On Fri, Oct 4, 2019 at 7:27 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> FWIW, vfs.git#fixes (or #next.dcache) ought to deal with that one.
Dang, I thought this already got merged. But we only discussed it
extensively and I guess it got delayed by all the discussions about
possible fixes for the d_lock contention.
Al, mind sending me that one - and honestly, I'd take the "cursors off
the list at the end" patch too. That may not be stable material, but I
still think it's going to help the d_lock contention at least
partially in practice.
Linus
