Hello, Paul.

On Fri, Sep 02, 2016 at 10:00:56AM -0700, Paul E. McKenney wrote:
> Note that lockdep believes that the rcu_read_lock() has also been
> dropped, otherwise it would have listed it.

There isn't any actual code between rcu_read_lock() and the following
rcu deref.  That's the first thing which happens.  It looked
mysterious to me so I was wondering whether you knew any brekages on
rcu lockdep side.  Unfortunately, that doesn't seem to be the case.

> At this point, I have to suggest "git bisect"...

I'll try to reproduce it and see wth is going on.

Thanks!

-- 
tejun