On Tue, Dec 01, 2020 at 11:50:25AM +0100, Christoph Hellwig wrote: > The READ_ONCE still looks suspect as it generally needs to be paired > with a WRITE_ONCE. The rest looks sane to me. Yeah, the comment from the other location is, /* Pairs with smp_load_acquire() in mnt_user_ns(). */ So I think it's just a typo that needs to be fixed. Tycho
