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 <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org

5 files changed, 29 insertions, 5 deletions

diff --git a/drivers/md/dm-thin-metadata.c b/drivers/md/dm-thin-metadata.c
index 60bce435f4fa..8a30ad54bd46 100644
--- a/drivers/md/dm-thin-metadata.c
+++ b/drivers/md/dm-thin-metadata.c
@@ -1697,6 +1697,14 @@ void dm_pool_metadata_read_only(struct dm_pool_metadata *pmd)
 	up_write(&pmd->root_lock);
 }
 
+void dm_pool_metadata_read_write(struct dm_pool_metadata *pmd)
+{
+	down_write(&pmd->root_lock);
+	pmd->read_only = false;
+	dm_bm_set_read_write(pmd->bm);
+	up_write(&pmd->root_lock);
+}
+
 int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
 					dm_block_t threshold,
 					dm_sm_threshold_fn fn,
diff --git a/drivers/md/dm-thin-metadata.h b/drivers/md/dm-thin-metadata.h
index 845ebbe589a9..7bcc0e1d6238 100644
--- a/drivers/md/dm-thin-metadata.h
+++ b/drivers/md/dm-thin-metadata.h
@@ -193,6 +193,7 @@ int dm_pool_resize_metadata_dev(struct dm_pool_metadata *pmd, dm_block_t new_siz
  * that nothing is changing.
  */
 void dm_pool_metadata_read_only(struct dm_pool_metadata *pmd);
+void dm_pool_metadata_read_write(struct dm_pool_metadata *pmd);
 
 int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
 					dm_block_t threshold,
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;
diff --git a/drivers/md/persistent-data/dm-block-manager.c b/drivers/md/persistent-data/dm-block-manager.c
index a7e8bf296388..064a3c271baa 100644
--- a/drivers/md/persistent-data/dm-block-manager.c
+++ b/drivers/md/persistent-data/dm-block-manager.c
@@ -626,6 +626,12 @@ void dm_bm_set_read_only(struct dm_block_manager *bm)
 }
 EXPORT_SYMBOL_GPL(dm_bm_set_read_only);
 
+void dm_bm_set_read_write(struct dm_block_manager *bm)
+{
+	bm->read_only = false;
+}
+EXPORT_SYMBOL_GPL(dm_bm_set_read_write);
+
 u32 dm_bm_checksum(const void *data, size_t len, u32 init_xor)
 {
 	return crc32c(~(u32) 0, data, len) ^ init_xor;
diff --git a/drivers/md/persistent-data/dm-block-manager.h b/drivers/md/persistent-data/dm-block-manager.h
index 9a82083a66b6..13cd58e1fe69 100644
--- a/drivers/md/persistent-data/dm-block-manager.h
+++ b/drivers/md/persistent-data/dm-block-manager.h
@@ -108,9 +108,9 @@ int dm_bm_unlock(struct dm_block *b);
 int dm_bm_flush_and_unlock(struct dm_block_manager *bm,
 			   struct dm_block *superblock);
 
- /*
-  * Request data be prefetched into the cache.
-  */
+/*
+ * Request data is prefetched into the cache.
+ */
 void dm_bm_prefetch(struct dm_block_manager *bm, dm_block_t b);
 
 /*
@@ -125,6 +125,7 @@ void dm_bm_prefetch(struct dm_block_manager *bm, dm_block_t b);
  * be returned if you do.
  */
 void dm_bm_set_read_only(struct dm_block_manager *bm);
+void dm_bm_set_read_write(struct dm_block_manager *bm);
 
 u32 dm_bm_checksum(const void *data, size_t len, u32 init_xor);