Btrfs: fix unprotected device->bytes_used update
authorMiao Xie <miaox@cn.fujitsu.com>
Wed, 3 Sep 2014 13:35:36 +0000 (21:35 +0800)
committerChris Mason <clm@fb.com>
Wed, 17 Sep 2014 20:38:36 +0000 (13:38 -0700)
We should update device->bytes_used in the lock context of
chunk_mutex, or we would get wrong data.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
fs/btrfs/volumes.c

index 1524b3f4403674ed2420d25a955ae39108b757ab..45e0b5d0f6dd1a985a3cfa0f90a777fea6f78434 100644 (file)
@@ -4429,6 +4429,9 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
        if (ret)
                goto error_del_extent;
 
+       for (i = 0; i < map->num_stripes; i++)
+               map->stripes[i].dev->bytes_used += stripe_size;
+
        free_extent_map(em);
        check_raid56_incompat_flag(extent_root->fs_info, type);
 
@@ -4500,7 +4503,6 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
                device = map->stripes[i].dev;
                dev_offset = map->stripes[i].physical;
 
-               device->bytes_used += stripe_size;
                ret = btrfs_update_device(trans, device);
                if (ret)
                        goto out;