Re: hugepage related lockdep trace.

[Michal Hocko <mhocko@suse.cz>]

On Mon 29-07-13 17:20:01, Peter Zijlstra wrote:
> On Mon, Jul 29, 2013 at 04:53:08PM +0200, Michal Hocko wrote:
> > Peter, for you context the lockdep splat has been reported
> > here: https://lkml.org/lkml/2013/7/17/381
> > 
> > Minchan has proposed to workaround it by using SINGLE_DEPTH_NESTING
> > https://lkml.org/lkml/2013/7/23/812
> > 
> > my idea was to use a separate class key for hugetlb as it is quite
> > special in many ways:
> > https://lkml.org/lkml/2013/7/25/277
> > 
> > What is the preferred way of fixing such an issue?
> 
> The class is the safer annotation.

OK, I will use the class then. It should prevent other false positives
AFAIU.

> That said; it is a rather horrible issue any which way. This PMD sharing
> is very unique to hugetlbfs (also is that really worth the effort these
> days?) and it will make it impossible to make hugetlbfs swappable.

No idea.

> The other solution is to make the pmd allocation GFP_NOFS.

That would be just papering over the lockdep limitation. So I would
rather stick with something lockdep specific.

I will cook up a patch.

Thanks!
-- 
Michal Hocko
SUSE Labs