On Wed, Jul 16, 2025 at 09:04:31AM -0700, Darrick J. Wong wrote: > On Wed, Jul 16, 2025 at 03:03:19PM +0200, Christoph Hellwig wrote: > > Split up the XFS_IS_CORRUPT statement so that it immediately shows > > if the reference counter overflowed or underflowed. > > Should we be using refcount_t for xg_active_ref to detect > over/underflows, then? In theory yes. In practice refcount_t still has a lot of performance overhead, so using it in something used a lot in the fast path might not be a good idea.