diff mbox

[v3] Btrfs: avoid losing data raid profile when deleting a device

Message ID 20171115232811.9756-1-bo.li.liu@oracle.com (mailing list archive)
State New, archived
Headers show

Commit Message

Liu Bo Nov. 15, 2017, 11:28 p.m. UTC
We've avoided data losing raid profile when doing balance, but it
turns out that deleting a device could also result in the same
problem.

Say we have 3 disks, and they're created with '-d raid1' profile.

- We have chunk P (the only data chunk on the empty btrfs).

- Suppose that chunk P's two raid1 copies reside in disk A and disk B.

- Now, 'btrfs device remove disk B'
         btrfs_rm_device()
	   -> btrfs_shrink_device()
	      -> btrfs_relocate_chunk() #relocate any chunk on disk B
	      	 			 to other places.

- Chunk P will be removed and a new chunk will be created to hold
  those data, but as chunk P is the only one holding raid1 profile,
  after it goes away, the new chunk will be created as single profile
  which is our default profile.

This fixes the problem by creating an empty data chunk before
relocating the data chunk.

Metadata/System chunk are supposed to have non-zero bytes all the time
so their raid profile is preserved.

Reported-by: James Alandt <James.Alandt@wdc.com>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
---
v3: More details about how losing data profile happens.

v2: - return the correct error.
    - move helper ahead of __btrfs_balance().

 fs/btrfs/volumes.c | 84 ++++++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 65 insertions(+), 19 deletions(-)

Comments

David Sterba Jan. 5, 2018, 6:14 p.m. UTC | #1
On Wed, Nov 15, 2017 at 04:28:11PM -0700, Liu Bo wrote:
> We've avoided data losing raid profile when doing balance, but it
> turns out that deleting a device could also result in the same
> problem.
> 
> Say we have 3 disks, and they're created with '-d raid1' profile.
> 
> - We have chunk P (the only data chunk on the empty btrfs).
> 
> - Suppose that chunk P's two raid1 copies reside in disk A and disk B.
> 
> - Now, 'btrfs device remove disk B'
>          btrfs_rm_device()
> 	   -> btrfs_shrink_device()
> 	      -> btrfs_relocate_chunk() #relocate any chunk on disk B
> 	      	 			 to other places.
> 
> - Chunk P will be removed and a new chunk will be created to hold
>   those data, but as chunk P is the only one holding raid1 profile,
>   after it goes away, the new chunk will be created as single profile
>   which is our default profile.
> 
> This fixes the problem by creating an empty data chunk before
> relocating the data chunk.
> 
> Metadata/System chunk are supposed to have non-zero bytes all the time
> so their raid profile is preserved.
> 
> Reported-by: James Alandt <James.Alandt@wdc.com>
> Signed-off-by: Liu Bo <bo.li.liu@oracle.com>

Added to 4.16 queue, thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 4a72c45..a74396d 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3018,6 +3018,48 @@  static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info)
 	return ret;
 }
 
+/*
+ * return 1 : allocate a data chunk successfully,
+ * return <0: errors during allocating a data chunk,
+ * return 0 : no need to allocate a data chunk.
+ */
+static int btrfs_may_alloc_data_chunk(struct btrfs_fs_info *fs_info,
+				      u64 chunk_offset)
+{
+	struct btrfs_block_group_cache *cache;
+	u64 bytes_used;
+	u64 chunk_type;
+
+	cache = btrfs_lookup_block_group(fs_info, chunk_offset);
+	ASSERT(cache);
+	chunk_type = cache->flags;
+	btrfs_put_block_group(cache);
+
+	if (chunk_type & BTRFS_BLOCK_GROUP_DATA) {
+		spin_lock(&fs_info->data_sinfo->lock);
+		bytes_used = fs_info->data_sinfo->bytes_used;
+		spin_unlock(&fs_info->data_sinfo->lock);
+
+		if (!bytes_used) {
+			struct btrfs_trans_handle *trans;
+			int ret;
+
+			trans =	btrfs_join_transaction(fs_info->tree_root);
+			if (IS_ERR(trans))
+				return PTR_ERR(trans);
+
+			ret = btrfs_force_chunk_alloc(trans, fs_info,
+						      BTRFS_BLOCK_GROUP_DATA);
+			btrfs_end_transaction(trans);
+			if (ret < 0)
+				return ret;
+
+			return 1;
+		}
+	}
+	return 0;
+}
+
 static int insert_balance_item(struct btrfs_fs_info *fs_info,
 			       struct btrfs_balance_control *bctl)
 {
@@ -3476,7 +3518,6 @@  static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 	u32 count_meta = 0;
 	u32 count_sys = 0;
 	int chunk_reserved = 0;
-	u64 bytes_used = 0;
 
 	/* step one make some room on all the devices */
 	devices = &fs_info->fs_devices->devices;
@@ -3635,28 +3676,21 @@  static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 			goto loop;
 		}
 
-		ASSERT(fs_info->data_sinfo);
-		spin_lock(&fs_info->data_sinfo->lock);
-		bytes_used = fs_info->data_sinfo->bytes_used;
-		spin_unlock(&fs_info->data_sinfo->lock);
-
-		if ((chunk_type & BTRFS_BLOCK_GROUP_DATA) &&
-		    !chunk_reserved && !bytes_used) {
-			trans = btrfs_start_transaction(chunk_root, 0);
-			if (IS_ERR(trans)) {
-				mutex_unlock(&fs_info->delete_unused_bgs_mutex);
-				ret = PTR_ERR(trans);
-				goto error;
-			}
-
-			ret = btrfs_force_chunk_alloc(trans, fs_info,
-						      BTRFS_BLOCK_GROUP_DATA);
-			btrfs_end_transaction(trans);
+		if (!chunk_reserved) {
+			/*
+			 * We may be relocating the only data chunk we have,
+			 * which could potentially end up with losing data's
+			 * raid profile, so lets allocate an empty one in
+			 * advance.
+			 */
+			ret = btrfs_may_alloc_data_chunk(fs_info,
+							 found_key.offset);
 			if (ret < 0) {
 				mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 				goto error;
+			} else if (ret == 1) {
+				chunk_reserved = 1;
 			}
-			chunk_reserved = 1;
 		}
 
 		ret = btrfs_relocate_chunk(fs_info, found_key.offset);
@@ -4419,6 +4453,18 @@  int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 		chunk_offset = btrfs_dev_extent_chunk_offset(l, dev_extent);
 		btrfs_release_path(path);
 
+		/*
+		 * We may be relocating the only data chunk we have,
+		 * which could potentially end up with losing data's
+		 * raid profile, so lets allocate an empty one in
+		 * advance.
+		 */
+		ret = btrfs_may_alloc_data_chunk(fs_info, chunk_offset);
+		if (ret < 0) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			goto done;
+		}
+
 		ret = btrfs_relocate_chunk(fs_info, chunk_offset);
 		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)