On Sat, 22 Mar 2025, Al Viro wrote: > On Wed, Mar 19, 2025 at 02:01:35PM +1100, NeilBrown wrote: > > Quite a few of those should be switched to a common helper, > lifted out of debugfs/tracefs start_creating()... The next batch of patches will add "lookup_and_lock" family of functions and change various calls to use them, including the "start_creating" functions. The goal there is to get all the code for lock/unlock around create/remove/rename to be in namei.c so we can change it in one place. Thanks, NeilBrown > > I can live with that, but it'll cause fuckloads of noise in > persistency queue... ;-/ >