[Letux-kernel] Strange things happening with latest kernels

[H. Nikolaus Schaller]

Hi Andreas,

> Am 13.06.2019 um 07:47 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> Hi Nikolaus,
> 
> On Thu, 13 Jun 2019 07:37:46 +0200
> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> 
>> Hi Andreas,
>> great analysis!
>> 
>> Maybe I was lucky with my GTA04A4 where I tried these days.
>> 
>> A theory is that if the opp voltages are not supported, they are not changed and the value initialized
>> by U-Boot is kept. This will be wrong depending on clock speed.
>> 
> Well, the whole frequency-scaling thing seems to be disabled.
> 5.1 seems still to work. I have reasonable percentages for 300Mhz here.
> 
>> I remember there was a discussion these days on linux-omap for the DRA7 about similar regulator issues,
>> but have forgotten the details.
>> 
>> So we should notify Tony about this.
>> 
> will probably write an email this evening about that. Unfortunately my autobuild
> was not active when the things were merged into linus/master. So I have only the
> interesting logs about linux-next

I can confirm that the issue is OMAP3 (twl4030) only and not on the Pyra.
And your patch makes cpufreq-info become available again.

And I have done some more tests and it looks as if power consumption
(battery current if not charging) in idle has gone up from letux-4.19(.0)
to 4.19.44 and again for 4.19.49 by ca. 50mA.

This happens if X11 is disabled or enabled. With X11 I have ca. 50-60% in
800 MHz OPP although the CPU load is below 10%.

Well, it may also be some repeating background activity which may prevent
lower clock in my GUI setup (have used QuantumSTEP and not e.g. LXDE
and have updated some parts in QuantumSTEP recently).

Something which needs deeper inspection to understand the influencing
factors.

BR and thanks,
Nikolaus