[Letux-kernel] weird problem with pwm_bl on omap3

[H. Nikolaus Schaller]

Hi,

> Am 30.05.2018 um 08:08 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> On Tue, 29 May 2018 22:03:52 +0200
>> 
> hmm, your latest backtrace shows letux-specific modules, so maybe you are still using them from older compiles?

seems as if this has been a significant hint.

I was indeed missing the --delete option in the rsync. This had the result that even after
a clean build the modules from last build did remain on the µSD. If they were not deleted,
they were replaced. But if we deconfigure something (or don't have the source code in the
tree) they stay available...

Now I have a much cleaner setup for debugging.

First results seem to indicate:
* without generic-adc-battery driver compiled or if blacklisted, it boots fine in 10/10 attempts
* with gab driver enabled there are spurious errors - sometimes from gab and once the strcmp(NULL) - a total of approx. 3/10 attempts

So this seems to prove that the pwm_bl is not really the issue but the victim of an issue in the gab driver
because it is probed at almost the same time.

What I do not understand is how the gab driver or iio can influence the pinctrl search of pwm_bl without
other signs of error.

I have added code to test for NULL pointers but it did not trigger.

So let's continue to hunt the ghost :)

BR and thanks for the hint,
Nikolaus