Arch Linux

graysky

Wiki Maintainer

From: :wq

Registered: 2008-12-01

Posts: 10,731

Website

Sporadic poweroff behavior under kernel 4.4.1

Problem: Machine does not reliability poweroff on a shutdown under 4.4.1.
Work-around: Downgrading to a 4.3.x kernel fixes the hung shutdown.

Since updating to the 4.4.1 kernel, my Haswell machine fails to poweroff about 40% of the time.  I don't see anything in `journalctl -b -1` that indicates an error or warning, so I'm thinking whatever is happening occurs after the journal logging as evident by the last few lines lines:

...
Feb 14 11:02:41 quartz systemd[1]: Reached target Shutdown.
Feb 14 11:02:41 quartz systemd[1]: Reached target Final Step.
Feb 14 11:02:41 quartz systemd[1]: Starting Power-Off...
Feb 14 11:02:41 quartz systemd[1]: Shutting down.
Feb 14 11:02:41 quartz systemd-shutdown[1]: Sending SIGTERM to remaining processes...
Feb 14 11:02:41 quartz systemd-journald[186]: Journal stopped

On the screen, I get many lines of the following:

INFO: task shutdown:1 block for more then 120 seconds.
  Not tainted 4.4.1-2-ARCH #1
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
INFO: task kworker/0:3:1642 block for more than 120 seconds.
  Not tainted 4.4.1-2-ARCH #1
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
...
             <<repeats 7 times>>

Anyone else experiencing this under the 4.4.1 kernel?  I'd like to open a report upstream but have no good idea which component of the kernel the report should go against.

EDIT: FS#45863 has the same error (task shutdown:1 block for more then 120 seconds.) and symptoms but mine is cleaner lacking all of the other output in that FS.

https://bugs.archlinux.org/task/48180

Last edited by graysky (2016-02-14 17:02:46)

---

AUR packages • Zsh and other configs

dogabone

Member

Registered: 2016-03-13

Posts: 1

Re: Sporadic poweroff behavior under kernel 4.4.1

Hello graysky, I was having the same problem as you. After much searching, I've done the following to debug.
First, add the options "debug ignore_loglevel" to your kernel line as told in the boot debugging wiki article. Reboot your machine after these options are active and then try to poweroff/reboot (none of them were working for me).

In my case, the last info shown on screen was about the radeon driver. So I blacklisted it and tested many reboots/poweroffs, all of them were clean. If I modprobe the radeon module and try to poweroff, I can reproduce the erratic behavior.

By looking at the log you attached in the bug report, I see that you do not have a radeon card, so please provide more info after turning debug on for your kernel line.

Regards,
dogabone.

EDIT: kernel version 4.4.5-1-ARCH

Last edited by dogabone (2016-03-13 23:46:58)