[PATCH v8.1 2/2] btrfs: mark device failed for write and flush errors

[Anand Jain <anand.jain@oracle.com>]

This patch marks device as failed for the write/flush errors
so that it can be brought out of the RW device list. However
if the chunk render is incomplete due to missing of a device
then the volume/raid fails, so this situation is handled by
calling btrfs_handle_fs_error() which puts the whole FS to
RDONLY.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
This patch is to replace
  [PATCH v8 2/2] btrfs: check device for critical errors and mark failed

 fs/btrfs/volumes.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 05b150c03995..53397d077e91 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -501,6 +501,7 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
 				u64 chunk_offset, u64 chunk_size);
 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 		       struct btrfs_fs_info *fs_info, u64 chunk_offset);
+void btrfs_mark_device_failed(struct btrfs_device *dev);
 
 static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
 {
@@ -513,6 +514,10 @@ static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
 	atomic_inc(dev->dev_stat_values + index);
 	smp_mb__before_atomic();
 	atomic_inc(&dev->dev_stats_ccnt);
+
+	if (index == BTRFS_DEV_STAT_WRITE_ERRS ||
+		index == BTRFS_DEV_STAT_FLUSH_ERRS)
+		btrfs_mark_device_failed(dev);
 }
 
 static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
-- 
2.13.1