For absolutely no reason ubuntu decided to prevent me from logging in suddenly.
I cant login with my user, cant login with the guest user, cant login using ubuntu ui, cant login using gnome. I dont know why gnome is an option at all, i did never install gnome, must have been installed automatically with some update.
I have absolutely no idea what causes the problem, ubuntu keeps telling me nothing, simply firing up the login screen over and over again. Using an older kernel does not work either.
In the recovery mode i cant update packages because dns is not working, resolv.conf points to a non-existing file, so no dns lookups as well.
I search the web and found many issues like mine, that could be solved by (re)moving some Xorg files but i cant event login into a console session. Once i press Ctrl+Alt+F1 the login screen fires up again.
Most problems seem to be related to lightdm+nvidia, i dont have a nvidia product, so that leads to nothing as well.
My system consists of an AMD FX CPU, a AMD Radeon R9 Graphics-Card (i used amdgpu driver), a creative sound-blaster sound card and a NVME insterted via PCIe > M.2 Adapter. Boot partition is located on a standard ssd because the mainboard does not support boot from PCIe.
At the moment i booted into windows 10 (dualboot setup).
What logfiles could be of help to solve that problem? I looked into some files but did not learn anything from those logs.
I am at the edge of pure furstration, please give me some hints, i need to fix this machine asap :-(
Is it me or is ubuntu becoming less stable with every single update? I though sticking with LTS releases would prevent me from repairing my setup every few months but that turned out to be wrong :(
Update
I fixed the resolv.conf manually (created the missing file and added 8.8.8.8 as nameserver), upgraded all packages, removed amdgpu, reinstalled amdgpu and now my keyboard wont work at login as well (usb keyboard connected via the built-in usb-hub in one of my screens). In grub the keyboard works.
After some seconds the pc seems to freeze (the caret does not "blink" anymore).
How did i manage to do that xD
Update
I managed to fix the keyboard problem by doing the following:
- Start into recovery mode
- Make filesystems writeable by checking the discs
- activate network
- goto root terminal
- fix resolv.conf (create /run/resolvconf/resolv.conf and insert "nameserver 8.8.8.8")
- goto recoverymode menu (Ctrl+D)
- select repair broken packages
New kernel was installed, some packages removed, etc.
Then i rebooted and my keyboard worked again - but still no login in gui possible.
Some logs i discovered:
~/.xsession-errors
openConnection: connect: Datei oder Verzeichnis nicht gefunden
cannot connect to brltty at :0
... followed by some upstart errors due to the error above i guess
Sorry for the german output, here the attempt of translating it correctly:
openConnection: connect: File or directory not found
/var/log/lightdm/lightdm.log
Some Debug...
....
CRITICAL: session_get_login1_session_id: assertion 'sesson != NULL' failed
...
Some Debug...
dmesg
...
amdgpu 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0xffff
...
journalctl -xb
...
systemd-fsck [857]: e2fsck: Neuere Version von e2fsck benötigt!
systemd-fsck [857]: fsck failed with error code 8
...
...
NetworkManager[1121]: nm_device_get_device_type: assertion 'NM_IS_DEVICE (self)' failed
...
lightdm[1623]: PAM unable to dlopen(pam_kwallet.so): /lib/security/pam_kwallet.so: cannot open shared object file: No such file or directory
lightdm[1623]: PAM adding falty module: pam_kwallet.so
lightdm[1623]: PAM unable to dlopen(pam_kwallet5.so): /lib/security/pam_kwallet5.so: cannot open shared object file: No such file or directory
lightdm[1623]: PAM adding falty module: pam_kwallet5.so
lightdm[1623]: pam_unix(lightdm-greeter:session): session opened for user lightdm by (uid=0)
...
...
...
gnome-session-binary[11821]: CRITICAL: We failed, but the fail whale is dead. Sorry...
...
UPDATE 3
I finally managed to solve the problem. Unfortunately - the community decided, that this is off-topic, so i have to answer this in the question:
I did lots of things and i am not 100% sure, what solved the problem, so i write down everything. I suspect a few specific steps to have solved the problem and will summarize that at the end.
- uninstall amdgpu-pro driver (there is a binary to achieve this)
- install the open source driver (see here). I also installed all those hwe packages.
- reinstall and reconfigure lightdm, ubuntu-dektop, unity
- repair packages via the recovery menu
So far i did not have any success. Additionally i added my user to the lightdm group, but that did not succeed either.
Then i found other similar issues and tried the following steps:
- add-apt-repository ppa:paulo-miguel-dias/pkppa
- apt-get update
- apt-get upgrade
- reboot
Still no success so far, so i decieded to try another upgrade method
apt full-upgrade
reboot
And then i was able to login again.
My personal conclusion
It seems the problem exists within the amdgpu driver and ubuntu will not be able to install a working driver unless you add the ppa mentioned above. It says to work only with Ubuntu 18.04 but i tried it anyway and it worked for now.
So - to be honest - i guess ubuntu 16 LTS is broken (at the day of 14th of june 2018) vor AMD-Graphics users. Therefore i would recommend to setup a fresh installation based on 18 LTS ASAP.
This is a big con on the reasons-for-and-against-ubuntu list. I may have broken the setup myself by installing amdgpu-pro in the first place but hey - if that breaks the whole operating system we have a bigger problem.
I hope this is of some help to everyone facing the same issue.