Re: [PATCH tip/core/rcu 16/27] rcu: Add comment documenting how rcu_seq_snap works

From: Paul E. McKenney
Date: Tue Jun 26 2018 - 15:29:53 EST


On Tue, Jun 26, 2018 at 09:21:13PM +0200, Peter Zijlstra wrote:
> On Tue, Jun 26, 2018 at 11:08:55AM -0700, Paul E. McKenney wrote:
> > On Tue, Jun 26, 2018 at 07:14:54PM +0200, Peter Zijlstra wrote:
> > > On Mon, Jun 25, 2018 at 05:35:02PM -0700, Paul E. McKenney wrote:
> > > > From: "Joel Fernandes (Google)" <joel@xxxxxxxxxxxxxxxxx>
> > > >
> > > > rcu_seq_snap may be tricky to decipher. Lets document how it works with
> > > > an example to make it easier.
> > >
> > > Since you had me looking at them functions; why isn't rcu_seq_snap()
> > > using smp_load_acquire() and rcu_seq_end() using smp_store_release() ?
> > > Their respective comments seem to suggest that would be sufficent.
> >
> > I do not believe that this would suffice. Would it make sense to refer
> > to Documentation/RCU/Design/Memory-Ordering in the comment header?
>
> No, because I can't read that thing in an editor.
>
> > Except that this would invite sprinkling this pathname far and wide...
> >
> > The key point is that these functions are part of the any-to-any
> > memory-ordering guarantee that RCU grace periods provide.
>
> Then the existing comment is misleading and really needs change.

Would it be sufficient to add something like "The memory barrier is
required to support the many-to-many ordering guaranteed by RCU grace
periods"?

Thanx, Paul