[RFC][PATCH 0/3] spin_unlock_wait and assorted borkage

[Peter Zijlstra]

As per recent discussions spin_unlock_wait() has an unintuitive 'feature'.

  lkml.kernel.org/r/20160520053926.GC31084@xxxxxxxxxxxxxxx

These patches pull the existing solution into generic code; annotate all
spin_unlock_wait() users and fix nf_conntrack more.

The alternative -- putting smp_acquire__after_ctrl_dep() into
spin_unlock_wait() can be done later if desired, now that al users are audited
and corrected.

Please (very) carefully consider.