After Upgrade 15.6 > 16.0 sddm login returns lgin screen

[hcvv]

I have one system (boven) upgraded to 16.0 last week without much hassle.

To day I try to do the same with another, very similar system `beneden’.

The zypper --releasever 16.0 dup --allow-vendor-change has run without problems.

After re-boot I get the sddm login screen. Filling in the credentials and sending of gives me a short moment a black screen and then the login screen again. Both users the same.
The DE chosen is KDE Wayland, when switching to KDE X11 or XFCE the system freezes and no key nor mouse actions possible (have to press the OFF switch).

Booted into recovery. There are several messages

FAILED failed to start D-Bus System Message Bus

And there are many messages in the log that point to a missing SELinux file in /etc/selinux . But I do not have SELinux on.

The only difference I see between boven and beneden is that the first does not have the file /etc/selinux/conf and the second has it with contents

SELINUX = disabled

Sorry not to have direct copy/paste info because the system is in single-user mode.

It will however switch to multi-user mode and have network connection. I will try to get more info in that way. But in the mean time: any ideas?

[hcvv]

This is very, very strange.

I removed /etc/selinux/config and now it works!

[newsense]

I would rather have the same file and configuration as the system that works, otherwise the next selinux update is likely to recreate the removed file with unknown consequences on reboot

[waspo]

@hcvv

Filling in the credentials and sending of gives me a short moment a black screen and then the login screen again.

I’ve had the same problem under Slowroll with lightdm, but not with sddm.

Deactivating selinux solved the problem:

update-bootloader --add-option 'selinux=0'
update-bootloader

[hcvv]

@waspo

Deactivating selinux solved the problem:

I guess that switching it off like that is better to be sure it does not create trouble anywhere.
Thanks for the hint.

[hcvv]

@newsense

I would rather have the same file and configuration as the system that works,

I do not quite follow you here.

The first system did (and does) not have /etc/selinux/config and gave no trouble.

The second system had the file with the contents

SELINUX = disabled

and gave trouble.

I now removed that second file to make it the same as the first system and the problem is gone.

Thus your advice to have that file would IMHO create the problem on both systems, which is not what I want.

BTW, reading through articles on the web, they say that for switching of the contents should be

SELINUX=disabled

thus no white space. No idea if that white space is doing something and I am not going to test it.

In any case, I will implement @waspo 's advice and switch it off at boot already.

[hcvv]

@newsense

the next selinux update is likely to recreate the removed file with unknown consequences on reboot

boven:~ # zypper se -si selinux
Loading repository data...
Reading installed packages...

S  | Name              | Type    | Version          | Arch   | Repository
---+-------------------+---------+------------------+--------+-----------
i  | libselinux-devel  | package | 3.8.1-160000.2.2 | x86_64 | Main (OSS)
i  | libselinux1       | package | 3.8.1-160000.2.2 | x86_64 | Main (OSS)
i  | libselinux1-32bit | package | 3.8.1-160000.2.2 | x86_64 | Main (OSS)

    Note: For an extended search including not yet activated remote resources please use 'zypper    search-packages'.
boven:~ #

are the only SELinux related packages installed. I think we can forget the 32-bit one. The other one only has the library file. Thus that config file is not part of any package. I have no idea while it was there, but I am not very afraid it being (re-)created during an update.