You are not logged in.
Hi,
After the latest kernel update, my Thinkpad P1 is constantly on fire. It seems that, according to htop, one of the cores is constantly running at around 82%. I find this quite shocking, considering that my the output of ps aux | sort -k 3 tells me no process is using such resources:
gavin 2852 1.1 1.1 2508428 185764 ? Sl 23:33 0:09 /usr/lib/firefox/firefox -contentproc -childID 7 -isForBrowser
-prefsLen 7769 -prefMapSize 230880 -parentBuildID 20200824164321 -appdir /usr/lib/firefox/browser 2230 true tab
root 2093 1.7 0.5 156572 93980 tty1 S 23:16 0:32 /usr/lib/Xorg -nolisten tcp :0 vt1 -keeptty -auth /tmp/serverauth.iTLD8UOoQw
gavin 2230 3.8 2.5 3395060 400288 ? Sl 23:21 1:01 /usr/lib/firefox/firefox
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMANDSo, the top process currently running is one from firefox at around 3.8% of a CPU. I have perused my dmesg some and am unable to discern anything that gives away the problem.
I would appreciate some help diagnosing this problem... for the sake of my laptop battery and the now-burnt skin on my legs from my newly fiery laptop. Thanks, kernel update!
In addition, my trackpad has also recently stopped working and I must now use an external mouse. Not sure if this is a hint as to what's going on here.
Offline
You can see "kernel threads" in htop if you type "K" (shift+k), maybe one of those is going crazy.
Strange CPU usage that's not visible in the process listing can happen if a device and its driver is misbehaving. The device can send a lot of interrupts and cause heavy load on the CPU. You can check on the interrupts by looking into the file "/proc/interrupts". Run the following command line (make sure your terminal window shows all lines), and see if there's a number somewhere that increases a lot with each second:
watch -n1 cat /proc/interruptsYou will see a hint about the driver module or device name at the end of each line in /proc/interrupts. Try searching for that name and see what discussion you can find.
Last edited by Ropid (2020-09-01 04:16:11)
Offline
Thank you Ropid. This seems to happen stochastically for me, and it's not happening this morning. I will report back once this happens again.
Offline
OK, this seems to be the same issue as encountered in this post. Thank you very much for your help, Ropid. I wouldn't have found this if it were not for your tip. It seems there are a few proposed solutions ranging from blacklisting the designware i2c driver to using a bleeding edge kernel that's in testing.
https://bbs.archlinux.org/viewtopic.php?id=252711&p=2
This kernel change potentially fixes this. It seems that the broken touchpad is indeed related to this problem.
Offline
I've seen that i2c designware thingy mentioned in several threads. From what I can remember, the people that tried blacklisted that driver didn't report any problems. The touchpad of their laptop would still work fine with it blacklisted.
Offline
Yeah, since there was some talk about this potentially screwing with power management on intel CPUs, I ended up solving this by just installing dkms, linux-lts-headers, and linux-lts, which preserved functionality for my nvidia driver. I had attempted to build the linux-mainline from AUR which has the fix for the bug causing this issue, but found it won't compile with nvidia stuff. Darn.
Anyways, thanks again.
Offline
As a followup, it seems that using the LTS kernel only temporarily fixed this issue, and it's back now. Darn.
Offline
If you didn't find how to blacklist it, it should be this kernel parameter here:
initcall_blacklist=dw_i2c_init_driverThe normal blacklisting through a modprobe config file does not work because the driver is built into the kernel and is not a module file.
Offline