On 21 January 2015 at 15:17, Lukasz Majewski [off-list ref] wrote:

In previous versions I've only checked for cpu 0.

If you think that it is enough to explicitly check only for cpu 0 and
forget about above "fail safe" code (when. e.g. CPU3 has defined
cooling-cells), then I'm fine with it.

I don't know what bindings are you following, but cpufreq-dt's bindings
say that it has to be present in cpu0. Anyway, this driver isn't for a
multi-cluster system and so cpu0 should be fine.

As I've mention - it would be maintainer's call if one trades potential
regression for patch separation.

I am just asking it to split into a separate patch, not that I will
get it through
cpufreq. Eduardo can take it, but it should be a separate patch.