On Thu, 8 Oct 2026 at 03:03, Alison Schofield <[email protected]> wrote: > As written, it suggests that READ_ONCE() makes a non-NULL result safe against > concurrent fs_put_dax(). However, READ_ONCE() only controls the pointer load, > but > IIUC it does not protect the lifetime of the returned holder.
Right, the changelog was a bit hurried. Will fix. > Also, is this RCU-based lifetime handling specific to the new FUSE > implementation, > or is it intended to establish a general requirement for DAX holders and > callers > of dax_holder()? I haven't looked. The RCU based fix only works as long as the memory failure handler doesn't need to sleep. I cannot make a comment on the xfs code. Thanks, Miklos
