Hi all, There's a problem with hibernating using the power button on at least some notebooks. Using a fresh Artix Linux installation on a new Lenovo ThinkPad E14 Gen 6, I configured and set up Suspend to Disk, and it works as expected: The notebook goes to sleep and wakes up as it should when being powered on again. On my old ThinkPad (E14 Gen 5), I configured hibernation to happen when the power button is pressed. I wanted to do this for the new one, too. Now, something quite strange happens: The default Power Management setting is to show the Logout Screen when the power button is pressed. If I choose “Hibernate” there, everything works: The power LED flashes for a few seconds, then, it stays on a short time, and then, the system turns off completely – as it should. After pressing the power button again, it starts to boot, and then wakes up as it should. Same as when I choose “Hibernate” from the start menu. But when I change the power button press action from “Show logout screen” to “Hibernate” in the Power Management settings and press the button, the process does not work when actually pressing the button: The power LED flashes for a few seconds, then turns on again – and stays on. The system does not turn off. It stays on and becomes unresponsive: I can’t wake it up again. Nothing happens. Screen off, but system on. Only way is to keep the power button pressed to hard-reset the system. After that, it reboots again, and repairs orphan inodes (due to the power loss). At least one discuss user could confirm this very problem on another notebook, cf. https://discuss.kde.org/t/strange-problem-when-hibernating-using-the-power-button/42235/8 Maybe, this is some race condition or whatnot. I currently have no idea how to debug this, but I'll gladly help to do so, if somebody with more insight tells me how. Thanks for looking into this! Cheers, Tobias
Can you share the journal output of when this happens? Something like `sudo journalctl --since -10m`. I know there's a case where suspend fails with NFS mounts because systemd tries to freeze all userspace processes when suspending and that freezing fails, so I'm wondering if something similar is happening here.
I don't use Systemd, my Artix runs on OpenRC. I'll attach a syslog log for a succeeded hibernation (through a mouse click) including a subsequent wakeup, and a failed one (by pressing the power button set to hibernate), including a hard reset and reboot. However, I'm pretty sure the other forum user that confirmed the problem runs Systemd, as most do. I'll ask there if he/she could comment here if this is the case. If I hibernate through a mouse click, hibernation and wakeup works – no matter if an NFS mount is mounted or not. The NFS mount is not re-mounted after waking up though. Vice versa, hibernation through pressing the power button (having it set to hiberate directly) fails, no matter if an NFS mount is mounted or not. So I suppose this doesn't cause it.
Created attachment 189406 [details] Syslog of a failed hibernation
Created attachment 189407 [details] Syslog of a succeeded hiberantion
From the failed log: > Feb 09 14:38:18 [kernel] [ 8861.978945] PM: hibernation: Wakeup event detected during hibernation, rolling back. I wonder if it fails hibernation because it's triggered on button press, then when you release the button it gets a button event and interprets that as a wakeup event. Maybe you can try pressing and holding the button for a few seconds (but not enough to trigger force power-off) to see if the behaviour is different then?
I tried pressing the power button for some seconds, not triggering a hard reset yet. The result is the same (if hibernating is triggered directly by pressing the power button of course). The hibernation process does not start unless I release the button. And then: The power LED flashes for a few seconds, then turns on again – and stays on. The system does not turn off and becomes unresponsive. Exactly the same behavior like pressing it shortly.