The b_error check right after xfs_trans_get_buf() is redundant:
1) If the buffer is found in transaction via xfs_trans_buf_item_match(),
any corrupted metadata error would have already been exposed during
previous reads like xfs_da3_node_read().
2) If the buffer is obtained via xfs_buf_get_map():
- It's called without XBF_READ flag, so won't return buffer with
b_error set, since xfs_buf_get_map() will clear it anyway.
- Buffer found in cache normally won't have error since previous reads
had checked it, unless someone corrupts the buffer and the AIL
pushes it out to disk while the buffer's unlocked. But in this case,
AIL will shut down the log.
Remove this redundant check to simplify the code, make the code consistent
with most other xfs_trans_get_buf() callers in XFS.
Signed-off-by: Long Li <leo.lilong@xxxxxxxxxx>
---
fs/xfs/xfs_attr_inactive.c | 5 -----
1 file changed, 5 deletions(-)
diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
index 24fb12986a56..319004bf089f 100644
--- a/fs/xfs/xfs_attr_inactive.c
+++ b/fs/xfs/xfs_attr_inactive.c
@@ -305,11 +305,6 @@ xfs_attr3_root_inactive(
XFS_FSB_TO_BB(mp, mp->m_attr_geo->fsbcount), 0, &bp);
if (error)
return error;
- error = bp->b_error;
- if (error) {
- xfs_trans_brelse(*trans, bp);
- return error;
- }
xfs_trans_binval(*trans, bp); /* remove from cache */
/*
* Commit the invalidate and start the next transaction.
--
2.39.2
