On 05/15/2015 09:15 AM, Viresh Kumar wrote:

On 14-05-15, 03:25, Michael Turquette wrote:

...

No, we don't understand the problem space well enough to form an ABI.

And why do you think so? We have been facing many problems since a long time which we are trying to solve here.

I would state "problem space is better defined now based on data made public by developers on various SoCs", this new binding seems to address majority of the concerns (esp with vendor specific extensions). OPP behavior is very SoC vendor specific -> it can only evolve with an extensible framework - which is what this new binding provides. This is something that was badly missing in the older binding and framework (I should blame myself for it), even though the previous definitions were simple, in effect it was inflexible to the detriment of many SoCs.

Do we know 100% if the new binding solves every SoC's issues - we wont be able to do that unless folks speak up - but then, providing ability for vendor specific extension allows to consolidate and make common as necessary.

Point blank rejection might be a bit of an overkill, IMHO.

I agree that it might not be right to try too many things which may not be required later, but most of the things we have now in new bindings are actually required.

...

Putting this stuff in C without any philosophical constraints on whether or not we can change it later is the way to go.

I don't agree to that :)

I second Viresh on this. Benefit of forcing data separation into device tree has provided the flexibility now to be able to loadup OPPs from bootloader OR over DTC overlay as desired - that is the right choice rather than embedding it within C code, providing kludgy extension options to provide dynamic data updates.