Re: [gentoo-user] kworker using 100% cpu

[Mick]

Hi Andreas,

On Monday, 4 March 2019 19:44:19 GMT Andreas Fink wrote:
> Hello,
> I have a problem which uses 100% of one cpu core on my HP-15-bs114ng
> notebook.

I had come across the same problem on a mid-2014 MacBook Pro:

https://archives.gentoo.org/gentoo-user/message/
417dfa5af8e9fb76a66fe4816bdc7c44


> I figured out already that it is somehow related to ACPI
> interrupts, since the following command will return the system to normal:
> echo disable > /sys/firmware/acpi/interrupts/gpe16

Yes, I had to do the same and repeat after each boot.  I can't recall if I 
added this in a script so I didn't have to run it manually each time.


> The content of this "file" is the following:
> 3749821     STS disabled     unmasked
>
> The first column is the number of interrupts, that happened, which is very
> high, hence the 100% cpu usage in one kworker process.
>
> My question would be the following now:
> 1. What could be the side effects of disabling this interrupt?

I did not discover any side effects.  The CPU would return back to normal and 
the overheating problem went away.  This does not mean some key functionality 
was not affected, but I never discovered anything relevant.  I did not debug 
the kernel at the time to bottom out what exactly caused this.


> 2. What could trigger the interrupt?

I don't know.  I only had this MacBook Pro for a few months, so I never got to 
the bottom of it.


> 3. How can this be debugged further and reported, i.e. who would be able to
> fix it?

Have a look here:

 https://lkml.org/lkml/2011/3/31/144

Then it will need to be reported to BGO and perhaps upstream to kernel devs.


> (This happens on all kernels, since I have the notebook, i.e. >= 4.17)

If your HP notebook's MoBo or CPU are similar to my 2014 MacBook Pro, then 
this has been a problem at least since the 3.x series kernel.

I hope you get to the bottom of it.

-- 
Regards,
Mick