I don't want to duplicate the post, but since I am using Slackware and installing the Slackware packages, I think it is relevant to post here a link to the original one.

The thing is that since Kernel 6.1.39 one of the cores of my CPU is always running at 100%, and it goes back to normal as soon as I ddowngrade to kernel 6.1.38.

All the details in the post at Kernel sub-forum.

Thanks!
//Jesus M

Try to find the process that eat the CPU, for instance with top or htop.

That's one of the mysteries ... none

Set htop to show all threads, including kernel threads (which it hides by default).

1 members found this post helpful.

Can't see anything obvious.

EDIT: rrdtools CPU use chart added

Attached Thumbnails

   

If you want to see processes sorted by CPU usage, you shouldn't use the tree view (you can switch the view with 'F5')

No-tree

Attached Thumbnails

 

Htop shows low CPU usage, the process consuming most CPU power is htop itself which consumes 3.2% of one of your four CPU cores (probably core 3).

That "Total CPU Use" graph, is that ksystemstats? According to htop ksystemstats only consumes about 0.6% of a CPU core (probably core 1 or 2).

Then we have chromium and konsole, each consuming 0.6% of a core (probably core 1 or 2).

At this time, nothing is using core 0 at all.

So your CPU cores are used to 0% + 1.3% + 0.6% + 3.2% which sums up to 5.1/4 = 1.3% of your entire CPU capacity.

Yet your "Total CPU Use" graph claims that you are using 25.1% of 100% capacity. My guess is that this "Total CPU Use" graph itself needs to do some heavy work when it samples the CPU usage.

However, there is one more discrepancy. Your load average is 0.71. The load average is the sum of processes waiting for CPU and processes waiting for disk. If you load average only consisted of processes waiting for CPU your cores should have an average use of about 18%. If you had some process waiting for disk, you should see a "D" in the "S"(tatus) column of htop for that process, but as the load is less than 1 you might have sampled when the process wasn't waiting for disk.

But if you really had a process claiming all the capacity of a core you would have a load average above 1.

regards Henrik

It seems to be a bug, as commented by @lvm_ here

2 members found this post helpful.