From fafc7a815e40255d24e80a1cb7365892362fa398 Mon Sep 17 00:00:00 2001 From: Joe Thornber Date: Mon, 2 Dec 2013 17:57:42 -0500 Subject: dm thin: switch to read only mode if a mapping insert fails Switch the thin pool to read-only mode when dm_thin_insert_block() fails since there is little reason to expect the cause of the failure to be resolved without further action by user space. This issue was noticed with the device-mapper-test-suite using: dmtest run --suite thin-provisioning -n /exhausting_metadata_space_causes_fail_mode/ The quantity of errors logged in this case must be reduced. before patch: device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: dm_thin_insert_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: thin: 253:4: no free metadata space available. device-mapper: thin: 253:4: switching pool to read-only mode after patch: device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: 253:4: dm_thin_insert_block() failed: error = -28 device-mapper: thin: 253:4: switching pool to read-only mode Signed-off-by: Joe Thornber Signed-off-by: Mike Snitzer Cc: stable@vger.kernel.org --- drivers/md/dm-thin.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'drivers/md/dm-thin.c') diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index 2c0cf511ec23..24327adc925c 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -640,7 +640,9 @@ 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_LIMIT("dm_thin_insert_block() failed"); + DMERR_LIMIT("%s: dm_thin_insert_block() failed: error = %d", + dm_device_name(pool->pool_md), r); + set_pool_mode(pool, PM_READ_ONLY); cell_error(pool, m->cell); goto out; } -- cgit v1.2.2 From 4a02b34e0cf1d0d0dd3737702841da4bf615a50a Mon Sep 17 00:00:00 2001 From: Mike Snitzer Date: Tue, 3 Dec 2013 12:20:57 -0500 Subject: dm thin: switch to read-only mode if metadata space is exhausted Switch the thin pool to read-only mode in alloc_data_block() if dm_pool_alloc_data_block() fails because the pool's metadata space is exhausted. Differentiate between data and metadata space in messages about no free space available. This issue was noticed with the device-mapper-test-suite using: dmtest run --suite thin-provisioning -n /exhausting_metadata_space_causes_fail_mode/ The quantity of errors logged in this case must be reduced. before patch: device-mapper: thin: 253:4: reached low water mark for metadata device: sending event. device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: space map common: dm_tm_shadow_block() failed device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: 253:4: commit failed: error = -28 device-mapper: thin: 253:4: switching pool to read-only mode after patch: device-mapper: thin: 253:4: reached low water mark for metadata device: sending event. device-mapper: space map metadata: unable to allocate new metadata block device-mapper: thin: 253:4: no free metadata space available. device-mapper: thin: 253:4: switching pool to read-only mode Signed-off-by: Mike Snitzer Acked-by: Joe Thornber Cc: stable@vger.kernel.org --- drivers/md/dm-thin.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'drivers/md/dm-thin.c') diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index 24327adc925c..44ee489f4a6e 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -959,7 +959,7 @@ static int alloc_data_block(struct thin_c *tc, dm_block_t *result) * table reload). */ if (!free_blocks) { - DMWARN("%s: no free space available.", + DMWARN("%s: no free data space available.", dm_device_name(pool->pool_md)); spin_lock_irqsave(&pool->lock, flags); pool->no_free_space = 1; @@ -969,8 +969,16 @@ static int alloc_data_block(struct thin_c *tc, dm_block_t *result) } r = dm_pool_alloc_data_block(pool->pmd, result); - if (r) + if (r) { + if (r == -ENOSPC && + !dm_pool_get_free_metadata_block_count(pool->pmd, &free_blocks) && + !free_blocks) { + DMWARN("%s: no free metadata space available.", + dm_device_name(pool->pool_md)); + set_pool_mode(pool, PM_READ_ONLY); + } return r; + } return 0; } -- cgit v1.2.2 From 020cc3b5e28c2e24f59f53a9154faf08564f308e Mon Sep 17 00:00:00 2001 From: Joe Thornber Date: Wed, 4 Dec 2013 15:05:36 -0500 Subject: dm thin: always fallback the pool mode if commit fails Rename commit_or_fallback() to commit(). Now all previous calls to commit() will trigger the pool mode to fallback if the commit fails. Also, check the error returned from commit() in alloc_data_block(). Signed-off-by: Joe Thornber Signed-off-by: Mike Snitzer Cc: stable@vger.kernel.org --- drivers/md/dm-thin.c | 37 +++++++++++++++---------------------- 1 file changed, 15 insertions(+), 22 deletions(-) (limited to 'drivers/md/dm-thin.c') diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index 44ee489f4a6e..af79bae5ab74 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -883,32 +883,23 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block, } } -static int commit(struct pool *pool) -{ - int r; - - r = dm_pool_commit_metadata(pool->pmd); - if (r) - DMERR_LIMIT("%s: commit failed: error = %d", - dm_device_name(pool->pool_md), r); - - return r; -} - /* * A non-zero return indicates read_only or fail_io mode. * Many callers don't care about the return value. */ -static int commit_or_fallback(struct pool *pool) +static int commit(struct pool *pool) { int r; if (get_pool_mode(pool) != PM_WRITE) return -EINVAL; - r = commit(pool); - if (r) + r = dm_pool_commit_metadata(pool->pmd); + if (r) { + DMERR_LIMIT("%s: dm_pool_commit_metadata failed: error = %d", + dm_device_name(pool->pool_md), r); set_pool_mode(pool, PM_READ_ONLY); + } return r; } @@ -945,7 +936,9 @@ static int alloc_data_block(struct thin_c *tc, dm_block_t *result) * Try to commit to see if that will free up some * more space. */ - (void) commit_or_fallback(pool); + r = commit(pool); + if (r) + return r; r = dm_pool_get_free_block_count(pool->pmd, &free_blocks); if (r) @@ -1359,7 +1352,7 @@ static void process_deferred_bios(struct pool *pool) if (bio_list_empty(&bios) && !need_commit_due_to_time(pool)) return; - if (commit_or_fallback(pool)) { + if (commit(pool)) { while ((bio = bio_list_pop(&bios))) bio_io_error(bio); return; @@ -2276,7 +2269,7 @@ static int pool_preresume(struct dm_target *ti) return r; if (need_commit1 || need_commit2) - (void) commit_or_fallback(pool); + (void) commit(pool); return 0; } @@ -2303,7 +2296,7 @@ static void pool_postsuspend(struct dm_target *ti) cancel_delayed_work(&pool->waker); flush_workqueue(pool->wq); - (void) commit_or_fallback(pool); + (void) commit(pool); } static int check_arg_count(unsigned argc, unsigned args_required) @@ -2437,7 +2430,7 @@ static int process_reserve_metadata_snap_mesg(unsigned argc, char **argv, struct if (r) return r; - (void) commit_or_fallback(pool); + (void) commit(pool); r = dm_pool_reserve_metadata_snap(pool->pmd); if (r) @@ -2499,7 +2492,7 @@ static int pool_message(struct dm_target *ti, unsigned argc, char **argv) DMWARN("Unrecognised thin pool target message received: %s", argv[0]); if (!r) - (void) commit_or_fallback(pool); + (void) commit(pool); return r; } @@ -2554,7 +2547,7 @@ static void pool_status(struct dm_target *ti, status_type_t type, /* Commit to ensure statistics aren't out-of-date */ if (!(status_flags & DM_STATUS_NOFLUSH_FLAG) && !dm_suspended(ti)) - (void) commit_or_fallback(pool); + (void) commit(pool); r = dm_pool_get_metadata_transaction_id(pool->pmd, &transaction_id); if (r) { -- cgit v1.2.2 From 5383ef3a929a1366e2ced45cd6d74be7aa2a2281 Mon Sep 17 00:00:00 2001 From: Joe Thornber Date: Wed, 4 Dec 2013 16:30:01 -0500 Subject: dm thin: re-establish read-only state when switching to fail mode If the thin-pool transitioned to fail mode and the thin-pool's table were reloaded for some reason: the new table's default pool mode would be read-write, though it will transition to fail mode during resume. When the pool mode transitions directly from PM_WRITE to PM_FAIL we need to re-establish the intermediate read-only state in both the metadata and persistent-data block manager (as is usually done with the normal pool mode transition sequence: PM_WRITE -> PM_READ_ONLY -> PM_FAIL). Signed-off-by: Joe Thornber Signed-off-by: Mike Snitzer Cc: stable@vger.kernel.org --- drivers/md/dm-thin.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/md/dm-thin.c') diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index af79bae5ab74..0d7852e2b275 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -1400,6 +1400,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode mode) case PM_FAIL: DMERR("%s: switching pool to failure mode", dm_device_name(pool->pool_md)); + 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; -- cgit v1.2.2 From 9b7aaa64f96f7ca280d75326fca42f42017b89ef Mon Sep 17 00:00:00 2001 From: Joe Thornber Date: Wed, 4 Dec 2013 16:58:19 -0500 Subject: dm thin: allow pool in read-only mode to transition to read-write mode A thin-pool may be in read-only mode because the pool's data or metadata space was exhausted. To allow for recovery, by adding more space to the pool, we must allow a pool to transition from PM_READ_ONLY to PM_WRITE mode. Otherwise, running out of space will render the pool permanently read-only. Signed-off-by: Joe Thornber Signed-off-by: Mike Snitzer Cc: stable@vger.kernel.org --- drivers/md/dm-thin.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'drivers/md/dm-thin.c') diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c index 0d7852e2b275..ee29037ffc2e 100644 --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -1425,6 +1425,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode mode) break; case PM_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; @@ -1641,12 +1642,19 @@ static int bind_control_target(struct pool *pool, struct dm_target *ti) struct pool_c *pt = ti->private; /* - * We want to make sure that degraded pools are never upgraded. + * We want to make sure that a pool in PM_FAIL mode is never upgraded. */ enum pool_mode old_mode = pool->pf.mode; enum pool_mode new_mode = pt->adjusted_pf.mode; - if (old_mode > new_mode) + /* + * 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. On the other hand a PM_READ_ONLY + * may have been due to a lack of metadata or data space, and may + * now work (ie. if the underlying devices have been resized). + */ + if (old_mode == PM_FAIL) new_mode = old_mode; pool->ti = ti; -- cgit v1.2.2