On Friday 13 Mar 2020 at 16:49:24 (+0000), Lukasz Luba wrote: <snip>
Well if someone would add EM to its platform and call this in hotplug, which is used as cooling method, will see a lot of warnings.
Right, but I guess I was arguing that calling this for CPUs, even on hotplug, is kinda wrong.
I would rather avoid stressing people with this kind of warnings. This is under control and nothing really happens even when they do hotplug very often, like LTP stress tests.
I agree to add a print there but warning for me is when something is not OK and should be investigated. I would prefer dev_dbg_once() to print thet the EM is not going to be removed. This will also not pollute dmesg in many logs.
Fair enough, a WARN is maybe a bit over the top. A debug message should work.
<snip>
So these small changes will be present in v5. I have to wait a few days because there is one change to devfreq_cooling.c queuing and I will send v5 with updated patch 3/4 rebased on top.
Sounds good, thanks. Quentin