On Sun, 2019-04-28 at 16:52 +0100, Al Viro wrote: > On Sun, Apr 28, 2019 at 11:47:58AM -0400, Jeff Layton wrote: > > > We could stick that in ceph_dentry_info (->d_fsdata). We have a flags > > field in there already. > > Yes, but... You have it freed in ->d_release(), AFAICS, and without > any delays. So lockless accesses will be trouble. That's easy enough to fix -- we could add a rcu_head to it and call_rcu in ceph_d_release instead of just freeing it immediately. I guess if we find that d_fsdata is NULL, we can just treat it like we currently do a negative dentry? -- Jeff Layton <jlayton@xxxxxxxxxx>
