diff options
Diffstat (limited to 'drivers/md/dm-thin.c')
| -rw-r--r-- | drivers/md/dm-thin.c | 2492 |
1 files changed, 1591 insertions, 901 deletions
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index ce59824fb41..fc9c848a60c 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -1,29 +1,38 @@ /* - * Copyright (C) 2011 Red Hat UK. + * Copyright (C) 2011-2012 Red Hat UK. * * This file is released under the GPL. */ #include "dm-thin-metadata.h" +#include "dm-bio-prison.h" +#include "dm.h" #include <linux/device-mapper.h> #include <linux/dm-io.h> #include <linux/dm-kcopyd.h> #include <linux/list.h> +#include <linux/rculist.h> #include <linux/init.h> #include <linux/module.h> #include <linux/slab.h> +#include <linux/rbtree.h> #define DM_MSG_PREFIX "thin" /* * Tunable constants */ -#define ENDIO_HOOK_POOL_SIZE 10240 -#define DEFERRED_SET_SIZE 64 +#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"); /* * The block size of the device holding pool data must be @@ -57,7 +66,7 @@ * i) plug io further to this physical block. (see bio_prison code). * * ii) quiesce any read io to that shared data block. Obviously - * including all devices that share this block. (see deferred_set code) + * including all devices that share this block. (see dm_deferred_set code) * * iii) copy the data block to a newly allocate block. This step can be * missed out if the io covers the block. (schedule_copy). @@ -98,381 +107,10 @@ /*----------------------------------------------------------------*/ /* - * Sometimes we can't deal with a bio straight away. We put them in prison - * where they can't cause any mischief. Bios are put in a cell identified - * by a key, multiple bios can be in the same cell. When the cell is - * subsequently unlocked the bios become available. - */ -struct bio_prison; - -struct cell_key { - int virtual; - dm_thin_id dev; - dm_block_t block; -}; - -struct dm_bio_prison_cell { - struct hlist_node list; - struct bio_prison *prison; - struct cell_key key; - struct bio *holder; - struct bio_list bios; -}; - -struct bio_prison { - spinlock_t lock; - mempool_t *cell_pool; - - unsigned nr_buckets; - unsigned hash_mask; - struct hlist_head *cells; -}; - -static uint32_t calc_nr_buckets(unsigned nr_cells) -{ - uint32_t n = 128; - - nr_cells /= 4; - nr_cells = min(nr_cells, 8192u); - - while (n < nr_cells) - n <<= 1; - - return n; -} - -static struct kmem_cache *_cell_cache; - -/* - * @nr_cells should be the number of cells you want in use _concurrently_. - * Don't confuse it with the number of distinct keys. - */ -static struct bio_prison *prison_create(unsigned nr_cells) -{ - unsigned i; - uint32_t nr_buckets = calc_nr_buckets(nr_cells); - size_t len = sizeof(struct bio_prison) + - (sizeof(struct hlist_head) * nr_buckets); - struct bio_prison *prison = kmalloc(len, GFP_KERNEL); - - if (!prison) - return NULL; - - spin_lock_init(&prison->lock); - prison->cell_pool = mempool_create_slab_pool(nr_cells, _cell_cache); - if (!prison->cell_pool) { - kfree(prison); - return NULL; - } - - prison->nr_buckets = nr_buckets; - prison->hash_mask = nr_buckets - 1; - prison->cells = (struct hlist_head *) (prison + 1); - for (i = 0; i < nr_buckets; i++) - INIT_HLIST_HEAD(prison->cells + i); - - return prison; -} - -static void prison_destroy(struct bio_prison *prison) -{ - mempool_destroy(prison->cell_pool); - kfree(prison); -} - -static uint32_t hash_key(struct bio_prison *prison, struct cell_key *key) -{ - const unsigned long BIG_PRIME = 4294967291UL; - uint64_t hash = key->block * BIG_PRIME; - - return (uint32_t) (hash & prison->hash_mask); -} - -static int keys_equal(struct cell_key *lhs, struct cell_key *rhs) -{ - return (lhs->virtual == rhs->virtual) && - (lhs->dev == rhs->dev) && - (lhs->block == rhs->block); -} - -static struct dm_bio_prison_cell *__search_bucket(struct hlist_head *bucket, - struct cell_key *key) -{ - struct dm_bio_prison_cell *cell; - struct hlist_node *tmp; - - hlist_for_each_entry(cell, tmp, bucket, list) - if (keys_equal(&cell->key, key)) - return cell; - - return NULL; -} - -/* - * This may block if a new cell needs allocating. You must ensure that - * cells will be unlocked even if the calling thread is blocked. - * - * Returns 1 if the cell was already held, 0 if @inmate is the new holder. - */ -static int bio_detain(struct bio_prison *prison, struct cell_key *key, - struct bio *inmate, struct dm_bio_prison_cell **ref) -{ - int r = 1; - unsigned long flags; - uint32_t hash = hash_key(prison, key); - struct dm_bio_prison_cell *cell, *cell2; - - BUG_ON(hash > prison->nr_buckets); - - spin_lock_irqsave(&prison->lock, flags); - - cell = __search_bucket(prison->cells + hash, key); - if (cell) { - bio_list_add(&cell->bios, inmate); - goto out; - } - - /* - * Allocate a new cell - */ - spin_unlock_irqrestore(&prison->lock, flags); - cell2 = mempool_alloc(prison->cell_pool, GFP_NOIO); - spin_lock_irqsave(&prison->lock, flags); - - /* - * We've been unlocked, so we have to double check that - * nobody else has inserted this cell in the meantime. - */ - cell = __search_bucket(prison->cells + hash, key); - if (cell) { - mempool_free(cell2, prison->cell_pool); - bio_list_add(&cell->bios, inmate); - goto out; - } - - /* - * Use new cell. - */ - cell = cell2; - - cell->prison = prison; - memcpy(&cell->key, key, sizeof(cell->key)); - cell->holder = inmate; - bio_list_init(&cell->bios); - hlist_add_head(&cell->list, prison->cells + hash); - - r = 0; - -out: - spin_unlock_irqrestore(&prison->lock, flags); - - *ref = cell; - - return r; -} - -/* - * @inmates must have been initialised prior to this call - */ -static void __cell_release(struct dm_bio_prison_cell *cell, struct bio_list *inmates) -{ - struct bio_prison *prison = cell->prison; - - hlist_del(&cell->list); - - if (inmates) { - bio_list_add(inmates, cell->holder); - bio_list_merge(inmates, &cell->bios); - } - - mempool_free(cell, prison->cell_pool); -} - -static void cell_release(struct dm_bio_prison_cell *cell, struct bio_list *bios) -{ - unsigned long flags; - struct bio_prison *prison = cell->prison; - - spin_lock_irqsave(&prison->lock, flags); - __cell_release(cell, bios); - spin_unlock_irqrestore(&prison->lock, flags); -} - -/* - * There are a couple of places where we put a bio into a cell briefly - * before taking it out again. In these situations we know that no other - * bio may be in the cell. This function releases the cell, and also does - * a sanity check. - */ -static void __cell_release_singleton(struct dm_bio_prison_cell *cell, struct bio *bio) -{ - BUG_ON(cell->holder != bio); - BUG_ON(!bio_list_empty(&cell->bios)); - - __cell_release(cell, NULL); -} - -static void cell_release_singleton(struct dm_bio_prison_cell *cell, struct bio *bio) -{ - unsigned long flags; - struct bio_prison *prison = cell->prison; - - spin_lock_irqsave(&prison->lock, flags); - __cell_release_singleton(cell, bio); - spin_unlock_irqrestore(&prison->lock, flags); -} - -/* - * Sometimes we don't want the holder, just the additional bios. - */ -static void __cell_release_no_holder(struct dm_bio_prison_cell *cell, - struct bio_list *inmates) -{ - struct bio_prison *prison = cell->prison; - - hlist_del(&cell->list); - bio_list_merge(inmates, &cell->bios); - - mempool_free(cell, prison->cell_pool); -} - -static void cell_release_no_holder(struct dm_bio_prison_cell *cell, - struct bio_list *inmates) -{ - unsigned long flags; - struct bio_prison *prison = cell->prison; - - spin_lock_irqsave(&prison->lock, flags); - __cell_release_no_holder(cell, inmates); - spin_unlock_irqrestore(&prison->lock, flags); -} - -static void cell_error(struct dm_bio_prison_cell *cell) -{ - struct bio_prison *prison = cell->prison; - struct bio_list bios; - struct bio *bio; - unsigned long flags; - - bio_list_init(&bios); - - spin_lock_irqsave(&prison->lock, flags); - __cell_release(cell, &bios); - spin_unlock_irqrestore(&prison->lock, flags); - - while ((bio = bio_list_pop(&bios))) - bio_io_error(bio); -} - -/*----------------------------------------------------------------*/ - -/* - * We use the deferred set to keep track of pending reads to shared blocks. - * We do this to ensure the new mapping caused by a write isn't performed - * until these prior reads have completed. Otherwise the insertion of the - * new mapping could free the old block that the read bios are mapped to. - */ - -struct deferred_set; -struct deferred_entry { - struct deferred_set *ds; - unsigned count; - struct list_head work_items; -}; - -struct deferred_set { - spinlock_t lock; - unsigned current_entry; - unsigned sweeper; - struct deferred_entry entries[DEFERRED_SET_SIZE]; -}; - -static void ds_init(struct deferred_set *ds) -{ - int i; - - spin_lock_init(&ds->lock); - ds->current_entry = 0; - ds->sweeper = 0; - for (i = 0; i < DEFERRED_SET_SIZE; i++) { - ds->entries[i].ds = ds; - ds->entries[i].count = 0; - INIT_LIST_HEAD(&ds->entries[i].work_items); - } -} - -static struct deferred_entry *ds_inc(struct deferred_set *ds) -{ - unsigned long flags; - struct deferred_entry *entry; - - spin_lock_irqsave(&ds->lock, flags); - entry = ds->entries + ds->current_entry; - entry->count++; - spin_unlock_irqrestore(&ds->lock, flags); - - return entry; -} - -static unsigned ds_next(unsigned index) -{ - return (index + 1) % DEFERRED_SET_SIZE; -} - -static void __sweep(struct deferred_set *ds, struct list_head *head) -{ - while ((ds->sweeper != ds->current_entry) && - !ds->entries[ds->sweeper].count) { - list_splice_init(&ds->entries[ds->sweeper].work_items, head); - ds->sweeper = ds_next(ds->sweeper); - } - - if ((ds->sweeper == ds->current_entry) && !ds->entries[ds->sweeper].count) - list_splice_init(&ds->entries[ds->sweeper].work_items, head); -} - -static void ds_dec(struct deferred_entry *entry, struct list_head *head) -{ - unsigned long flags; - - spin_lock_irqsave(&entry->ds->lock, flags); - BUG_ON(!entry->count); - --entry->count; - __sweep(entry->ds, head); - spin_unlock_irqrestore(&entry->ds->lock, flags); -} - -/* - * Returns 1 if deferred or 0 if no pending items to delay job. - */ -static int ds_add_work(struct deferred_set *ds, struct list_head *work) -{ - int r = 1; - unsigned long flags; - unsigned next_entry; - - spin_lock_irqsave(&ds->lock, flags); - if ((ds->sweeper == ds->current_entry) && - !ds->entries[ds->current_entry].count) - r = 0; - else { - list_add(work, &ds->entries[ds->current_entry].work_items); - next_entry = ds_next(ds->current_entry); - if (!ds->entries[next_entry].count) - ds->current_entry = next_entry; - } - spin_unlock_irqrestore(&ds->lock, flags); - - return r; -} - -/*----------------------------------------------------------------*/ - -/* * Key building. */ static void build_data_key(struct dm_thin_device *td, - dm_block_t b, struct cell_key *key) + dm_block_t b, struct dm_cell_key *key) { key->virtual = 0; key->dev = dm_thin_dev_id(td); @@ -480,7 +118,7 @@ static void build_data_key(struct dm_thin_device *td, } static void build_virtual_key(struct dm_thin_device *td, dm_block_t b, - struct cell_key *key) + struct dm_cell_key *key) { key->virtual = 1; key->dev = dm_thin_dev_id(td); @@ -496,12 +134,29 @@ static void build_virtual_key(struct dm_thin_device *td, dm_block_t b, */ struct dm_thin_new_mapping; +/* + * 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 */ +}; + struct pool_features { - unsigned zero_new_blocks:1; - unsigned discard_enabled:1; - unsigned discard_passdown:1; + enum pool_mode mode; + + 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_mapping_fn)(struct dm_thin_new_mapping *m); + struct pool { struct list_head list; struct dm_target *ti; /* Only set if a pool target is bound */ @@ -510,41 +165,46 @@ struct pool { struct block_device *md_dev; struct dm_pool_metadata *pmd; - uint32_t sectors_per_block; - unsigned block_shift; - dm_block_t offset_mask; dm_block_t low_water_blocks; + uint32_t sectors_per_block; + 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 */ - struct bio_prison *prison; + struct dm_bio_prison *prison; struct dm_kcopyd_client *copier; struct workqueue_struct *wq; struct work_struct worker; struct delayed_work waker; + struct delayed_work no_space_timeout; - unsigned ref_count; 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 list_head active_thins; - struct bio_list retry_on_resume_list; - - struct deferred_set shared_read_ds; - struct deferred_set all_io_ds; + struct dm_deferred_set *shared_read_ds; + struct dm_deferred_set *all_io_ds; struct dm_thin_new_mapping *next_mapping; mempool_t *mapping_pool; - mempool_t *endio_hook_pool; + + process_bio_fn process_bio; + process_bio_fn process_discard; + + process_mapping_fn process_prepared_mapping; + process_mapping_fn process_prepared_discard; }; +static enum pool_mode get_pool_mode(struct pool *pool); +static void metadata_operation_failed(struct pool *pool, const char *op, int r); + /* * Target context for a pool. */ @@ -556,24 +216,115 @@ struct pool_c { struct dm_target_callbacks callbacks; dm_block_t low_water_blocks; - struct pool_features pf; + struct pool_features requested_pf; /* Features requested during table load */ + struct pool_features adjusted_pf; /* Features used after adjusting for constituent devices */ }; /* * Target context for a thin. */ struct thin_c { + struct list_head list; struct dm_dev *pool_dev; struct dm_dev *origin_dev; dm_thin_id dev_id; struct pool *pool; struct dm_thin_device *td; + bool requeue_mode:1; + spinlock_t lock; + 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; }; /*----------------------------------------------------------------*/ /* + * wake_worker() is used when new work is queued and when pool_resume is + * ready to continue deferred IO processing. + */ +static void wake_worker(struct pool *pool) +{ + queue_work(pool->wq, &pool->worker); +} + +/*----------------------------------------------------------------*/ + +static int bio_detain(struct pool *pool, struct dm_cell_key *key, struct bio *bio, + struct dm_bio_prison_cell **cell_result) +{ + int r; + struct dm_bio_prison_cell *cell_prealloc; + + /* + * Allocate a cell from the prison's mempool. + * This might block but it can't fail. + */ + cell_prealloc = dm_bio_prison_alloc_cell(pool->prison, GFP_NOIO); + + r = dm_bio_detain(pool->prison, key, bio, cell_prealloc, cell_result); + if (r) + /* + * We reused an old cell; we can get rid of + * the new one. + */ + dm_bio_prison_free_cell(pool->prison, cell_prealloc); + + return r; +} + +static void cell_release(struct pool *pool, + struct dm_bio_prison_cell *cell, + struct bio_list *bios) +{ + dm_cell_release(pool->prison, cell, bios); + 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) +{ + dm_cell_release_no_holder(pool->prison, cell, bios); + 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) +{ + struct pool *pool = tc->pool; + unsigned long flags; + + spin_lock_irqsave(&tc->lock, flags); + dm_cell_release_no_holder(pool->prison, cell, &tc->deferred_bio_list); + spin_unlock_irqrestore(&tc->lock, flags); + + wake_worker(pool); +} + +static void cell_error_with_code(struct pool *pool, + struct dm_bio_prison_cell *cell, int error_code) +{ + dm_cell_error(pool->prison, cell, error_code); + dm_bio_prison_free_cell(pool->prison, cell); +} + +static void cell_error(struct pool *pool, struct dm_bio_prison_cell *cell) +{ + cell_error_with_code(pool, cell, -EIO); +} + +/*----------------------------------------------------------------*/ + +/* * A global list of pools that uses a struct mapped_device as a key. */ static struct dm_thin_pool_table { @@ -635,39 +386,60 @@ static struct pool *__pool_table_lookup_metadata_dev(struct block_device *md_dev struct dm_thin_endio_hook { struct thin_c *tc; - struct deferred_entry *shared_read_entry; - struct deferred_entry *all_io_entry; + 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; }; -static void __requeue_bio_list(struct thin_c *tc, struct bio_list *master) +static void requeue_bio_list(struct thin_c *tc, struct bio_list *master) { struct bio *bio; struct bio_list bios; + unsigned long flags; bio_list_init(&bios); + + spin_lock_irqsave(&tc->lock, flags); bio_list_merge(&bios, master); bio_list_init(master); + spin_unlock_irqrestore(&tc->lock, flags); - while ((bio = bio_list_pop(&bios))) { - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; - - if (h->tc == tc) - bio_endio(bio, DM_ENDIO_REQUEUE); - else - bio_list_add(master, bio); - } + while ((bio = bio_list_pop(&bios))) + bio_endio(bio, DM_ENDIO_REQUEUE); } static void requeue_io(struct thin_c *tc) { - struct pool *pool = tc->pool; + requeue_bio_list(tc, &tc->deferred_bio_list); + requeue_bio_list(tc, &tc->retry_on_resume_list); +} + +static void error_thin_retry_list(struct thin_c *tc) +{ + struct bio *bio; unsigned long flags; + struct bio_list bios; - 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); + bio_list_init(&bios); + + spin_lock_irqsave(&tc->lock, flags); + bio_list_merge(&bios, &tc->retry_on_resume_list); + bio_list_init(&tc->retry_on_resume_list); + spin_unlock_irqrestore(&tc->lock, flags); + + while ((bio = bio_list_pop(&bios))) + bio_io_error(bio); +} + +static void error_retry_list(struct pool *pool) +{ + struct thin_c *tc; + + rcu_read_lock(); + list_for_each_entry_rcu(tc, &pool->active_thins, list) + error_thin_retry_list(tc); + rcu_read_unlock(); } /* @@ -677,18 +449,37 @@ static void requeue_io(struct thin_c *tc) * 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) { - return bio->bi_sector >> tc->pool->block_shift; + struct pool *pool = tc->pool; + sector_t block_nr = bio->bi_iter.bi_sector; + + if (block_size_is_power_of_two(pool)) + block_nr >>= pool->sectors_per_block_shift; + else + (void) sector_div(block_nr, pool->sectors_per_block); + + return block_nr; } 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_iter.bi_sector; bio->bi_bdev = tc->pool_dev->bdev; - bio->bi_sector = (block << pool->block_shift) + - (bio->bi_sector & pool->offset_mask); + if (block_size_is_power_of_two(pool)) + bio->bi_iter.bi_sector = + (block << pool->sectors_per_block_shift) | + (bi_sector & (pool->sectors_per_block - 1)); + else + bio->bi_iter.bi_sector = (block * pool->sectors_per_block) + + sector_div(bi_sector, pool->sectors_per_block); } static void remap_to_origin(struct thin_c *tc, struct bio *bio) @@ -696,21 +487,50 @@ static void remap_to_origin(struct thin_c *tc, struct bio *bio) bio->bi_bdev = tc->origin_dev->bdev; } +static int bio_triggers_commit(struct thin_c *tc, struct bio *bio) +{ + return (bio->bi_rw & (REQ_FLUSH | REQ_FUA)) && + dm_thin_changed_this_transaction(tc->td); +} + +static void inc_all_io_entry(struct pool *pool, struct bio *bio) +{ + struct dm_thin_endio_hook *h; + + if (bio->bi_rw & REQ_DISCARD) + return; + + h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); + h->all_io_entry = dm_deferred_entry_inc(pool->all_io_ds); +} + static void issue(struct thin_c *tc, struct bio *bio) { struct pool *pool = tc->pool; unsigned long flags; + if (!bio_triggers_commit(tc, bio)) { + generic_make_request(bio); + return; + } + /* - * Batch together any FUA/FLUSH bios we find and then issue - * a single commit for them in process_deferred_bios(). + * Complete bio with an error if earlier I/O caused changes to + * the metadata that can't be committed e.g, due to I/O errors + * on the metadata device. */ - if (bio->bi_rw & (REQ_FLUSH | REQ_FUA)) { - spin_lock_irqsave(&pool->lock, flags); - bio_list_add(&pool->deferred_flush_bios, bio); - spin_unlock_irqrestore(&pool->lock, flags); - } else - generic_make_request(bio); + if (dm_thin_aborted_changes(tc->td)) { + bio_io_error(bio); + return; + } + + /* + * Batch together any bios that trigger commits and then issue a + * single commit for them in process_deferred_bios(). + */ + spin_lock_irqsave(&pool->lock, flags); + bio_list_add(&pool->deferred_flush_bios, bio); + spin_unlock_irqrestore(&pool->lock, flags); } static void remap_to_origin_and_issue(struct thin_c *tc, struct bio *bio) @@ -726,15 +546,6 @@ static void remap_and_issue(struct thin_c *tc, struct bio *bio, issue(tc, bio); } -/* - * wake_worker() is used when new work is queued and when pool_resume is - * ready to continue deferred IO processing. - */ -static void wake_worker(struct pool *pool) -{ - queue_work(pool->wq, &pool->worker); -} - /*----------------------------------------------------------------*/ /* @@ -743,15 +554,16 @@ static void wake_worker(struct pool *pool) struct dm_thin_new_mapping { struct list_head list; - unsigned quiesced:1; - unsigned prepared:1; - unsigned pass_discard:1; + bool quiesced:1; + bool prepared:1; + bool pass_discard:1; + bool definitely_not_shared:1; + int err; struct thin_c *tc; dm_block_t virt_block; dm_block_t data_block; struct dm_bio_prison_cell *cell, *cell2; - int err; /* * If the bio covers the whole area of a block then we can avoid @@ -768,7 +580,7 @@ static void __maybe_add_mapping(struct dm_thin_new_mapping *m) struct pool *pool = m->tc->pool; if (m->quiesced && m->prepared) { - list_add(&m->list, &pool->prepared_mappings); + list_add_tail(&m->list, &pool->prepared_mappings); wake_worker(pool); } } @@ -782,7 +594,7 @@ static void copy_complete(int read_err, unsigned long write_err, void *context) m->err = read_err || write_err ? -EIO : 0; spin_lock_irqsave(&pool->lock, flags); - m->prepared = 1; + m->prepared = true; __maybe_add_mapping(m); spin_unlock_irqrestore(&pool->lock, flags); } @@ -790,14 +602,14 @@ static void copy_complete(int read_err, unsigned long write_err, void *context) static void overwrite_endio(struct bio *bio, int err) { unsigned long flags; - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; + 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; spin_lock_irqsave(&pool->lock, flags); - m->prepared = 1; + m->prepared = true; __maybe_add_mapping(m); spin_unlock_irqrestore(&pool->lock, flags); } @@ -815,51 +627,60 @@ static void overwrite_endio(struct bio *bio, int err) /* * This sends the bios in the cell back to the deferred_bios list. */ -static void cell_defer(struct thin_c *tc, struct dm_bio_prison_cell *cell, - dm_block_t data_block) +static void cell_defer(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(cell, &pool->deferred_bios); - spin_unlock_irqrestore(&tc->pool->lock, flags); + spin_lock_irqsave(&tc->lock, flags); + cell_release(pool, cell, &tc->deferred_bio_list); + spin_unlock_irqrestore(&tc->lock, flags); wake_worker(pool); } /* - * Same as cell_defer above, except it omits one particular detainee, - * a write bio that covers the block and has already been processed. + * Same as cell_defer above, except it omits the original holder of the cell. */ -static void cell_defer_except(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 bio_list bios; struct pool *pool = tc->pool; unsigned long flags; - bio_list_init(&bios); - - spin_lock_irqsave(&pool->lock, flags); - cell_release_no_holder(cell, &pool->deferred_bios); - spin_unlock_irqrestore(&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); } +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; + atomic_inc(&m->bio->bi_remaining); + } + cell_error(m->tc->pool, m->cell); + list_del(&m->list); + mempool_free(m, m->tc->pool->mapping_pool); +} + static void process_prepared_mapping(struct dm_thin_new_mapping *m) { struct thin_c *tc = m->tc; + struct pool *pool = tc->pool; struct bio *bio; int r; bio = m->bio; - if (bio) + if (bio) { bio->bi_end_io = m->saved_bi_end_io; + atomic_inc(&bio->bi_remaining); + } if (m->err) { - cell_error(m->cell); - return; + cell_error(pool, m->cell); + goto out; } /* @@ -869,9 +690,9 @@ static void process_prepared_mapping(struct dm_thin_new_mapping *m) */ r = dm_thin_insert_block(tc->td, m->virt_block, m->data_block); if (r) { - DMERR("dm_thin_insert_block() failed"); - cell_error(m->cell); - return; + metadata_operation_failed(pool, "dm_thin_insert_block", r); + cell_error(pool, m->cell); + goto out; } /* @@ -881,39 +702,64 @@ static void process_prepared_mapping(struct dm_thin_new_mapping *m) * the bios in the cell. */ if (bio) { - cell_defer_except(tc, m->cell); + cell_defer_no_holder(tc, m->cell); bio_endio(bio, 0); } else - cell_defer(tc, m->cell, m->data_block); + cell_defer(tc, m->cell); +out: list_del(&m->list); + mempool_free(m, pool->mapping_pool); +} + +static void process_prepared_discard_fail(struct dm_thin_new_mapping *m) +{ + struct thin_c *tc = m->tc; + + bio_io_error(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(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; - r = dm_thin_remove_block(tc->td, m->virt_block); - if (r) - DMERR("dm_thin_remove_block() failed"); + inc_all_io_entry(tc->pool, m->bio); + cell_defer_no_holder(tc, m->cell); + cell_defer_no_holder(tc, m->cell2); - /* - * Pass the discard down to the underlying device? - */ if (m->pass_discard) - remap_and_issue(tc, m->bio, m->data_block); + 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); + } else bio_endio(m->bio, 0); - cell_defer_except(tc, m->cell); - cell_defer_except(tc, m->cell2); mempool_free(m, tc->pool->mapping_pool); } +static void process_prepared_discard(struct dm_thin_new_mapping *m) +{ + int r; + struct thin_c *tc = m->tc; + + r = dm_thin_remove_block(tc->td, m->virt_block); + if (r) + DMERR_LIMIT("dm_thin_remove_block() failed"); + + process_prepared_discard_passdown(m); +} + static void process_prepared(struct pool *pool, struct list_head *head, - void (*fn)(struct dm_thin_new_mapping *)) + process_mapping_fn *fn) { unsigned long flags; struct list_head maps; @@ -925,7 +771,7 @@ static void process_prepared(struct pool *pool, struct list_head *head, spin_unlock_irqrestore(&pool->lock, flags); list_for_each_entry_safe(m, tmp, &maps, list) - fn(m); + (*fn)(m); } /* @@ -933,9 +779,8 @@ static void process_prepared(struct pool *pool, struct list_head *head, */ static int io_overlaps_block(struct pool *pool, struct bio *bio) { - return !(bio->bi_sector & pool->offset_mask) && - (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) @@ -963,13 +808,17 @@ static int ensure_next_mapping(struct pool *pool) static struct dm_thin_new_mapping *get_next_mapping(struct pool *pool) { - struct dm_thin_new_mapping *r = pool->next_mapping; + struct dm_thin_new_mapping *m = pool->next_mapping; BUG_ON(!pool->next_mapping); + memset(m, 0, sizeof(struct dm_thin_new_mapping)); + INIT_LIST_HEAD(&m->list); + m->bio = NULL; + pool->next_mapping = NULL; - return r; + return m; } static void schedule_copy(struct thin_c *tc, dm_block_t virt_block, @@ -981,18 +830,13 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block, struct pool *pool = tc->pool; struct dm_thin_new_mapping *m = get_next_mapping(pool); - INIT_LIST_HEAD(&m->list); - m->quiesced = 0; - m->prepared = 0; m->tc = tc; m->virt_block = virt_block; m->data_block = data_dest; m->cell = cell; - m->err = 0; - m->bio = NULL; - if (!ds_add_work(&pool->shared_read_ds, &m->list)) - m->quiesced = 1; + if (!dm_deferred_set_add_work(pool->shared_read_ds, &m->list)) + m->quiesced = true; /* * IO to pool_dev remaps to the pool target's data_dev. @@ -1001,11 +845,12 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block, * bio immediately. Otherwise we use kcopyd to clone the data first. */ if (io_overwrites_block(pool, bio)) { - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; + 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 { struct dm_io_region from, to; @@ -1022,8 +867,8 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block, 0, copy_complete, m); if (r < 0) { mempool_free(m, pool->mapping_pool); - DMERR("dm_kcopyd_copy() failed"); - cell_error(cell); + DMERR_LIMIT("dm_kcopyd_copy() failed"); + cell_error(pool, cell); } } } @@ -1051,15 +896,12 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block, struct pool *pool = tc->pool; struct dm_thin_new_mapping *m = get_next_mapping(pool); - INIT_LIST_HEAD(&m->list); - m->quiesced = 1; - m->prepared = 0; + m->quiesced = true; + m->prepared = false; m->tc = tc; m->virt_block = virt_block; m->data_block = data_block; m->cell = cell; - m->err = 0; - m->bio = NULL; /* * If the whole block of data is being overwritten or we are not @@ -1070,11 +912,12 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block, process_prepared_mapping(m); else if (io_overwrites_block(pool, bio)) { - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; + 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; @@ -1087,69 +930,89 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block, r = dm_kcopyd_zero(pool->copier, 1, &to, 0, copy_complete, m); if (r < 0) { mempool_free(m, pool->mapping_pool); - DMERR("dm_kcopyd_zero() failed"); - cell_error(cell); + DMERR_LIMIT("dm_kcopyd_zero() failed"); + cell_error(pool, cell); } } } -static int alloc_data_block(struct thin_c *tc, dm_block_t *result) +/* + * A non-zero return indicates read_only or fail_io mode. + * Many callers don't care about the return value. + */ +static int commit(struct pool *pool) { 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 (get_pool_mode(pool) >= PM_READ_ONLY) + return -EINVAL; + + r = dm_pool_commit_metadata(pool->pmd); if (r) - return r; + metadata_operation_failed(pool, "dm_pool_commit_metadata", r); + + return r; +} + +static void check_low_water_mark(struct pool *pool, dm_block_t free_blocks) +{ + unsigned long flags; if (free_blocks <= pool->low_water_blocks && !pool->low_water_triggered) { - DMWARN("%s: reached low water mark, sending event.", + 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 void set_pool_mode(struct pool *pool, enum pool_mode new_mode); + +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. - */ - r = dm_pool_commit_metadata(pool->pmd); - if (r) { - DMERR("%s: dm_pool_commit_metadata() failed, error = %d", - __func__, r); - return r; - } + /* + * 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; } @@ -1160,26 +1023,70 @@ static int alloc_data_block(struct thin_c *tc, dm_block_t *result) */ static void retry_on_resume(struct bio *bio) { - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; + 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_endio(bio, error); + else + retry_on_resume(bio); } -static void no_space(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(cell, &bios); + cell_release(pool, cell, &bios); - while ((bio = bio_list_pop(&bios))) - retry_on_resume(bio); + error = should_error_unserviceable_bio(pool); + if (error) + while ((bio = bio_list_pop(&bios))) + bio_endio(bio, error); + else + while ((bio = bio_list_pop(&bios))) + retry_on_resume(bio); } static void process_discard(struct thin_c *tc, struct bio *bio) @@ -1188,13 +1095,13 @@ static void process_discard(struct thin_c *tc, struct bio *bio) unsigned long flags; struct pool *pool = tc->pool; struct dm_bio_prison_cell *cell, *cell2; - struct cell_key key, key2; + 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; build_virtual_key(tc->td, block, &key); - if (bio_detain(tc->pool->prison, &key, bio, &cell)) + if (bio_detain(tc->pool, &key, bio, &cell)) return; r = dm_thin_find_block(tc->td, block, 1, &lookup_result); @@ -1206,8 +1113,8 @@ static void process_discard(struct thin_c *tc, struct bio *bio) * on this block. */ build_data_key(tc->td, lookup_result.block, &key2); - if (bio_detain(tc->pool->prison, &key2, bio, &cell2)) { - cell_release_singleton(cell, bio); + if (bio_detain(tc->pool, &key2, bio, &cell2)) { + cell_defer_no_holder(tc, cell); break; } @@ -1218,34 +1125,34 @@ static void process_discard(struct thin_c *tc, struct bio *bio) */ m = get_next_mapping(pool); m->tc = tc; - m->pass_discard = (!lookup_result.shared) & pool->pf.discard_passdown; + 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->err = 0; m->bio = bio; - if (!ds_add_work(&pool->all_io_ds, &m->list)) { + 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); + list_add_tail(&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); + /* - * This path is hit if people are ignoring - * limits->discard_granularity. It ignores any - * part of the discard that is in a subsequent - * block. + * The DM core makes sure that the discard doesn't span + * a block boundary. So we submit the discard of a + * partial block appropriately. */ - sector_t offset = bio->bi_sector - (block << pool->block_shift); - unsigned remaining = (pool->sectors_per_block - offset) << 9; - bio->bi_size = min(bio->bi_size, remaining); - - cell_release_singleton(cell, bio); - cell_release_singleton(cell2, bio); - remap_and_issue(tc, bio, lookup_result.block); + if ((!lookup_result.shared) && pool->pf.discard_passdown) + remap_and_issue(tc, bio, lookup_result.block); + else + bio_endio(bio, 0); } break; @@ -1253,25 +1160,27 @@ static void process_discard(struct thin_c *tc, struct bio *bio) /* * It isn't provisioned, just forget it. */ - cell_release_singleton(cell, bio); + cell_defer_no_holder(tc, cell); bio_endio(bio, 0); break; default: - DMERR("discard: find block unexpectedly returned %d", r); - cell_release_singleton(cell, bio); + DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d", + __func__, r); + cell_defer_no_holder(tc, cell); bio_io_error(bio); break; } } static void break_sharing(struct thin_c *tc, struct bio *bio, dm_block_t block, - struct cell_key *key, + struct dm_cell_key *key, struct dm_thin_lookup_result *lookup_result, struct dm_bio_prison_cell *cell) { int r; dm_block_t data_block; + struct pool *pool = tc->pool; r = alloc_data_block(tc, &data_block); switch (r) { @@ -1281,12 +1190,13 @@ static void break_sharing(struct thin_c *tc, struct bio *bio, dm_block_t block, break; case -ENOSPC: - no_space(cell); + retry_bios_on_resume(pool, cell); break; default: - DMERR("%s: alloc_data_block() failed, error = %d", __func__, r); - cell_error(cell); + DMERR_LIMIT("%s: alloc_data_block() failed: error = %d", + __func__, r); + cell_error(pool, cell); break; } } @@ -1297,24 +1207,25 @@ static void process_shared_bio(struct thin_c *tc, struct bio *bio, { struct dm_bio_prison_cell *cell; struct pool *pool = tc->pool; - struct cell_key key; + struct dm_cell_key key; /* * If cell is already occupied, then sharing is already in the process * of being broken so we have nothing further to do here. */ build_data_key(tc->td, lookup_result->block, &key); - if (bio_detain(pool->prison, &key, bio, &cell)) + if (bio_detain(pool, &key, bio, &cell)) return; - if (bio_data_dir(bio) == WRITE) + if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) break_sharing(tc, bio, block, &key, lookup_result, cell); else { - struct dm_thin_endio_hook *h = dm_get_mapinfo(bio)->ptr; + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); - h->shared_read_entry = ds_inc(&pool->shared_read_ds); + h->shared_read_entry = dm_deferred_entry_inc(pool->shared_read_ds); + inc_all_io_entry(pool, bio); + cell_defer_no_holder(tc, cell); - cell_release_singleton(cell, bio); remap_and_issue(tc, bio, lookup_result->block); } } @@ -1324,12 +1235,15 @@ static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block { int r; dm_block_t data_block; + struct pool *pool = tc->pool; /* * Remap empty bios (flushes) immediately, without provisioning. */ - if (!bio->bi_size) { - cell_release_singleton(cell, bio); + if (!bio->bi_iter.bi_size) { + inc_all_io_entry(pool, bio); + cell_defer_no_holder(tc, cell); + remap_and_issue(tc, bio, 0); return; } @@ -1339,7 +1253,7 @@ static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block */ if (bio_data_dir(bio) == READ) { zero_fill_bio(bio); - cell_release_singleton(cell, bio); + cell_defer_no_holder(tc, cell); bio_endio(bio, 0); return; } @@ -1354,12 +1268,13 @@ static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block break; case -ENOSPC: - no_space(cell); + retry_bios_on_resume(pool, cell); break; default: - DMERR("%s: alloc_data_block() failed, error = %d", __func__, r); - cell_error(cell); + DMERR_LIMIT("%s: alloc_data_block() failed: error = %d", + __func__, r); + cell_error(pool, cell); break; } } @@ -1367,9 +1282,10 @@ static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block static void process_bio(struct thin_c *tc, struct bio *bio) { int r; + struct pool *pool = tc->pool; dm_block_t block = get_bio_block(tc, bio); struct dm_bio_prison_cell *cell; - struct cell_key key; + struct dm_cell_key key; struct dm_thin_lookup_result lookup_result; /* @@ -1377,86 +1293,270 @@ static void process_bio(struct thin_c *tc, struct bio *bio) * being provisioned so we have nothing further to do here. */ build_virtual_key(tc->td, block, &key); - if (bio_detain(tc->pool->prison, &key, bio, &cell)) + if (bio_detain(pool, &key, bio, &cell)) return; r = dm_thin_find_block(tc->td, block, 1, &lookup_result); switch (r) { case 0: - /* - * We can release this cell now. This thread is the only - * one that puts bios into a cell, and we know there were - * no preceding bios. - */ - /* - * TODO: this will probably have to change when discard goes - * back in. - */ - cell_release_singleton(cell, bio); - - if (lookup_result.shared) + if (lookup_result.shared) { process_shared_bio(tc, bio, block, &lookup_result); - else + cell_defer_no_holder(tc, cell); /* FIXME: pass this cell into process_shared? */ + } else { + inc_all_io_entry(pool, bio); + cell_defer_no_holder(tc, cell); + remap_and_issue(tc, bio, lookup_result.block); + } break; case -ENODATA: if (bio_data_dir(bio) == READ && tc->origin_dev) { - cell_release_singleton(cell, bio); + inc_all_io_entry(pool, bio); + cell_defer_no_holder(tc, cell); + remap_to_origin_and_issue(tc, bio); } else provision_block(tc, bio, block, cell); break; default: - DMERR("dm_thin_find_block() failed, error = %d", r); - cell_release_singleton(cell, bio); + DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d", + __func__, r); + cell_defer_no_holder(tc, cell); + bio_io_error(bio); + break; + } +} + +static void process_bio_read_only(struct thin_c *tc, struct bio *bio) +{ + int r; + int rw = bio_data_dir(bio); + dm_block_t block = get_bio_block(tc, bio); + struct dm_thin_lookup_result lookup_result; + + r = dm_thin_find_block(tc->td, block, 1, &lookup_result); + switch (r) { + case 0: + if (lookup_result.shared && (rw == WRITE) && bio->bi_iter.bi_size) + handle_unserviceable_bio(tc->pool, bio); + else { + inc_all_io_entry(tc->pool, bio); + remap_and_issue(tc, bio, lookup_result.block); + } + break; + + case -ENODATA: + if (rw != READ) { + handle_unserviceable_bio(tc->pool, bio); + break; + } + + if (tc->origin_dev) { + inc_all_io_entry(tc->pool, bio); + remap_to_origin_and_issue(tc, bio); + break; + } + + zero_fill_bio(bio); + bio_endio(bio, 0); + break; + + default: + DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d", + __func__, r); bio_io_error(bio); break; } } +static void process_bio_success(struct thin_c *tc, struct bio *bio) +{ + bio_endio(bio, 0); +} + +static void process_bio_fail(struct thin_c *tc, struct bio *bio) +{ + bio_io_error(bio); +} + +/* + * 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; } -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; - int r; + struct blk_plug plug; + + if (tc->requeue_mode) { + requeue_bio_list(tc, &tc->deferred_bio_list); + 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_get_mapinfo(bio)->ptr; - 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_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; } if (bio->bi_rw & REQ_DISCARD) - process_discard(tc, bio); + pool->process_discard(tc, bio); else - process_bio(tc, bio); + pool->process_bio(tc, bio); + } + blk_finish_plug(&plug); +} + +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_bios(tc); + tc = get_next_thin(pool, tc); } /* @@ -1469,13 +1569,11 @@ static void process_deferred_bios(struct pool *pool) bio_list_init(&pool->deferred_flush_bios); spin_unlock_irqrestore(&pool->lock, flags); - if (bio_list_empty(&bios) && !need_commit_due_to_time(pool)) + if (bio_list_empty(&bios) && + !(dm_pool_changed_this_transaction(pool->pmd) && need_commit_due_to_time(pool))) return; - r = dm_pool_commit_metadata(pool->pmd); - if (r) { - DMERR("%s: dm_pool_commit_metadata() failed, error = %d", - __func__, r); + if (commit(pool)) { while ((bio = bio_list_pop(&bios))) bio_io_error(bio); return; @@ -1490,8 +1588,8 @@ static void do_worker(struct work_struct *ws) { struct pool *pool = container_of(ws, struct pool, worker); - process_prepared(pool, &pool->prepared_mappings, process_prepared_mapping); - process_prepared(pool, &pool->prepared_discards, process_prepared_discard); + process_prepared(pool, &pool->prepared_mappings, &pool->process_prepared_mapping); + process_prepared(pool, &pool->prepared_discards, &pool->process_prepared_discard); process_deferred_bios(pool); } @@ -1506,6 +1604,211 @@ static void do_waker(struct work_struct *ws) queue_delayed_work(pool->wq, &pool->waker, COMMIT_PERIOD); } +/* + * 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_READ_ONLY and start erroring IO. + */ +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) + set_pool_mode(pool, PM_READ_ONLY); +} + +/*----------------------------------------------------------------*/ + +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) +{ + return pool->pf.mode; +} + +static void notify_of_pool_mode_change(struct pool *pool, const char *new_mode) +{ + dm_table_event(pool->ti->table); + DMINFO("%s: switching pool to %s mode", + dm_device_name(pool->pool_md), new_mode); +} + +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: + 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_prepared_mapping = process_prepared_mapping_fail; + pool->process_prepared_discard = process_prepared_discard_fail; + + error_retry_list(pool); + break; + + case PM_READ_ONLY: + 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_prepared_mapping = process_prepared_mapping_fail; + pool->process_prepared_discard = process_prepared_discard_passdown; + + 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(pool, "out-of-data-space"); + pool->process_bio = process_bio_read_only; + pool->process_discard = process_discard; + pool->process_prepared_mapping = process_prepared_mapping; + pool->process_prepared_discard = process_prepared_discard_passdown; + + 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"); + dm_pool_metadata_read_write(pool->pmd); + pool->process_bio = process_bio; + pool->process_discard = process_discard; + pool->process_prepared_mapping = process_prepared_mapping; + pool->process_prepared_discard = process_prepared_discard; + 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); +} + /*----------------------------------------------------------------*/ /* @@ -1520,39 +1823,49 @@ static void thin_defer_bio(struct thin_c *tc, struct bio *bio) 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 struct dm_thin_endio_hook *thin_hook_bio(struct thin_c *tc, struct bio *bio) +static void thin_hook_bio(struct thin_c *tc, struct bio *bio) { - struct pool *pool = tc->pool; - struct dm_thin_endio_hook *h = mempool_alloc(pool->endio_hook_pool, GFP_NOIO); + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); h->tc = tc; h->shared_read_entry = NULL; - h->all_io_entry = bio->bi_rw & REQ_DISCARD ? NULL : ds_inc(&pool->all_io_ds); + h->all_io_entry = NULL; h->overwrite_mapping = NULL; - - return h; } /* * Non-blocking function called from the thin target's map function. */ -static int thin_bio_map(struct dm_target *ti, struct bio *bio, - union map_info *map_context) +static int thin_bio_map(struct dm_target *ti, struct bio *bio) { int r; struct thin_c *tc = ti->private; 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_cell_key key; + + thin_hook_bio(tc, bio); + + if (tc->requeue_mode) { + bio_endio(bio, DM_ENDIO_REQUEUE); + return DM_MAPIO_SUBMITTED; + } + + if (get_pool_mode(tc->pool) == PM_FAIL) { + bio_io_error(bio); + return DM_MAPIO_SUBMITTED; + } - map_context->ptr = thin_hook_bio(tc, bio); if (bio->bi_rw & (REQ_DISCARD | REQ_FLUSH | REQ_FUA)) { thin_defer_bio(tc, bio); return DM_MAPIO_SUBMITTED; @@ -1581,76 +1894,154 @@ static int thin_bio_map(struct dm_target *ti, struct bio *bio, * shared flag will be set in their case. */ thin_defer_bio(tc, bio); - r = DM_MAPIO_SUBMITTED; - } else { - remap(tc, bio, result.block); - r = DM_MAPIO_REMAPPED; + 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); + return DM_MAPIO_SUBMITTED; } - break; + + inc_all_io_entry(tc->pool, bio); + cell_defer_no_holder_no_free(tc, &cell2); + cell_defer_no_holder_no_free(tc, &cell1); + + 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. + */ + handle_unserviceable_bio(tc->pool, 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. */ - case -EWOULDBLOCK: thin_defer_bio(tc, bio); - r = DM_MAPIO_SUBMITTED; - break; - } + return DM_MAPIO_SUBMITTED; - return r; + default: + /* + * Must always call bio_io_error on failure. + * dm_thin_find_block can fail with -EINVAL if the + * pool is switched to fail-io mode. + */ + bio_io_error(bio); + 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 (get_pool_mode(pt->pool) == PM_OUT_OF_DATA_SPACE) + return 1; - if (!r) { - struct request_queue *q = bdev_get_queue(pt->data_dev->bdev); - r = bdi_congested(&q->backing_dev_info, bdi_bits); - } - - 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(); } /*---------------------------------------------------------------- * Binding of control targets to a pool object *--------------------------------------------------------------*/ +static bool data_dev_supports_discard(struct pool_c *pt) +{ + struct request_queue *q = bdev_get_queue(pt->data_dev->bdev); + + return q && blk_queue_discard(q); +} + +static bool is_factor(sector_t block_size, uint32_t n) +{ + return !sector_div(block_size, n); +} + +/* + * If discard_passdown was enabled verify that the data device + * supports discards. Disable discard_passdown if not. + */ +static void disable_passdown_if_not_supported(struct pool_c *pt) +{ + 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]; + + if (!pt->adjusted_pf.discard_passdown) + return; + + if (!data_dev_supports_discard(pt)) + reason = "discard unsupported"; + + 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; + } +} + static int bind_control_target(struct pool *pool, struct dm_target *ti) { struct pool_c *pt = ti->private; - pool->ti = ti; - pool->low_water_blocks = pt->low_water_blocks; - pool->pf = pt->pf; + /* + * We want to make sure that a pool in PM_FAIL mode is never upgraded. + */ + enum pool_mode old_mode = get_pool_mode(pool); + enum pool_mode new_mode = pt->adjusted_pf.mode; /* - * If discard_passdown was enabled verify that the data device - * supports discards. Disable discard_passdown if not; otherwise - * -EOPNOTSUPP will be returned. + * 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. */ - if (pt->pf.discard_passdown) { - struct request_queue *q = bdev_get_queue(pt->data_dev->bdev); - if (!q || !blk_queue_discard(q)) { - char buf[BDEVNAME_SIZE]; - DMWARN("Discard unsupported by data device (%s): Disabling discard passdown.", - bdevname(pt->data_dev->bdev, buf)); - pool->pf.discard_passdown = 0; - } - } + pt->adjusted_pf.mode = old_mode; + + pool->ti = ti; + pool->pf = pt->adjusted_pf; + pool->low_water_blocks = pt->low_water_blocks; + + set_pool_mode(pool, new_mode); return 0; } @@ -1667,9 +2058,11 @@ static void unbind_control_target(struct pool *pool, struct dm_target *ti) /* Initialize pool features. */ static void pool_features_init(struct pool_features *pf) { - pf->zero_new_blocks = 1; - pf->discard_enabled = 1; - pf->discard_passdown = 1; + pf->mode = PM_WRITE; + 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) @@ -1679,7 +2072,7 @@ static void __pool_destroy(struct pool *pool) if (dm_pool_metadata_close(pool->pmd) < 0) DMWARN("%s: dm_pool_metadata_close() failed.", __func__); - prison_destroy(pool->prison); + dm_bio_prison_destroy(pool->prison); dm_kcopyd_client_destroy(pool->copier); if (pool->wq) @@ -1688,23 +2081,25 @@ static void __pool_destroy(struct pool *pool) if (pool->next_mapping) mempool_free(pool->next_mapping, pool->mapping_pool); mempool_destroy(pool->mapping_pool); - mempool_destroy(pool->endio_hook_pool); + dm_deferred_set_destroy(pool->shared_read_ds); + dm_deferred_set_destroy(pool->all_io_ds); kfree(pool); } static struct kmem_cache *_new_mapping_cache; -static struct kmem_cache *_endio_hook_cache; static struct pool *pool_create(struct mapped_device *pool_md, struct block_device *metadata_dev, - unsigned long block_size, char **error) + unsigned long block_size, + int read_only, char **error) { int r; void *err_p; struct pool *pool; struct dm_pool_metadata *pmd; + bool format_device = read_only ? false : true; - pmd = dm_pool_metadata_open(metadata_dev, block_size); + pmd = dm_pool_metadata_open(metadata_dev, block_size, format_device); if (IS_ERR(pmd)) { *error = "Error creating metadata object"; return (struct pool *)pmd; @@ -1719,18 +2114,20 @@ static struct pool *pool_create(struct mapped_device *pool_md, pool->pmd = pmd; pool->sectors_per_block = block_size; - pool->block_shift = ffs(block_size) - 1; - pool->offset_mask = block_size - 1; + if (block_size & (block_size - 1)) + pool->sectors_per_block_shift = -1; + else + pool->sectors_per_block_shift = __ffs(block_size); pool->low_water_blocks = 0; pool_features_init(&pool->pf); - pool->prison = prison_create(PRISON_CELLS); + pool->prison = dm_bio_prison_create(PRISON_CELLS); if (!pool->prison) { *error = "Error creating pool's bio prison"; err_p = ERR_PTR(-ENOMEM); goto bad_prison; } - pool->copier = dm_kcopyd_client_create(); + pool->copier = dm_kcopyd_client_create(&dm_kcopyd_throttle); if (IS_ERR(pool->copier)) { r = PTR_ERR(pool->copier); *error = "Error creating pool's kcopyd client"; @@ -1751,16 +2148,27 @@ static struct pool *pool_create(struct mapped_device *pool_md, 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); - ds_init(&pool->shared_read_ds); - ds_init(&pool->all_io_ds); + INIT_LIST_HEAD(&pool->active_thins); + pool->low_water_triggered = false; + + pool->shared_read_ds = dm_deferred_set_create(); + if (!pool->shared_read_ds) { + *error = "Error creating pool's shared read deferred set"; + err_p = ERR_PTR(-ENOMEM); + goto bad_shared_read_ds; + } + + pool->all_io_ds = dm_deferred_set_create(); + if (!pool->all_io_ds) { + *error = "Error creating pool's all io deferred set"; + err_p = ERR_PTR(-ENOMEM); + goto bad_all_io_ds; + } pool->next_mapping = NULL; pool->mapping_pool = mempool_create_slab_pool(MAPPING_POOL_SIZE, @@ -1771,13 +2179,6 @@ static struct pool *pool_create(struct mapped_device *pool_md, goto bad_mapping_pool; } - pool->endio_hook_pool = mempool_create_slab_pool(ENDIO_HOOK_POOL_SIZE, - _endio_hook_cache); - if (!pool->endio_hook_pool) { - *error = "Error creating pool's endio_hook mempool"; - err_p = ERR_PTR(-ENOMEM); - goto bad_endio_hook_pool; - } pool->ref_count = 1; pool->last_commit_jiffies = jiffies; pool->pool_md = pool_md; @@ -1786,14 +2187,16 @@ static struct pool *pool_create(struct mapped_device *pool_md, return pool; -bad_endio_hook_pool: - mempool_destroy(pool->mapping_pool); bad_mapping_pool: + dm_deferred_set_destroy(pool->all_io_ds); +bad_all_io_ds: + dm_deferred_set_destroy(pool->shared_read_ds); +bad_shared_read_ds: destroy_workqueue(pool->wq); bad_wq: dm_kcopyd_client_destroy(pool->copier); bad_kcopyd_client: - prison_destroy(pool->prison); + dm_bio_prison_destroy(pool->prison); bad_prison: kfree(pool); bad_pool: @@ -1819,25 +2222,29 @@ static void __pool_dec(struct pool *pool) static struct pool *__pool_find(struct mapped_device *pool_md, struct block_device *metadata_dev, - unsigned long block_size, char **error, - int *created) + unsigned long block_size, int read_only, + char **error, int *created) { struct pool *pool = __pool_table_lookup_metadata_dev(metadata_dev); if (pool) { - if (pool->pool_md != pool_md) + if (pool->pool_md != pool_md) { + *error = "metadata device already in use by a pool"; return ERR_PTR(-EBUSY); + } __pool_inc(pool); } else { pool = __pool_table_lookup(pool_md); if (pool) { - if (pool->md_dev != metadata_dev) + if (pool->md_dev != metadata_dev) { + *error = "different pool cannot replace a pool"; return ERR_PTR(-EINVAL); + } __pool_inc(pool); } else { - pool = pool_create(pool_md, metadata_dev, block_size, error); + pool = pool_create(pool_md, metadata_dev, block_size, read_only, error); *created = 1; } } @@ -1871,7 +2278,7 @@ static int parse_pool_features(struct dm_arg_set *as, struct pool_features *pf, 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"}, }; /* @@ -1888,24 +2295,92 @@ static int parse_pool_features(struct dm_arg_set *as, struct pool_features *pf, arg_name = dm_shift_arg(as); argc--; - if (!strcasecmp(arg_name, "skip_block_zeroing")) { - pf->zero_new_blocks = 0; - continue; - } else if (!strcasecmp(arg_name, "ignore_discard")) { - pf->discard_enabled = 0; - continue; - } else if (!strcasecmp(arg_name, "no_discard_passdown")) { - pf->discard_passdown = 0; - continue; - } + if (!strcasecmp(arg_name, "skip_block_zeroing")) + pf->zero_new_blocks = false; - ti->error = "Unrecognised pool feature requested"; - r = -EINVAL; + else if (!strcasecmp(arg_name, "ignore_discard")) + pf->discard_enabled = false; + + else if (!strcasecmp(arg_name, "no_discard_passdown")) + pf->discard_passdown = false; + + 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; + break; + } } return r; } +static void metadata_low_callback(void *context) +{ + struct pool *pool = context; + + DMWARN("%s: reached low water mark for metadata device: sending event.", + dm_device_name(pool->pool_md)); + + dm_table_event(pool->ti->table); +} + +static sector_t get_dev_size(struct block_device *bdev) +{ + 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) + DMWARN("Metadata device %s is larger than %u sectors: excess space will not be used.", + bdevname(bdev, buffer), THIN_METADATA_MAX_SECTORS); +} + +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; +} + +static dm_block_t get_metadata_dev_size_in_blocks(struct block_device *bdev) +{ + sector_t metadata_dev_size = get_metadata_dev_size(bdev); + + sector_div(metadata_dev_size, THIN_METADATA_BLOCK_SIZE); + + return metadata_dev_size; +} + +/* + * When a metadata threshold is crossed a dm event is triggered, and + * userland should respond by growing the metadata device. We could let + * userland set the threshold, like we do with the data threshold, but I'm + * not sure they know enough to do this well. + */ +static dm_block_t calc_metadata_threshold(struct pool_c *pt) +{ + /* + * 4M is ample for all ops with the possible exception of thin + * device deletion which is harmless if it fails (just retry the + * delete after you've grown the device). + */ + dm_block_t quarter = get_metadata_dev_size_in_blocks(pt->metadata_dev->bdev) / 4; + return min((dm_block_t)1024ULL /* 4M */, quarter); +} + /* * thin-pool <metadata dev> <data dev> * <data block size (sectors)> @@ -1916,6 +2391,8 @@ static int parse_pool_features(struct dm_arg_set *as, struct pool_features *pf, * 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) { @@ -1928,8 +2405,7 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) unsigned long block_size; dm_block_t low_water_blocks; struct dm_dev *metadata_dev; - sector_t metadata_dev_size; - char b[BDEVNAME_SIZE]; + fmode_t metadata_mode; /* * FIXME Remove validation from scope of lock. @@ -1941,19 +2417,27 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) r = -EINVAL; goto out_unlock; } + as.argc = argc; as.argv = argv; - r = dm_get_device(ti, argv[0], FMODE_READ | FMODE_WRITE, &metadata_dev); + /* + * Set default pool features. + */ + pool_features_init(&pf); + + dm_consume_args(&as, 4); + r = parse_pool_features(&as, &pf, ti); + if (r) + goto out_unlock; + + metadata_mode = FMODE_READ | ((pf.mode == PM_READ_ONLY) ? 0 : FMODE_WRITE); + r = dm_get_device(ti, argv[0], metadata_mode, &metadata_dev); if (r) { ti->error = "Error opening metadata block device"; goto out_unlock; } - - metadata_dev_size = i_size_read(metadata_dev->bdev->bd_inode) >> SECTOR_SHIFT; - 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(metadata_dev->bdev, b), THIN_METADATA_MAX_SECTORS); + warn_if_metadata_device_too_big(metadata_dev->bdev); r = dm_get_device(ti, argv[1], FMODE_READ | FMODE_WRITE, &data_dev); if (r) { @@ -1964,7 +2448,7 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) if (kstrtoul(argv[2], 10, &block_size) || !block_size || block_size < DATA_DEV_BLOCK_SIZE_MIN_SECTORS || block_size > DATA_DEV_BLOCK_SIZE_MAX_SECTORS || - !is_power_of_2(block_size)) { + block_size & (DATA_DEV_BLOCK_SIZE_MIN_SECTORS - 1)) { ti->error = "Invalid block size"; r = -EINVAL; goto out; @@ -1976,16 +2460,6 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) goto out; } - /* - * Set default pool features. - */ - pool_features_init(&pf); - - dm_consume_args(&as, 4); - r = parse_pool_features(&as, &pf, ti); - if (r) - goto out; - pt = kzalloc(sizeof(*pt), GFP_KERNEL); if (!pt) { r = -ENOMEM; @@ -1993,7 +2467,7 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) } pool = __pool_find(dm_table_get_md(ti->table), metadata_dev->bdev, - block_size, &ti->error, &pool_created); + block_size, pf.mode == PM_READ_ONLY, &ti->error, &pool_created); if (IS_ERR(pool)) { r = PTR_ERR(pool); goto out_free_pt; @@ -2016,24 +2490,34 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv) pt->metadata_dev = metadata_dev; pt->data_dev = data_dev; pt->low_water_blocks = low_water_blocks; - pt->pf = pf; - ti->num_flush_requests = 1; + pt->adjusted_pf = pt->requested_pf = pf; + ti->num_flush_bios = 1; + /* * Only need to enable discards if the pool should pass * 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_requests = 1; + ti->num_discard_bios = 1; + /* * Setting 'discards_supported' circumvents the normal * stacking of discard limits (this keeps the pool and * thin devices' discard limits consistent). */ - ti->discards_supported = 1; + ti->discards_supported = true; } ti->private = pt; + r = dm_pool_register_metadata_threshold(pt->pool->pmd, + calc_metadata_threshold(pt), + metadata_low_callback, + pool); + if (r) + goto out_free_pt; + pt->callbacks.congested_fn = pool_is_congested; dm_table_add_target_callbacks(ti->table, &pt->callbacks); @@ -2055,8 +2539,7 @@ out_unlock: return r; } -static int pool_map(struct dm_target *ti, struct bio *bio, - union map_info *map_context) +static int pool_map(struct dm_target *ti, struct bio *bio) { int r; struct pool_c *pt = ti->private; @@ -2074,6 +2557,101 @@ static int pool_map(struct dm_target *ti, struct bio *bio, return r; } +static int maybe_resize_data_dev(struct dm_target *ti, bool *need_commit) +{ + int r; + struct pool_c *pt = ti->private; + struct pool *pool = pt->pool; + sector_t data_size = ti->len; + dm_block_t sb_data_size; + + *need_commit = false; + + (void) sector_div(data_size, pool->sectors_per_block); + + r = dm_pool_get_data_dev_size(pool->pmd, &sb_data_size); + if (r) { + DMERR("%s: failed to retrieve data device size", + dm_device_name(pool->pool_md)); + return r; + } + + if (data_size < sb_data_size) { + 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) { + metadata_operation_failed(pool, "dm_pool_resize_data_dev", r); + return r; + } + + *need_commit = true; + } + + return 0; +} + +static int maybe_resize_metadata_dev(struct dm_target *ti, bool *need_commit) +{ + int r; + struct pool_c *pt = ti->private; + struct pool *pool = pt->pool; + dm_block_t metadata_dev_size, sb_metadata_dev_size; + + *need_commit = false; + + metadata_dev_size = get_metadata_dev_size_in_blocks(pool->md_dev); + + r = dm_pool_get_metadata_dev_size(pool->pmd, &sb_metadata_dev_size); + if (r) { + 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("%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) { + metadata_operation_failed(pool, "dm_pool_resize_metadata_dev", r); + return r; + } + + *need_commit = true; + } + + return 0; +} + /* * Retrieves the number of blocks of the data device from * the superblock and compares it to the actual device size, @@ -2088,9 +2666,9 @@ static int pool_map(struct dm_target *ti, struct bio *bio, static int pool_preresume(struct dm_target *ti) { int r; + bool need_commit1, need_commit2; struct pool_c *pt = ti->private; struct pool *pool = pt->pool; - dm_block_t data_size, sb_data_size; /* * Take control of the pool object. @@ -2099,32 +2677,16 @@ static int pool_preresume(struct dm_target *ti) if (r) return r; - data_size = ti->len >> pool->block_shift; - r = dm_pool_get_data_dev_size(pool->pmd, &sb_data_size); - if (r) { - DMERR("failed to retrieve data device size"); + r = maybe_resize_data_dev(ti, &need_commit1); + if (r) return r; - } - if (data_size < sb_data_size) { - DMERR("pool target too small, is %llu blocks (expected %llu)", - data_size, sb_data_size); - return -EINVAL; - - } else if (data_size > sb_data_size) { - r = dm_pool_resize_data_dev(pool->pmd, data_size); - if (r) { - DMERR("failed to resize data device"); - return r; - } + r = maybe_resize_metadata_dev(ti, &need_commit2); + if (r) + return r; - r = dm_pool_commit_metadata(pool->pmd); - if (r) { - DMERR("%s: dm_pool_commit_metadata() failed, error = %d", - __func__, r); - return r; - } - } + if (need_commit1 || need_commit2) + (void) commit(pool); return 0; } @@ -2136,29 +2698,22 @@ static void pool_resume(struct dm_target *ti) unsigned long flags; spin_lock_irqsave(&pool->lock, flags); - pool->low_water_triggered = 0; - pool->no_free_space = 0; - __requeue_bios(pool); + pool->low_water_triggered = false; spin_unlock_irqrestore(&pool->lock, flags); + requeue_bios(pool); do_waker(&pool->waker.work); } static void pool_postsuspend(struct dm_target *ti) { - int r; struct pool_c *pt = ti->private; struct pool *pool = pt->pool; cancel_delayed_work(&pool->waker); + cancel_delayed_work(&pool->no_space_timeout); flush_workqueue(pool->wq); - - r = dm_pool_commit_metadata(pool->pmd); - if (r < 0) { - DMERR("%s: dm_pool_commit_metadata() failed, error = %d", - __func__, r); - /* FIXME: invalidate device? error the next FUA or FLUSH bio ?*/ - } + (void) commit(pool); } static int check_arg_count(unsigned argc, unsigned args_required) @@ -2292,12 +2847,7 @@ static int process_reserve_metadata_snap_mesg(unsigned argc, char **argv, struct if (r) return r; - r = dm_pool_commit_metadata(pool->pmd); - if (r) { - DMERR("%s: dm_pool_commit_metadata() failed, error = %d", - __func__, r); - return r; - } + (void) commit(pool); r = dm_pool_reserve_metadata_snap(pool->pmd); if (r) @@ -2358,25 +2908,45 @@ static int pool_message(struct dm_target *ti, unsigned argc, char **argv) else DMWARN("Unrecognised thin pool target message received: %s", argv[0]); - if (!r) { - r = dm_pool_commit_metadata(pool->pmd); - if (r) - DMERR("%s message: dm_pool_commit_metadata() failed, error = %d", - argv[0], r); - } + if (!r) + (void) commit(pool); return r; } +static void emit_flags(struct pool_features *pf, char *result, + unsigned sz, unsigned maxlen) +{ + unsigned count = !pf->zero_new_blocks + !pf->discard_enabled + + !pf->discard_passdown + (pf->mode == PM_READ_ONLY) + + pf->error_if_no_space; + DMEMIT("%u ", count); + + if (!pf->zero_new_blocks) + DMEMIT("skip_block_zeroing "); + + if (!pf->discard_enabled) + DMEMIT("ignore_discard "); + + if (!pf->discard_passdown) + DMEMIT("no_discard_passdown "); + + if (pf->mode == PM_READ_ONLY) + DMEMIT("read_only "); + + if (pf->error_if_no_space) + DMEMIT("error_if_no_space "); +} + /* * Status line is: * <transaction id> <used metadata sectors>/<total metadata sectors> * <used data sectors>/<total data sectors> <held metadata root> */ -static int pool_status(struct dm_target *ti, status_type_t type, - char *result, unsigned maxlen) +static void pool_status(struct dm_target *ti, status_type_t type, + unsigned status_flags, char *result, unsigned maxlen) { - int r, count; + int r; unsigned sz = 0; uint64_t transaction_id; dm_block_t nr_free_blocks_data; @@ -2391,32 +2961,56 @@ static int pool_status(struct dm_target *ti, status_type_t type, switch (type) { case STATUSTYPE_INFO: - r = dm_pool_get_metadata_transaction_id(pool->pmd, - &transaction_id); - if (r) - return r; + if (get_pool_mode(pool) == PM_FAIL) { + DMEMIT("Fail"); + break; + } - r = dm_pool_get_free_metadata_block_count(pool->pmd, - &nr_free_blocks_metadata); - if (r) - return r; + /* Commit to ensure statistics aren't out-of-date */ + if (!(status_flags & DM_STATUS_NOFLUSH_FLAG) && !dm_suspended(ti)) + (void) commit(pool); + + r = dm_pool_get_metadata_transaction_id(pool->pmd, &transaction_id); + if (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("%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) - return r; + if (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) - return r; + r = dm_pool_get_free_block_count(pool->pmd, &nr_free_blocks_data); + if (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) - return r; + if (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) - return r; + if (r) { + DMERR("%s: dm_pool_get_metadata_snap returned %d", + dm_device_name(pool->pool_md), r); + goto err; + } DMEMIT("%llu %llu/%llu %llu/%llu ", (unsigned long long)transaction_id, @@ -2426,9 +3020,28 @@ static int pool_status(struct dm_target *ti, status_type_t type, (unsigned long long)nr_blocks_data); if (held_root) - DMEMIT("%llu", held_root); + DMEMIT("%llu ", held_root); + else + DMEMIT("- "); + + 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 "); + else if (pool->pf.discard_passdown) + DMEMIT("discard_passdown "); + else + DMEMIT("no_discard_passdown "); + + if (pool->pf.error_if_no_space) + DMEMIT("error_if_no_space "); else - DMEMIT("-"); + DMEMIT("queue_if_no_space "); break; @@ -2438,24 +3051,13 @@ static int pool_status(struct dm_target *ti, status_type_t type, format_dev_t(buf2, pt->data_dev->bdev->bd_dev), (unsigned long)pool->sectors_per_block, (unsigned long long)pt->low_water_blocks); - - count = !pool->pf.zero_new_blocks + !pool->pf.discard_enabled + - !pt->pf.discard_passdown; - DMEMIT("%u ", count); - - if (!pool->pf.zero_new_blocks) - DMEMIT("skip_block_zeroing "); - - if (!pool->pf.discard_enabled) - DMEMIT("ignore_discard "); - - if (!pt->pf.discard_passdown) - DMEMIT("no_discard_passdown "); - + emit_flags(&pt->requested_pf, result, sz, maxlen); break; } + return; - return 0; +err: + DMEMIT("Error"); } static int pool_iterate_devices(struct dm_target *ti, @@ -2480,37 +3082,66 @@ static int pool_merge(struct dm_target *ti, struct bvec_merge_data *bvm, return min(max_size, q->merge_bvec_fn(q, bvm, biovec)); } -static void set_discard_limits(struct pool *pool, struct queue_limits *limits) +static void set_discard_limits(struct pool_c *pt, struct queue_limits *limits) { - /* - * FIXME: these limits may be incompatible with the pool's data device - */ + struct pool *pool = pt->pool; + struct queue_limits *data_limits; + limits->max_discard_sectors = pool->sectors_per_block; /* - * This is just a hint, and not enforced. We have to cope with - * bios that overlap 2 blocks. + * discard_granularity is just a hint, and not enforced. */ - limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT; - limits->discard_zeroes_data = pool->pf.zero_new_blocks; + 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; + uint64_t io_opt_sectors = limits->io_opt >> 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 || + do_div(io_opt_sectors, pool->sectors_per_block)) { + blk_limits_io_min(limits, 0); + 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) { + /* + * 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); - blk_limits_io_min(limits, 0); - blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT); - if (pool->pf.discard_enabled) - set_discard_limits(pool, limits); + set_discard_limits(pt, limits); } static struct target_type pool_target = { .name = "thin-pool", .features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE | DM_TARGET_IMMUTABLE, - .version = {1, 2, 0}, + .version = {1, 12, 0}, .module = THIS_MODULE, .ctr = pool_ctr, .dtr = pool_dtr, @@ -2528,9 +3159,29 @@ static struct target_type pool_target = { /*---------------------------------------------------------------- * 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; + + thin_put(tc); + wait_for_completion(&tc->can_destroy); + + spin_lock_irqsave(&tc->pool->lock, flags); + list_del_rcu(&tc->list); + spin_unlock_irqrestore(&tc->pool->lock, flags); + synchronize_rcu(); mutex_lock(&dm_thin_pool_table.mutex); @@ -2562,6 +3213,7 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv) 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); @@ -2577,6 +3229,10 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv) r = -ENOMEM; goto out_unlock; } + spin_lock_init(&tc->lock); + 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); @@ -2615,27 +3271,57 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv) } __pool_inc(tc->pool); + 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; + } + 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; } - ti->split_io = tc->pool->sectors_per_block; - ti->num_flush_requests = 1; + r = dm_set_target_max_io_len(ti, tc->pool->sectors_per_block); + if (r) + goto bad_target_max_io_len; + + 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 = 1; - ti->num_discard_requests = 1; + ti->discards_supported = true; + ti->num_discard_bios = 1; + /* Discard bios must be split on a block boundary */ + ti->split_discard_bios = true; } dm_put(pool_md); mutex_unlock(&dm_thin_pool_table.mutex); + atomic_set(&tc->refcount, 1); + init_completion(&tc->can_destroy); + + spin_lock_irqsave(&tc->pool->lock, flags); + 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(); + return 0; +bad_target_max_io_len: + dm_pool_close_thin_device(tc->td); bad_thin_open: __pool_dec(tc->pool); bad_pool_lookup: @@ -2653,32 +3339,29 @@ out_unlock: return r; } -static int thin_map(struct dm_target *ti, struct bio *bio, - union map_info *map_context) +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, map_context); + return thin_bio_map(ti, bio); } -static int thin_endio(struct dm_target *ti, - struct bio *bio, int err, - union map_info *map_context) +static int thin_endio(struct dm_target *ti, struct bio *bio, int err) { unsigned long flags; - struct dm_thin_endio_hook *h = map_context->ptr; + struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook)); struct list_head work; struct dm_thin_new_mapping *m, *tmp; struct pool *pool = h->tc->pool; if (h->shared_read_entry) { INIT_LIST_HEAD(&work); - ds_dec(h->shared_read_entry, &work); + dm_deferred_entry_dec(h->shared_read_entry, &work); spin_lock_irqsave(&pool->lock, flags); list_for_each_entry_safe(m, tmp, &work, list) { list_del(&m->list); - m->quiesced = 1; + m->quiesced = true; __maybe_add_mapping(m); } spin_unlock_irqrestore(&pool->lock, flags); @@ -2686,29 +3369,43 @@ static int thin_endio(struct dm_target *ti, if (h->all_io_entry) { INIT_LIST_HEAD(&work); - ds_dec(h->all_io_entry, &work); - spin_lock_irqsave(&pool->lock, flags); - list_for_each_entry_safe(m, tmp, &work, list) - list_add(&m->list, &pool->prepared_discards); - spin_unlock_irqrestore(&pool->lock, flags); + dm_deferred_entry_dec(h->all_io_entry, &work); + if (!list_empty(&work)) { + spin_lock_irqsave(&pool->lock, flags); + list_for_each_entry_safe(m, tmp, &work, list) + list_add_tail(&m->list, &pool->prepared_discards); + spin_unlock_irqrestore(&pool->lock, flags); + wake_worker(pool); + } } - mempool_free(h, pool->endio_hook_pool); - 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); } /* * <nr mapped sectors> <highest mapped sector> */ -static int thin_status(struct dm_target *ti, status_type_t type, - char *result, unsigned maxlen) +static void thin_status(struct dm_target *ti, status_type_t type, + unsigned status_flags, char *result, unsigned maxlen) { int r; ssize_t sz = 0; @@ -2716,18 +3413,27 @@ static int thin_status(struct dm_target *ti, status_type_t type, char buf[BDEVNAME_SIZE]; struct thin_c *tc = ti->private; + if (get_pool_mode(tc->pool) == PM_FAIL) { + DMEMIT("Fail"); + return; + } + if (!tc->td) DMEMIT("-"); else { switch (type) { case STATUSTYPE_INFO: r = dm_thin_get_mapped_count(tc->td, &mapped); - if (r) - return r; + if (r) { + DMERR("dm_thin_get_mapped_count returned %d", r); + goto err; + } r = dm_thin_get_highest_mapped_block(tc->td, &highest); - if (r < 0) - return r; + if (r < 0) { + DMERR("dm_thin_get_highest_mapped_block returned %d", r); + goto err; + } DMEMIT("%llu ", mapped * tc->pool->sectors_per_block); if (r) @@ -2747,51 +3453,46 @@ static int thin_status(struct dm_target *ti, status_type_t type, } } - return 0; + return; + +err: + DMEMIT("Error"); } static int thin_iterate_devices(struct dm_target *ti, iterate_devices_callout_fn fn, void *data) { - dm_block_t blocks; + sector_t blocks; struct thin_c *tc = ti->private; + struct pool *pool = tc->pool; /* * We can't call dm_pool_get_data_dev_size() since that blocks. So * we follow a more convoluted path through to the pool's target. */ - if (!tc->pool->ti) + if (!pool->ti) return 0; /* nothing is bound */ - blocks = tc->pool->ti->len >> tc->pool->block_shift; + blocks = pool->ti->len; + (void) sector_div(blocks, pool->sectors_per_block); if (blocks) - return fn(ti, tc->pool_dev, 0, tc->pool->sectors_per_block * blocks, data); + return fn(ti, tc->pool_dev, 0, pool->sectors_per_block * blocks, data); 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; - - blk_limits_io_min(limits, 0); - blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT); - set_discard_limits(pool, limits); -} - static struct target_type thin_target = { .name = "thin", - .version = {1, 1, 0}, + .version = {1, 12, 0}, .module = THIS_MODULE, .ctr = thin_ctr, .dtr = thin_dtr, .map = thin_map, .end_io = thin_endio, + .presuspend = thin_presuspend, .postsuspend = thin_postsuspend, .status = thin_status, .iterate_devices = thin_iterate_devices, - .io_hints = thin_io_hints, }; /*----------------------------------------------------------------*/ @@ -2812,25 +3513,13 @@ static int __init dm_thin_init(void) r = -ENOMEM; - _cell_cache = KMEM_CACHE(dm_bio_prison_cell, 0); - if (!_cell_cache) - goto bad_cell_cache; - _new_mapping_cache = KMEM_CACHE(dm_thin_new_mapping, 0); if (!_new_mapping_cache) goto bad_new_mapping_cache; - _endio_hook_cache = KMEM_CACHE(dm_thin_endio_hook, 0); - if (!_endio_hook_cache) - goto bad_endio_hook_cache; - return 0; -bad_endio_hook_cache: - kmem_cache_destroy(_new_mapping_cache); bad_new_mapping_cache: - kmem_cache_destroy(_cell_cache); -bad_cell_cache: dm_unregister_target(&pool_target); bad_pool_target: dm_unregister_target(&thin_target); @@ -2843,14 +3532,15 @@ static void dm_thin_exit(void) dm_unregister_target(&thin_target); dm_unregister_target(&pool_target); - kmem_cache_destroy(_cell_cache); kmem_cache_destroy(_new_mapping_cache); - kmem_cache_destroy(_endio_hook_cache); } 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 <dm-devel@redhat.com>"); MODULE_LICENSE("GPL"); |