Re: [PATCH RFC] percpu: add data dependency barrier in percpu accessors and operations

[Christoph Lameter <cl@gentwo.org>]

On Thu, 19 Jun 2014, Tejun Heo wrote:

> Again, data dependency barrier is noop in all in-use archs.

A barrier limits what the compiler can do.

> > Remote write events are contrary to that design and are exceedingly rare.
> > An IPI is justifiable for such a rare event. At least in my use cases I
> > have always found that to be sufficient. Well, I designed the data
> > structures in a way that made this possible because of the design criteria
> > that did not allow me remote write access to other processors per cpu
> > data.
>
> You're repeatedly getting wayside in the discussion.  What are you
> suggesting?  Sending IPIs on each percpu allocation?

No this is about sending an IPI if you want to modify the percpu data of
another process. There was a mentionig of code that modifies the per cpu
data of another processor?

> Again, I'm leaning towards just clarifying the init write ownership to
> the allocating CPU as that seems the most straight forward way to deal
> with it, but please stop brining up the raw performance thing.  Nobody
> is doing anything to that.  It's not relevant in the discussion.

Ok sounds good.