Arch Linux

keeyan

Member

Registered: 2019-07-01

Posts: 20

[SOLVED] What is causing so many interrupts and how would work it out?

A couple of days ago I noticed a CPU thread is often always at 100% on htop. This seems to be a Kernel thread (since it's red).

After searching a lot I found a couple of similar questions:
Many interrupts caused by trackpad issues from i2c_designware
Interrupts caused when connecting a docking station
Another trackpad issue

So I ran a few commands suggested by those, and I get this result (htop and cat /proc/interrupts):

https://i.ibb.co/8D63YKY/Screenshot-fro … -35-21.png

It seems that "INT3455:00" is where most of these interrupts are coming from.

But this is where I'm stuck. Where do I go from here to work out what "INT3455:00" is (I tried searching for it, no relevant results), what's causing it to trigger so much, and if there is anything I can do about it?

moderator edit -- replaced oversized image with link.
Pasting pictures and code

Last edited by keeyan (2020-12-13 22:30:57)

Ropid

Member

Registered: 2015-03-09

Posts: 1,069

Re: [SOLVED] What is causing so many interrupts and how would work it out?

Try this kernel command line option here:

initcall_blacklist=icl_pinctrl_driver_init

Here is how I researched this:

An "INT3455" text shows up in the "pinctrl-icelake" driver source code:

https://github.com/torvalds/linux/blob/ … -icelake.c

I'm then guessing that's the hardware/driver for that "INT3455:00" line from your /proc/interrupts output.

This pinctrl-icelake module is not an external kernel module file on Arch. It's compiled directly into the kernel image, see the file-name line in the output of the "modinfo" command:

$ modinfo pinctrl-icelake
name:           pinctrl_icelake
filename:       (builtin)
...

To disable a built-in module, you have to keep it from loading with a kernel command line parameter "initcall_blacklist=...". For the name to use for initcall_blacklist=..., I got it from this line here at the end of the pinctrl-icelake.c source code:

module_platform_driver(icl_pinctrl_driver);

That "module_platform_driver()" thing is a macro. It takes that "icl_pinctrl_driver" text and adds several functions to the source code. It creates an init-call function with the name "icl_pinctrl_driver_init".

keeyan

Member

Registered: 2019-07-01

Posts: 20

Re: [SOLVED] What is causing so many interrupts and how would work it out?

Thanks a lot @Ropid, great detective work Hopefully next time I won't have to ask, but reading through the Linux source code would never have even occured to me. I added the Kernel command line option and rebooted. Unfortunately, since it's an intermittent issue I don't know if it's fixed. I'll wait a couple of days and if it doesn't happen again I'll mark the thread as solved

keeyan

Member

Registered: 2019-07-01

Posts: 20

Re: [SOLVED] What is causing so many interrupts and how would work it out?

Yeap that seems to have fixed it Do you think this is a bug worth reporting to them or just some weird quirk of my setup or hardware?

Ropid

Member

Registered: 2015-03-09

Posts: 1,069

Re: [SOLVED] What is causing so many interrupts and how would work it out?

I think this should be worth reporting because the 100% core load is bad. I tried searching for bug entries about "pinctrl" on the bugzilla.kernel.org website and there was no interesting old bug report. Maybe this kind of problem is rare and your bug report would be new? Or maybe there's a lot of people that never noticed and are suffering from this?