On 05/09/16 22:17, maxime.ripard wrote:

Hi Maxime,

thanks for your answer.

> I'm leaving aside the rest of your mail, since we apparently have a
> very fundamental disagreement and will never reach an agreement. So,
> let's agree to disagree.

I agree ;-)

> 
> That being said...
> 
> On Mon, Sep 05, 2016 at 01:25:36AM +0100, André Przywara wrote:
>> From thinking more about this I think we can have both approaches
>> implemented:
>> - An embedded design for people who either want full control or who
>> don't care too much about the potential danger.
>> - A server design which hides those bits from Linux, instead offers
>> abstracted firmware interfaces to monitor (or control, if necessary) AXP
>> functionality.
>>
>> The DT would then ultimately describe which version is used: A "server
>> firmware" would switch the RSB to secure and would expose SCPI
>> regulators and sensors in the DT. People in favour of the embedded
>> design could load a different firmware (or somehow switch over) and use
>> the "embedded" DT with a RSB node and have full control.
> 
> What I don't want is to wait for the year of SCPI on embedded
> boards. We already wasted way too much time on this, and people are
> asking for it.

I can see that and totally agree here.

> 
>> I don't see how those should clash from a Linux perspective, as we just
>> use different drivers and describe the wanted one in the DT, without
>> having conflicting bindings.
>>
>> Yes, having different DTs for the same hardware may rise some eyebrows,
>> but in the end we just describe firmware interfaces (like PSCI) and
>> avoid describing some hardware (RSB).
> 
> But PSCI is actually a very good example. The implementation we have
> right now just generates the right DT nodes if the bootloader is
> PSCI-enabled, and the kernel just picks whatever is provided, without
> having to actually declare anything.
> 
> I could imagine the same thing for SCPI. And it would even be easier
> with sunxi-ng to modify the DT to basically drop the CCU node, and
> insert the scpi-clocks node instead.
> 
> You would only need to keep the same phandle (which is trivial) and to
> keep the same clock indices (which is not that hard either).
> 
> That would be the most obvious way to have things working I guess: if
> you use an SCPI-enabled firmware, everything is just generated at
> runtime depending on what the firmware supports.
> 
> And everything just works.

Excellent! You actually anticipate the suggestion I wanted to make in
the reply to that other mail ;-)
I guess I have no real stand against the sunxi-ng clocks anyway, since I
seem to be the only opponent and maintainers seem to be happy with it.
And similar to the regulator issue I think we don't clash on any binding.
In the world I imagine the DT comes with the firmware anyway (EDK2 does
it that way), so as long as the kernel works with both DTs ...

>> I really want to depart from this all-embedded approach which
>> exposes every bit to Linux. So why not use the first 64-bit
>> Allwinner chip as a test vehicle to prototype this? Could you live
>> with a different approach living in parallel with the classic sunxi
>> way?
> 
> If we don't have to maintain and ship several DTs depending on the
> option, yes, sure.

Alright, I will try to go as far as I can with the firmware based
approach. As SCPI services are requested by the OS, we can even use the
very same firmware code, on a sunxi-ng based system SCPI will just never
be used and I will try my best to keep it compatible.

Thanks!

Cheers,
Andre.

-- 
You received this message because you are subscribed to the Google Groups 
"linux-sunxi" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to