Hi Boris,

> > 
> > Could you be more specific here? 
> > 
> > As I mentioned before, the problem is not with the lack of
> > "atomic" API.  
> 
> Below is a quick and dirty I made on top of this patch to show you how
> atomic update can be implemented in this driver. 

Thank you for example patch.

I will implement the ->apply() callback and post patches very soon :-).

I had two issues with the ->apply() implementation:

- Do my work on top of this patch (https://lkml.org/lkml/2016/10/7/454
  as you did) to avoid rewriting work already done.

- In the example ->apply() implementation for rockchip
  (https://patchwork.kernel.org/patch/7228221/) the ->config() callback
  was not removed when ->apply() was implemented. I was confused with
  such approach, but as you explained in this mail, the solely ->apply()
  is enough.

> It's not tested, and
> probably not working, but it should give you a better idea of what is
> expected.

Thanks for explanation,

Łukasz Majewski
[unhandled content-type:application/pgp-signature]