block_shift = __ffs(block_size); pool->low_water_blocks = 0; pool_features_init(&pool->pf); - pool->prison = dm_bio_prison_create(PRISON_CELLS); + pool->prison = dm_bio_prison_create(); if (!pool->prison) { *error = "Error creating pool's bio prison"; err_p = ERR_PTR(-ENOMEM); @@ -1749,16 +2821,17 @@ goto bad_wq; } + throttle_init(&pool->throttle); INIT_WORK(&pool->worker, do_worker); INIT_DELAYED_WORK(&pool->waker, do_waker); + INIT_DELAYED_WORK(&pool->no_space_timeout, do_no_space_timeout); spin_lock_init(&pool->lock); - bio_list_init(&pool->deferred_bios); bio_list_init(&pool->deferred_flush_bios); INIT_LIST_HEAD(&pool->prepared_mappings); INIT_LIST_HEAD(&pool->prepared_discards); - pool->low_water_triggered = 0; - pool->no_free_space = 0; - bio_list_init(&pool->retry_on_resume_list); + INIT_LIST_HEAD(&pool->active_thins); + pool->low_water_triggered = false; + pool->suspended = true; pool->shared_read_ds = dm_deferred_set_create(); if (!pool->shared_read_ds) { @@ -1783,6 +2856,13 @@ goto bad_mapping_pool; } + pool->cell_sort_array = vmalloc(sizeof(*pool->cell_sort_array) * CELL_SORT_ARRAY_SIZE); + if (!pool->cell_sort_array) { + *error = "Error allocating cell sort array"; + err_p = ERR_PTR(-ENOMEM); + goto bad_sort_array; + } + pool->ref_count = 1; pool->last_commit_jiffies = jiffies; pool->pool_md = pool_md; @@ -1791,6 +2871,8 @@ return pool; +bad_sort_array: + mempool_destroy(pool->mapping_pool); bad_mapping_pool: dm_deferred_set_destroy(pool->all_io_ds); bad_all_io_ds: @@ -1882,7 +2964,7 @@ const char *arg_name; static struct dm_arg _args[] = { - {0, 3, "Invalid number of pool feature arguments"}, + {0, 4, "Invalid number of pool feature arguments"}, }; /* @@ -1911,6 +2993,9 @@ else if (!strcasecmp(arg_name, "read_only")) pf->mode = PM_READ_ONLY; + else if (!strcasecmp(arg_name, "error_if_no_space")) + pf->error_if_no_space = true; + else { ti->error = "Unrecognised pool feature requested"; r = -EINVAL; @@ -1931,16 +3016,27 @@ dm_table_event(pool->ti->table); } -static sector_t get_metadata_dev_size(struct block_device *bdev) +static sector_t get_dev_size(struct block_device *bdev) { - sector_t metadata_dev_size = i_size_read(bdev->bd_inode) >> SECTOR_SHIFT; + return i_size_read(bdev->bd_inode) >> SECTOR_SHIFT; +} + +static void warn_if_metadata_device_too_big(struct block_device *bdev) +{ + sector_t metadata_dev_size = get_dev_size(bdev); char buffer[BDEVNAME_SIZE]; - if (metadata_dev_size > THIN_METADATA_MAX_SECTORS_WARNING) { + if (metadata_dev_size > THIN_METADATA_MAX_SECTORS_WARNING) DMWARN("Metadata device %s is larger than %u sectors: excess space will not be used.", bdevname(bdev, buffer), THIN_METADATA_MAX_SECTORS); - metadata_dev_size = THIN_METADATA_MAX_SECTORS_WARNING; - } +} + +static sector_t get_metadata_dev_size(struct block_device *bdev) +{ + sector_t metadata_dev_size = get_dev_size(bdev); + + if (metadata_dev_size > THIN_METADATA_MAX_SECTORS) + metadata_dev_size = THIN_METADATA_MAX_SECTORS; return metadata_dev_size; } @@ -1949,7 +3045,7 @@ { sector_t metadata_dev_size = get_metadata_dev_size(bdev); - sector_div(metadata_dev_size, THIN_METADATA_BLOCK_SIZE >> SECTOR_SHIFT); + sector_div(metadata_dev_size, THIN_METADATA_BLOCK_SIZE); return metadata_dev_size; } @@ -1981,6 +3077,8 @@ * skip_block_zeroing: skips the zeroing of newly-provisioned blocks. * ignore_discard: disable discard * no_discard_passdown: don't pass discards down to the data device + * read_only: Don't allow any changes to be made to the pool metadata. + * error_if_no_space: error IOs, instead of queueing, if no space. */ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) { @@ -2025,12 +3123,7 @@ ti->error = "Error opening metadata block device"; goto out_unlock; } - - /* - * Run for the side-effect of possibly issuing a warning if the - * device is too big. - */ - (void) get_metadata_dev_size(metadata_dev->bdev); + warn_if_metadata_device_too_big(metadata_dev->bdev); r = dm_get_device(ti, argv[1], FMODE_READ | FMODE_WRITE, &data_dev); if (r) { @@ -2091,6 +3184,7 @@ * them down to the data device. The thin device's discard * processing will cause mappings to be removed from the btree. */ + ti->discard_zeroes_data_unsupported = true; if (pf.discard_enabled && pf.discard_passdown) { ti->num_discard_bios = 1; @@ -2100,7 +3194,6 @@ * thin devices' discard limits consistent). */ ti->discards_supported = true; - ti->discard_zeroes_data_unsupported = true; } ti->private = pt; @@ -2164,20 +3257,31 @@ r = dm_pool_get_data_dev_size(pool->pmd, &sb_data_size); if (r) { - DMERR("failed to retrieve data device size"); + DMERR("%s: failed to retrieve data device size", + dm_device_name(pool->pool_md)); return r; } if (data_size < sb_data_size) { - DMERR("pool target (%llu blocks) too small: expected %llu", + DMERR("%s: pool target (%llu blocks) too small: expected %llu", + dm_device_name(pool->pool_md), (unsigned long long)data_size, sb_data_size); return -EINVAL; } else if (data_size > sb_data_size) { + if (dm_pool_metadata_needs_check(pool->pmd)) { + DMERR("%s: unable to grow the data device until repaired.", + dm_device_name(pool->pool_md)); + return 0; + } + + if (sb_data_size) + DMINFO("%s: growing the data device from %llu to %llu blocks", + dm_device_name(pool->pool_md), + sb_data_size, (unsigned long long)data_size); r = dm_pool_resize_data_dev(pool->pmd, data_size); if (r) { - DMERR("failed to resize data device"); - set_pool_mode(pool, PM_READ_ONLY); + metadata_operation_failed(pool, "dm_pool_resize_data_dev", r); return r; } @@ -2200,19 +3304,31 @@ r = dm_pool_get_metadata_dev_size(pool->pmd, &sb_metadata_dev_size); if (r) { - DMERR("failed to retrieve data device size"); + DMERR("%s: failed to retrieve metadata device size", + dm_device_name(pool->pool_md)); return r; } if (metadata_dev_size < sb_metadata_dev_size) { - DMERR("metadata device (%llu blocks) too small: expected %llu", + DMERR("%s: metadata device (%llu blocks) too small: expected %llu", + dm_device_name(pool->pool_md), metadata_dev_size, sb_metadata_dev_size); return -EINVAL; } else if (metadata_dev_size > sb_metadata_dev_size) { + if (dm_pool_metadata_needs_check(pool->pmd)) { + DMERR("%s: unable to grow the metadata device until repaired.", + dm_device_name(pool->pool_md)); + return 0; + } + + warn_if_metadata_device_too_big(pool->md_dev); + DMINFO("%s: growing the metadata device from %llu to %llu blocks", + dm_device_name(pool->pool_md), + sb_metadata_dev_size, metadata_dev_size); r = dm_pool_resize_metadata_dev(pool->pmd, metadata_dev_size); if (r) { - DMERR("failed to resize metadata device"); + metadata_operation_failed(pool, "dm_pool_resize_metadata_dev", r); return r; } @@ -2256,34 +3372,91 @@ return r; if (need_commit1 || need_commit2) - (void) commit_or_fallback(pool); + (void) commit(pool); return 0; } +static void pool_suspend_active_thins(struct pool *pool) +{ + struct thin_c *tc; + + /* Suspend all active thin devices */ + tc = get_first_thin(pool); + while (tc) { + dm_internal_suspend_noflush(tc->thin_md); + tc = get_next_thin(pool, tc); + } +} + +static void pool_resume_active_thins(struct pool *pool) +{ + struct thin_c *tc; + + /* Resume all active thin devices */ + tc = get_first_thin(pool); + while (tc) { + dm_internal_resume(tc->thin_md); + tc = get_next_thin(pool, tc); + } +} + static void pool_resume(struct dm_target *ti) { struct pool_c *pt = ti->private; struct pool *pool = pt->pool; unsigned long flags; + /* + * Must requeue active_thins' bios and then resume + * active_thins _before_ clearing 'suspend' flag. + */ + requeue_bios(pool); + pool_resume_active_thins(pool); + spin_lock_irqsave(&pool->lock, flags); - pool->low_water_triggered = 0; - pool->no_free_space = 0; - __requeue_bios(pool); + pool->low_water_triggered = false; + pool->suspended = false; spin_unlock_irqrestore(&pool->lock, flags); do_waker(&pool->waker.work); } +static void pool_presuspend(struct dm_target *ti) +{ + struct pool_c *pt = ti->private; + struct pool *pool = pt->pool; + unsigned long flags; + + spin_lock_irqsave(&pool->lock, flags); + pool->suspended = true; + spin_unlock_irqrestore(&pool->lock, flags); + + pool_suspend_active_thins(pool); +} + +static void pool_presuspend_undo(struct dm_target *ti) +{ + struct pool_c *pt = ti->private; + struct pool *pool = pt->pool; + unsigned long flags; + + pool_resume_active_thins(pool); + + spin_lock_irqsave(&pool->lock, flags); + pool->suspended = false; + spin_unlock_irqrestore(&pool->lock, flags); +} + static void pool_postsuspend(struct dm_target *ti) { struct pool_c *pt = ti->private; struct pool *pool = pt->pool; cancel_delayed_work_sync(&pool->waker); + cancel_delayed_work_sync(&pool->no_space_timeout); flush_workqueue(pool->wq); - (void) commit_or_fallback(pool); + (void) commit(pool); } static int check_arg_count(unsigned argc, unsigned args_required) @@ -2417,7 +3590,7 @@ if (r) return r; - (void) commit_or_fallback(pool); + (void) commit(pool); r = dm_pool_reserve_metadata_snap(pool->pmd); if (r) @@ -2446,7 +3619,6 @@ * create_thin * create_snap * delete - * trim * set_transaction_id * reserve_metadata_snap * release_metadata_snap @@ -2460,7 +3632,7 @@ if (get_pool_mode(pool) >= PM_READ_ONLY) { DMERR("%s: unable to service pool target messages in READ_ONLY or FAIL mode", dm_device_name(pool->pool_md)); - return -EINVAL; + return -EOPNOTSUPP; } if (!strcasecmp(argv[0], "create_thin")) @@ -2485,7 +3657,7 @@ DMWARN("Unrecognised thin pool target message received: %s", argv[0]); if (!r) - (void) commit_or_fallback(pool); + (void) commit(pool); return r; } @@ -2494,7 +3666,8 @@ unsigned sz, unsigned maxlen) { unsigned count = !pf->zero_new_blocks + !pf->discard_enabled + - !pf->discard_passdown + (pf->mode == PM_READ_ONLY); + !pf->discard_passdown + (pf->mode == PM_READ_ONLY) + + pf->error_if_no_space; DMEMIT("%u ", count); if (!pf->zero_new_blocks) @@ -2508,12 +3681,16 @@ if (pf->mode == PM_READ_ONLY) DMEMIT("read_only "); + + if (pf->error_if_no_space) + DMEMIT("error_if_no_space "); } /* * Status line is: * / * / + * */ static void pool_status(struct dm_target *ti, status_type_t type, unsigned status_flags, char *result, unsigned maxlen) @@ -2540,41 +3717,47 @@ /* Commit to ensure statistics aren't out-of-date */ if (!(status_flags & DM_STATUS_NOFLUSH_FLAG) && !dm_suspended(ti)) - (void) commit_or_fallback(pool); + (void) commit(pool); r = dm_pool_get_metadata_transaction_id(pool->pmd, &transaction_id); if (r) { - DMERR("dm_pool_get_metadata_transaction_id returned %d", r); + DMERR("%s: dm_pool_get_metadata_transaction_id returned %d", + dm_device_name(pool->pool_md), r); goto err; } r = dm_pool_get_free_metadata_block_count(pool->pmd, &nr_free_blocks_metadata); if (r) { - DMERR("dm_pool_get_free_metadata_block_count returned %d", r); + DMERR("%s: dm_pool_get_free_metadata_block_count returned %d", + dm_device_name(pool->pool_md), r); goto err; } r = dm_pool_get_metadata_dev_size(pool->pmd, &nr_blocks_metadata); if (r) { - DMERR("dm_pool_get_metadata_dev_size returned %d", r); + DMERR("%s: dm_pool_get_metadata_dev_size returned %d", + dm_device_name(pool->pool_md), r); goto err; } r = dm_pool_get_free_block_count(pool->pmd, &nr_free_blocks_data); if (r) { - DMERR("dm_pool_get_free_block_count returned %d", r); + DMERR("%s: dm_pool_get_free_block_count returned %d", + dm_device_name(pool->pool_md), r); goto err; } r = dm_pool_get_data_dev_size(pool->pmd, &nr_blocks_data); if (r) { - DMERR("dm_pool_get_data_dev_size returned %d", r); + DMERR("%s: dm_pool_get_data_dev_size returned %d", + dm_device_name(pool->pool_md), r); goto err; } r = dm_pool_get_metadata_snap(pool->pmd, &held_root); if (r) { - DMERR("dm_pool_get_metadata_snap returned %d", r); + DMERR("%s: dm_pool_get_metadata_snap returned %d", + dm_device_name(pool->pool_md), r); goto err; } @@ -2590,17 +3773,29 @@ else DMEMIT("- "); - if (pool->pf.mode == PM_READ_ONLY) + if (pool->pf.mode == PM_OUT_OF_DATA_SPACE) + DMEMIT("out_of_data_space "); + else if (pool->pf.mode == PM_READ_ONLY) DMEMIT("ro "); else DMEMIT("rw "); if (!pool->pf.discard_enabled) - DMEMIT("ignore_discard"); + DMEMIT("ignore_discard "); else if (pool->pf.discard_passdown) - DMEMIT("discard_passdown"); + DMEMIT("discard_passdown "); + else + DMEMIT("no_discard_passdown "); + + if (pool->pf.error_if_no_space) + DMEMIT("error_if_no_space "); + else + DMEMIT("queue_if_no_space "); + + if (dm_pool_metadata_needs_check(pool->pmd)) + DMEMIT("needs_check "); else - DMEMIT("no_discard_passdown"); + DMEMIT("- "); break; @@ -2627,74 +3822,82 @@ return fn(ti, pt->data_dev, 0, ti->len, data); } -static int pool_merge(struct dm_target *ti, struct bvec_merge_data *bvm, - struct bio_vec *biovec, int max_size) +static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits) { struct pool_c *pt = ti->private; - struct request_queue *q = bdev_get_queue(pt->data_dev->bdev); - - if (!q->merge_bvec_fn) - return max_size; - - bvm->bi_bdev = pt->data_dev->bdev; - - return min(max_size, q->merge_bvec_fn(q, bvm, biovec)); -} - -static void set_discard_limits(struct pool_c *pt, struct queue_limits *limits) -{ struct pool *pool = pt->pool; - struct queue_limits *data_limits; - - limits->max_discard_sectors = pool->sectors_per_block; + sector_t io_opt_sectors = limits->io_opt >> SECTOR_SHIFT; /* - * discard_granularity is just a hint, and not enforced. + * If max_sectors is smaller than pool->sectors_per_block adjust it + * to the highest possible power-of-2 factor of pool->sectors_per_block. + * This is especially beneficial when the pool's data device is a RAID + * device that has a full stripe width that matches pool->sectors_per_block + * -- because even though partial RAID stripe-sized IOs will be issued to a + * single RAID stripe; when aggregated they will end on a full RAID stripe + * boundary.. which avoids additional partial RAID stripe writes cascading */ - if (pt->adjusted_pf.discard_passdown) { - data_limits = &bdev_get_queue(pt->data_dev->bdev)->limits; - limits->discard_granularity = max(data_limits->discard_granularity, - pool->sectors_per_block << SECTOR_SHIFT); - } else - limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT; -} - -static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits) -{ - struct pool_c *pt = ti->private; - struct pool *pool = pt->pool; + if (limits->max_sectors < pool->sectors_per_block) { + while (!is_factor(pool->sectors_per_block, limits->max_sectors)) { + if ((limits->max_sectors & (limits->max_sectors - 1)) == 0) + limits->max_sectors--; + limits->max_sectors = rounddown_pow_of_two(limits->max_sectors); + } + } - blk_limits_io_min(limits, 0); - blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT); + /* + * If the system-determined stacked limits are compatible with the + * pool's blocksize (io_opt is a factor) do not override them. + */ + if (io_opt_sectors < pool->sectors_per_block || + !is_factor(io_opt_sectors, pool->sectors_per_block)) { + if (is_factor(pool->sectors_per_block, limits->max_sectors)) + blk_limits_io_min(limits, limits->max_sectors << SECTOR_SHIFT); + else + blk_limits_io_min(limits, pool->sectors_per_block << SECTOR_SHIFT); + blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT); + } /* * pt->adjusted_pf is a staging area for the actual features to use. * They get transferred to the live pool in bind_control_target() * called from pool_preresume(). */ - if (!pt->adjusted_pf.discard_enabled) + if (!pt->adjusted_pf.discard_enabled) { + /* + * Must explicitly disallow stacking discard limits otherwise the + * block layer will stack them if pool's data device has support. + * QUEUE_FLAG_DISCARD wouldn't be set but there is no way for the + * user to see that, so make sure to set all discard limits to 0. + */ + limits->discard_granularity = 0; return; + } disable_passdown_if_not_supported(pt); - set_discard_limits(pt, limits); + /* + * The pool uses the same discard limits as the underlying data + * device. DM core has already set this up. + */ } static struct target_type pool_target = { .name = "thin-pool", .features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE | DM_TARGET_IMMUTABLE, - .version = {1, 8, 0}, + .version = {1, 16, 0}, .module = THIS_MODULE, .ctr = pool_ctr, .dtr = pool_dtr, .map = pool_map, + .presuspend = pool_presuspend, + .presuspend_undo = pool_presuspend_undo, .postsuspend = pool_postsuspend, .preresume = pool_preresume, .resume = pool_resume, .message = pool_message, .status = pool_status, - .merge = pool_merge, .iterate_devices = pool_iterate_devices, .io_hints = pool_io_hints, }; @@ -2702,9 +3905,29 @@ /*---------------------------------------------------------------- * Thin target methods *--------------------------------------------------------------*/ +static void thin_get(struct thin_c *tc) +{ + atomic_inc(&tc->refcount); +} + +static void thin_put(struct thin_c *tc) +{ + if (atomic_dec_and_test(&tc->refcount)) + complete(&tc->can_destroy); +} + static void thin_dtr(struct dm_target *ti) { struct thin_c *tc = ti->private; + unsigned long flags; + + spin_lock_irqsave(&tc->pool->lock, flags); + list_del_rcu(&tc->list); + spin_unlock_irqrestore(&tc->pool->lock, flags); + synchronize_rcu(); + + thin_put(tc); + wait_for_completion(&tc->can_destroy); mutex_lock(&dm_thin_pool_table.mutex); @@ -2736,6 +3959,7 @@ struct thin_c *tc; struct dm_dev *pool_dev, *origin_dev; struct mapped_device *pool_md; + unsigned long flags; mutex_lock(&dm_thin_pool_table.mutex); @@ -2751,6 +3975,12 @@ r = -ENOMEM; goto out_unlock; } + tc->thin_md = dm_table_get_md(ti->table); + spin_lock_init(&tc->lock); + INIT_LIST_HEAD(&tc->deferred_cells); + bio_list_init(&tc->deferred_bio_list); + bio_list_init(&tc->retry_on_resume_list); + tc->sort_bio_list = RB_ROOT; if (argc == 3) { r = dm_get_device(ti, argv[2], FMODE_READ, &origin_dev); @@ -2792,41 +4022,60 @@ if (get_pool_mode(tc->pool) == PM_FAIL) { ti->error = "Couldn't open thin device, Pool is in fail mode"; r = -EINVAL; - goto bad_thin_open; + goto bad_pool; } r = dm_pool_open_thin_device(tc->pool->pmd, tc->dev_id, &tc->td); if (r) { ti->error = "Couldn't open thin internal device"; - goto bad_thin_open; + goto bad_pool; } r = dm_set_target_max_io_len(ti, tc->pool->sectors_per_block); if (r) - goto bad_target_max_io_len; + goto bad; ti->num_flush_bios = 1; ti->flush_supported = true; ti->per_bio_data_size = sizeof(struct dm_thin_endio_hook); /* In case the pool supports discards, pass them on. */ + ti->discard_zeroes_data_unsupported = true; if (tc->pool->pf.discard_enabled) { ti->discards_supported = true; ti->num_discard_bios = 1; - ti->discard_zeroes_data_unsupported = true; - /* Discard bios must be split on a block boundary */ - ti->split_discard_bios = true; + ti->split_discard_bios = false; } - dm_put(pool_md); - mutex_unlock(&dm_thin_pool_table.mutex); + spin_lock_irqsave(&tc->pool->lock, flags); + if (tc->pool->suspended) { + spin_unlock_irqrestore(&tc->pool->lock, flags); + mutex_lock(&dm_thin_pool_table.mutex); /* reacquire for __pool_dec */ + ti->error = "Unable to activate thin device while pool is suspended"; + r = -EINVAL; + goto bad; + } + atomic_set(&tc->refcount, 1); + init_completion(&tc->can_destroy); + list_add_tail_rcu(&tc->list, &tc->pool->active_thins); + spin_unlock_irqrestore(&tc->pool->lock, flags); + /* + * This synchronize_rcu() call is needed here otherwise we risk a + * wake_worker() call finding no bios to process (because the newly + * added tc isn't yet visible). So this reduces latency since we + * aren't then dependent on the periodic commit to wake_worker(). + */ + synchronize_rcu(); + + dm_put(pool_md); + return 0; -bad_target_max_io_len: +bad: dm_pool_close_thin_device(tc->td); -bad_thin_open: +bad_pool: __pool_dec(tc->pool); bad_pool_lookup: dm_put(pool_md); @@ -2845,7 +4094,7 @@ static int thin_map(struct dm_target *ti, struct bio *bio) { - bio->bi_sector = dm_target_offset(ti, bio->bi_sector); + bio->bi_iter.bi_sector = dm_target_offset(ti, bio->bi_iter.bi_sector); return thin_bio_map(ti, bio); } @@ -2865,8 +4114,7 @@ spin_lock_irqsave(&pool->lock, flags); list_for_each_entry_safe(m, tmp, &work, list) { list_del(&m->list); - m->quiesced = 1; - __maybe_add_mapping(m); + __complete_mapping_preparation(m); } spin_unlock_irqrestore(&pool->lock, flags); } @@ -2877,19 +4125,45 @@ if (!list_empty(&work)) { spin_lock_irqsave(&pool->lock, flags); list_for_each_entry_safe(m, tmp, &work, list) - list_add(&m->list, &pool->prepared_discards); + list_add_tail(&m->list, &pool->prepared_discards); spin_unlock_irqrestore(&pool->lock, flags); wake_worker(pool); } } + if (h->cell) + cell_defer_no_holder(h->tc, h->cell); + return 0; } -static void thin_postsuspend(struct dm_target *ti) +static void thin_presuspend(struct dm_target *ti) { + struct thin_c *tc = ti->private; + if (dm_noflush_suspending(ti)) - requeue_io((struct thin_c *)ti->private); + noflush_work(tc, do_noflush_start); +} + +static void thin_postsuspend(struct dm_target *ti) +{ + struct thin_c *tc = ti->private; + + /* + * The dm_noflush_suspending flag has been cleared by now, so + * unfortunately we must always run this. + */ + noflush_work(tc, do_noflush_stop); +} + +static int thin_preresume(struct dm_target *ti) +{ + struct thin_c *tc = ti->private; + + if (tc->origin_dev) + tc->origin_size = get_dev_size(tc->origin_dev->bdev); + + return 0; } /* @@ -2972,17 +4246,32 @@ return 0; } +static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits) +{ + struct thin_c *tc = ti->private; + struct pool *pool = tc->pool; + + if (!pool->pf.discard_enabled) + return; + + limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT; + limits->max_discard_sectors = 2048 * 1024 * 16; /* 16G */ +} + static struct target_type thin_target = { .name = "thin", - .version = {1, 8, 0}, + .version = {1, 16, 0}, .module = THIS_MODULE, .ctr = thin_ctr, .dtr = thin_dtr, .map = thin_map, .end_io = thin_endio, + .preresume = thin_preresume, + .presuspend = thin_presuspend, .postsuspend = thin_postsuspend, .status = thin_status, .iterate_devices = thin_iterate_devices, + .io_hints = thin_io_hints, }; /*----------------------------------------------------------------*/ @@ -3028,6 +4317,9 @@ module_init(dm_thin_init); module_exit(dm_thin_exit); +module_param_named(no_space_timeout, no_space_timeout_secs, uint, S_IRUGO | S_IWUSR); +MODULE_PARM_DESC(no_space_timeout, "Out of data space queue IO timeout in seconds"); + MODULE_DESCRIPTION(DM_NAME " thin provisioning target"); MODULE_AUTHOR("Joe Thornber "); MODULE_LICENSE("GPL");