mmikowski | genii: Yes; as I mentioned, I'm almost certain the patch I illustrated above got into the mainline eventually, probably months after the submission. The patched kernel for Ubuntu, however, was available in a few weeks. It might be helped along that laptops for a major system seller were having many customers report the same issue. | 01:45 |
---|---|---|
=== samuel is now known as Swyde | ||
=== giovanni is now known as giovanni_ | ||
giovanni_ | list #g | 11:21 |
giovanni_ | list /? | 11:21 |
giovanni_ | help | 11:21 |
=== giovanni_ is now known as giovanni | ||
=== giovanni is now known as giovanni_ | ||
=== giovanni_ is now known as giovanni__ | ||
BluesKaj | Hi all | 13:30 |
=== giovanni is now known as giovanni__ | ||
=== giovanni__ is now known as giovanni_ | ||
Crell | Hi folks. Odd question. I've a Kubuntu laptop, running 22.10. It's been running for several years, upgrading along the way. In the past few weeks, though, the CPU keeps pegging. (4 core) htop shows all 4 cores running at near maximum, or sometimes just one at maximum and the others almost idle. The CPU fan is running full blast. But | 17:00 |
Crell | htop sorted by CPU usage shows... almost nothing using the CPU. Lots of processes mostly idle. What's going on? Am I misreading the metrics? Is there a phantom process it's not showing me? | 17:00 |
=== giovanni_ is now known as giovanni__ | ||
alkisg | Crell, a screenshot would help when that happens | 17:08 |
Crell | Screenshot of htop? | 17:14 |
Crell | What's a good place to upload that? Since pastebins are for text... :-) | 17:15 |
Crell | alkisg ^^ | 17:24 |
alkisg | Usually imgur.com or similar sites | 17:24 |
Crell | My fear is that it means I have a trojan running in the background, but I don't want to jump to conclusions. | 17:24 |
Crell | Oh, that does work without an account. Nice. https://imgur.com/a/4rqGRQh | 17:25 |
Crell | alkisg Does that tell you anything? | 17:55 |
alkisg | Crell: I think it says one kernel thread running | 17:56 |
alkisg | Check journalctl -fb -pwarning and dmesg for disk issues or overuse | 17:56 |
Crell | journalctl reports | 17:58 |
Crell | bah | 17:58 |
Crell | journalctl reports "Faied to start Refresh fwupd metadata and update motd" about an hour ago, and various "No beacon heard" messags since. Nothing about a disk. | 17:58 |
alkisg | And dmesg? | 17:59 |
Crell | Hm. ots of trying and failing to connect to an AP. (I presume wireless.) | 17:59 |
alkisg | And if you disable wifi, does the CPU usage drop? | 18:00 |
Crell | If i scroll back far enough, I see this: BAR 13: no space for [io size 0x1000] | 18:00 |
Crell | Just disconnected from the AP. (I'm on a wired connection but it keeps connecting to wifi anyway.) Let's see if it calms down. | 18:01 |
Crell | (I wish dmesg had timestamps.) | 18:02 |
alkisg | It does | 18:02 |
alkisg | man dmesg for all the display options | 18:02 |
Crell | Ah ha. | 18:04 |
Crell | CPU hasn't changed yet. One core is pegged, the others are low, but the fan is at full blast. | 18:05 |
Crell | From dmesg, it looks like the system had a really hard time connecting to wifi about an hour ago. (That's shortly after I got to the computer today.) I'm not sure how to interpret the rest of this, but I don't see anything else suspicious about the disk other than the "no space for io" line above. | 18:06 |
alkisg | Try to disable wifi instead of disconnecting (from network-manager applet, or from nmcli) | 18:13 |
Crell | Disabled it from the gui. Let's see what it does. | 18:14 |
Crell | So far no change. | 18:16 |
Crell | It still has one pegged core, a load average of about 3, and htop shows all processes using 0% CPU. | 18:16 |
Crell | I just don't understand how those numbers add up. | 18:16 |
Crell | Oh! Well, this is interesting. If I use top, not htop, it shows kworker/0:3+kacpi_notify sitting at around 63% CPU usage, consistently. Other things (Firefox) occasionally peak, but that's pretty consistent. | 18:18 |
Crell | And firefox jumps seemingly entirely independently of whether I'm even touching the browser... | 18:18 |
Crell | So what is kacpi_notify and what is it doing? | 18:20 |
Crell | Hm, this looks relevant: https://unix.stackexchange.com/questions/588018/kworker-thread-kacpid-notify-kacpid-hogging-60-70-of-cpu | 18:26 |
Crell | Huh. Yep. The gpe6D interrupt is just freaking out and firing constantly. | 18:29 |
Crell | Which is probably causing all kinds of kernel-mode switching. | 18:30 |
Crell | Ah ha! And if I disable that interrupt as one of the comments suggest, CPU usage drops and my fan chills out. | 18:35 |
Crell | No idea if this is a permanent fix or if it will come back tomorrow, but this is definitely progress. Thanks for the rubber ducking, alkisg! | 18:37 |
alkisg | 👍️ | 18:37 |
Guest76 | Hi... im trying to install Kubuntu onto a lenovo T470s.... having trouble with it now boot looping | 23:08 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!