I wouldn't say that elogind is broken.

Most likely elogind works perfectly in Gentoo and Devuan, operating systems under which it is developed and for which it is developed. It seems that the elogind developer does not even have problems with the communication between elogind and NetworkManager.

Personally, I would say that there are compatibility issues between the latest versions of elogind and Slackware. Which problems will surely get worse if Slackers stay in old versions of elogind.

I hope we don't find out that elogind is no longer compatible with Slackware, while we wait for the Devuan developers to fix it for us.

2 members found this post helpful.

Good point! Just tried rebuilding polkit-124 with elogind.252.23 installed, and yes: power management and shutdown menu now works as they should.

Gentoo repository offers elogind-252.9. Devuan indeed offers 255.4.1, but whether it works is anybody's guess.

By fortune, the stock package of NetworkManager is built with "session-traking=consolekit" (which is not ConsoleKit2, but that old abandoned ConsoleKit) and if you add/change to "suspend-resume=upower", then NetworkManager is completely disconnected from elogind.

If it still doesn't work, it means that there are another problems in Slackware-current, beyond this elogind 255.4 vs. NetworkManager 1.46.0

In another way, can you explain to me why a NetworkManager that is completely disconnected from elogind does not work correctly?

@ZhaoLin1457: Unfortunately, building NM with the upower option is not a viable solution, see https://www.linuxquestions.org/quest...7/#post6496534

I haven't dug further into this so I don't know why this happens.

I'm not contradicting you. What I want to say is that your statement is a clear proof that there are other compatibility problems in Slackware-current, beyond this elogind.

Which compatibility problems, they probably have have reached a critical mass lately and we have come to notice them widely and pay more attention to them.

Maybe git bisect elogind v252.23 ... v255.4 is a good idea ...

I'm on Slackware-current, and have that power issue. Elogind-255.4-r2 is on github at least, but I can't build it because it wants auditd . What's that?

EDIT: I read somewhere (Github issues or comments?) that some issue with elogind-255.4 was leading to orphaned processes and 255.4-r1 was supposed to fix that.

If Memory serves, I didn't have the power-off issue on Current of 2024-03-24. A large amount of stuff has changed, so that's not diagnostically useful. But if this issue gets narrowed down to a dozen packages or so, I can still downgrade them, as I still have the 2024-03-24 iso on disk.

elogind-255.4_r2 is already in -current.

Which power issue? If it's about network not going up after suspend, upgrading to v255.4-r2 doesn't change a thing.

If "power issue" is about broken power management and the disappearing shutdown menu (mate, xfce, maybe KDE too) there are three possible solutions:
- use the latest elogind-255.4_r2 and polkit-124 from the -current repo (but then network breaks after suspend/resume)
- downgrade elogind to v252.23 and polkit to v123
- downgrade elogind to v252.23 and then rebuild polkit-124

2 members found this post helpful.

Someone has found something:
https://github.com/elogind/elogind/i...ent-2060986639

Hopefully a solution is just around the corner.

The culprit in elogind seems to have been identified, so maybe we can expect a 255.4_r3 before long:
https://github.com/elogind/elogind/i...ent-2061221562

1 members found this post helpful.

someone
networkmanager listens to elogind signals on dbus to handle sleep/resume.
elogind sends the "sleep" message correctly, but not the "resume" message.
So, after resume networkmanager doesn't bring up the network again, and needs to be restarted.
The same problems applies to other networkmanager-enabled applications, eg. Firefox and Thunderbird.

the problem in elogind is caused by a patch introduced to fix another problem. 252.23 is good, any new version is broken (252.23-r1, 255.4-*)

6 members found this post helpful.

And how you explain the issues with the initial release of 255.4 (without ulterior patches) ?

I for one I believe that an explanation could be that this issue introduced by that patch is in fact the second one (but having a similar misbehavior) , as speculated by Zhao Lin there:

https://www.linuxquestions.org/quest...ml#post6496630

For what is worth, I have just tested the elogind-255.4 (without -rX) and it presents a similar flawed behavior as its successors.