(In reply to Jonathan from comment #265)
> (In reply to Kai-Heng Feng from comment #263)
> > If there's no noticeable power consumption increase when Package C6 gets
> > disabled, and we are sure that it really workarounds the hard lockup, I'll
> > send the patch to upstream.
>
> What exactly would this mean (Linux newbie here)? Would this mean this would
> be active by default on Linux (C6 disabled for Ryzen/TR/Epyc)? Because
> electricity is bloody expensive here and the 5-10% power consumption
> difference would matter to me.
Kai-Heng Feng seems to mulishly ignore the fact, that not *all people* need this workaround. I repeatedly wrote, that this *must be switchable*, as opt in e.g.
But before anything is done, it must be proven in general, that the fix is correct at all - means, that it really addresses C6 package and nothing else.
My tests here have been shown, that there isn't any difference between enabled / disabled C6 package on voltages using zenstates.py. Did it really address the C6 package? But I'm not sure, if it does the same as the kernel patch.
(In reply to Jonathan from comment #265)
> (In reply to Kai-Heng Feng from comment #263)
> > If there's no noticeable power consumption increase when Package C6 gets
> > disabled, and we are sure that it really workarounds the hard lockup, I'll
> > send the patch to upstream.
>
> What exactly would this mean (Linux newbie here)? Would this mean this would
> be active by default on Linux (C6 disabled for Ryzen/TR/Epyc)? Because
> electricity is bloody expensive here and the 5-10% power consumption
> difference would matter to me.
Kai-Heng Feng seems to mulishly ignore the fact, that not *all people* need this workaround. I repeatedly wrote, that this *must be switchable*, as opt in e.g.
But before anything is done, it must be proven in general, that the fix is correct at all - means, that it really addresses C6 package and nothing else.
My tests here have been shown, that there isn't any difference between enabled / disabled C6 package on voltages using zenstates.py. Did it really address the C6 package? But I'm not sure, if it does the same as the kernel patch.