Hi Dirk, On Sun, Aug 02, 2020 at 03:44:09PM +0200, Salvatore Bonaccorso wrote: > Control: tags -1 + moreinfo > > Hi Dirk > > On Sun, Aug 02, 2020 at 10:00:27AM +0200, Dirk Kostrewa wrote: > > Package: src:linux > > Version: 4.19.132-1 > > Severity: normal > > > > Dear Maintainer, > > > > after booting the kernel 4.19.0-10-amd64, there is a kworker process running > > with a permanent high CPU load of almost 90% as reported by the "top" > > command: > > > > $ top > > top - 09:48:19 up 0 min, 4 users, load average: 1.91, 0.58, 0.20 > > Tasks: 218 total, 2 running, 216 sleeping, 0 stopped, 0 zombie > > %Cpu(s): 0.8 us, 12.4 sy, 0.0 ni, 84.5 id, 0.0 wa, 0.0 hi, 2.3 si, 0.0 > > st > > MiB Mem : 15889.4 total, 14173.1 free, 889.3 used, 827.0 buff/cache > > MiB Swap: 0.0 total, 0.0 free, 0.0 used. 14677.7 avail Mem > > > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > > 64 root 20 0 0 0 0 R 86.7 0.0 0:47.41 > > kworker/0:2+pm > > 9 root 20 0 0 0 0 S 20.0 0.0 0:08.84 > > ksoftirqd/0 > > 364 root -51 0 0 0 0 S 6.7 0.0 0:00.50 > > irq/126-nvidia > > 1177 dirk 20 0 2921696 122848 94268 S 6.7 0.8 0:02.23 kwin_x11 > > 1 root 20 0 169652 10280 7740 S 0.0 0.1 0:01.56 systemd > > 2 root 20 0 0 0 0 S 0.0 0.0 0:00.00 kthreadd > > ... > > > > The expected result after booting the kernel 4.19.0-10-amd64 is a kworker > > process with a CPU load close to 0%. > > > > As a control, booting the previous kernel 4.19.0-9-amd64 does not show a > > high CPU load for the kworker process. Instead, the kworker CPU load > > reported by the "top" command is 0.0%. > > > > Therefore, I suspect a bug in the kernel 4.19.0-10-amd64. > > > > Neither "dmesg" nor "journalctl -b" show any messages containing "kworker". > > > > I am using Debian/GNU Linux 10.5 with kernel 4.19.0-10-amd64 and libc6:amd64 > > 2.28-10. > > > > If you need more information, I would be happy to provide it. > > To find out what could be the cause, could you have a look at > https://www.kernel.org/doc/html/latest/core-api/workqueue.html#debugging > this could help determining isolating why the kworker goes crazy.
Please as well to the above one additional thing: Can you reproduce the issue when the kernel does not get tained? So without loading the propriertary, out-of-tree modules. This is particularly important if the issue can be tracked down, found in upstream and needs to be reported upstream. Regards, Salvatore
