On Tue, 20 Mar 2018 09:44:45 +0100 Michal Hocko <mhocko@xxxxxxxx> wrote: > > And I wonder if overloading CONFIG_PROVE_LOCKING is appropriate here. > > CONFIG_PROVE_LOCKING is a high-level thing under which a whole bunch of > > different debugging options may exist. > > Yes but it is meant to catch locking issues in general so I think doing > this check under the same config makes sense. > > > I guess we should add a new config item under PROVE_LOCKING, > > I am not convinced a new config is really worth it. We have way too many > already and PROVE_LOCKING sounds like a good fit to me. I few scruffy misc sites have used PROVE_LOCKING in this fashion, but they really shouldn't have. It means that if anyone wants to enable, say, "Locking API boot-time self-tests" then they must enable PROVE_LOCKING, so they accidentally get this feature as well.