Hi,

My $0.02 ...


> > > for today we have what is inside barebox and this patche is coherent
> > > with what is actually inside barebox (please check other drivers in mfd
> > > and mx35-3stack for example).
> > > For the future we may no more need this if we have a proper PMIC API
> > > but as this API actually is not written, this patchset makes sense.
> > 
> > today you do not need it
> > 
> > today you just need to known it the chip is here
> > 
> > so just check if the devfs is present
> 
> Don't know where your problem is with exporting it. 

I also don't see the problem. 99.9% of designs using this IC (family) will 
only have one of them on a board. If someone puts more than one on their board 
then they should be the person to manage dealing with two.

> Yes, it supports
> only a single pmic, but that's all we have on the boards I know. And
> this function can be used to modify registers on the pmic, some boards
> need this. The day might come when we need some regulator API, but we
> are not there and 

The board I have uses an mc13892 - there is stuff we need to we need to do 
which is unrelated to regulators like increasing the charge/supply current and 
detecting critically low battery. Calling xxxx_set_bits() is much better than 
a read/modify/write from dev file.


> often enough getting a board to run properly is just
> 'modify this damn bit in the pmic'

I've said "this damn pmic" quite a few times about the 13892 :) (Has anyone 
else had troubles too?)


Cheers,
Marc