Thinkpad laptop reboots insted of powerff after suspend

widlokm

Hello,
I have old thinkpad T530 with gentoo and custom build 5.15.26 kernel. Some time ago (I'm not sure when) I've noticed strange issue with poweroff. Laptop reboots whenever I'm switching it off after entering suspend-to-ram, like:
1. Turn on laptop
2. Enter suspend to ram (Fn-F4)
3. Wake up (by pressing power button)
4. Turn off with power button - now laptop performs normal shutdown, but instead of staying OFF it reboots.
5. Turn off again with power button - laptops stays off as expected

Poweroff works perfectly OK without step 2&3, plugging network cable or power supply does not seem to have any effect, "wake on lan" is disabled in BIOS.
This is not very problematic, more inconvenient, but could You give me any suggestions how to solve this. I'm out of ideas.
Thanks,
Michael Widlok

alamahant · Advocate Joined: 23 Mar 2019 Posts: 3950

Hu · Administrator Joined: 06 Mar 2007 Posts: 23103

ACPI soft-off through a brief tap on the power button should be legal, and should gracefully halt, hibernate, or suspend the system, depending on how it was configured.

OP: please describe how you enter suspend-to-RAM state. What Linux actions does Fn-F4 perform?

widlokm · Posted: Mon Apr 04, 2022 8:40 am Post subject:

Poweroff & suspend is controlled by elogind (without systemd). Elogind has one hook script for suspend & resume in "/lib64/elogind/system-sleep/" to shut down and restore one USB to Ethernet interface (connected to raspberry pi), but I don't thing it is a problem. The script just performs "rc-service net.enp0s.... stop" then "c-service net.enp0s..... start" on resume. ACPI (acpid daemon) does not handle suspend or power button, only elogind. I've also tested a few other situations:

1. Kill elogind, manually enter suspend (echo "mem" >/sys/power/state), then shutdown (by command "poweroff").
2. Enter sleep & shutdown by elogind: loginctl suspend, loginctl poweroff.
3. Enter sleep & shutdown with and without Xorg running.
4. Enter sleep by Fn-F4, shutdown by "poweroff" command.

The result is always the same, if sleep to ram was used laptop reboots. Now I suspect a kernel / kernel configuration problem, but I might be wrong. Should I provide kernel log or something? For me they look perfectly normal. By the way, I use custom kernel compiled specially for this laptop, without initrd. Should I post kernel config?
Thanks,
Michael W.

Hu · Administrator Joined: 06 Mar 2007 Posts: 23103

I once had a system where, due to a configuration omission on my part, if it had used suspend-to-both, future uses of suspend-to-disk (or maybe it was suspend-to-ram - I don't recall) acted as suspend-to-both. It was caused by the hibernate script not fully setting all the various parameters. For your case, please show the output of grep --directories=skip -e . /sys/power/*.

widlokm · Posted: Tue Apr 05, 2022 7:30 am Post subject:

Here they are:

deagol · n00b Joined: 12 Jul 2014 Posts: 62

I've a Thinkpad W530, which should from a HW perspective pretty close to what you have.
And shutdown after sleep is working normal here, running ~amd64.

Quite some time ago (> one year for sure) I had some issues with powering down. I remember it was not powering off properly but forgot the details.
I figured out that this was related to using the i915 driver. I switched over to i965 which fixed the issue for me. So when you are still using i915 I would give that a shot.

widlokm · Posted: Tue Apr 05, 2022 10:11 am Post subject:

Yes T530 and W530 are very similar, I will be switching to W530, but only as a "desktop" machine, where suspend is not needed.

In make.conf I have: