Re: [PATCH 5/6] clk: pxa: transfer CPU clock setting from pxa2xx-cpufreq

[Stephen Boyd]

On 10/19, Robert Jarzmik wrote:
> Stephen Boyd <sboyd@xxxxxxxxxxxxxx> writes:
> 
> > On 10/18, Robert Jarzmik wrote:
> >> Robert Jarzmik <robert.jarzmik@xxxxxxx> writes:
> >> Hi Michael and Stephen,
> >> 
> >> I'm planing on sending a v2 next week with minor corrections, mostly in the data
> >> tables (pxa25x_freqs and pxa27x_freqs), as testing prooved some values were wrong.
> >> 
> >> If you want me modify this serie, will you have time to review for next week or
> >> should I delay the v2 posting ?
> >> 
> >
> > No need to delay. clk patches look fine with a quick glance. It
> > would be really neat if we could make cpufreq-dt work without DT.
> > What's blocking that? OPP tables?
> 
> Heu I'm not the author of cpufreq-dt, so I'm not the best to answer.
> To answer the question "without DT", it depends if you mean "with ACPI" or "with
> platform_data" or something else.

I mean platform_data mostly. Do you use ACPI with the clk driver?

> 
> From what I've seen so far, the missing/blocking points are :
>  - the OPP points definition as you said

Hm.. I thought cpufreq-dt worked with OPP tables populated by
other code (i.e. platform code).

>  - probably same thing for the input power supply / regulator

Regulators should be optional I hope. Do you use regulators in
your design that has platform_data?

>  - the cooling parts probably
>  - and more generaly all the cpufreq-dt is built around device-tree
>  - last point, the name from KConfig, "Generic DT based cpufreq driver"
>    => that strongly suggest it's device-tree only
> 
> I'm deeply convinced that Viresh being one of the authors will shed more light
> on this.

Sure, thanks for the notes.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project