You are not logged in.
Pages: 1
Hello:
I have an up-to-date ASCII rig with a pair of Nvidia cards running three monitors on Xinerama.
With the latest kernel update I began to briefly get some artifacts on two of the screens when the desktop started.
Not a big deal but sort of annoying and we all know how this goes ...
So ..
What did I do?
As I did not have these artifacts in my PCLOS rig (exact same machine/hardware, just different drive) I checked the proprietary drivers' versions and sure enough, the PCLOS version is 340.107 and the Devuan version (was) 340.102 and I said to myself: "yep, this is it".
One thing I evidently did not take into account at the time was that my PCLOS setup's kernel (4.14.51):
[groucho@groucho ~]$ uname -a -p
Linux groucho 4.14.51-pclos1 #1 SMP Sat Jun 23 22:41:29 CDT 2018 x86_64 x86_64 x86_64 GNU/Linux
[groucho@groucho ~]$
Not the same as Devuan's ...
But I went and upgraded the driver via Synaptic but when the warnings showed up it was too late and I did not know how to back out of the dark alley I had wandered into. 8^/.
The result is that now I cannot start X.
I'd very much appreciate if some kindred soul would give me a hand at straightening this out, preferably ending up with the 340.107 Nvidia non-free drivers or at least the ones I had installed and worked perfectly well.
Thanks in advance,
A.
Last edited by Altoid (2018-08-15 13:26:22)
Offline
If the kernel is the issue, maybe you should just install the 4.17 kernel from ascii-backports. You might have to reinstall the nvidia driver, but that shouldn't be a problem.
Offline
I may be worthwhile to see if there is any log output for X. On my system, these are still in /var/log in the form of Xorg.n.log - where n is the virtual console number. When screwing around with elogind & co., I found a log in ~/.local/share/xorg/Xorg.3.log. So ~/.local/share/xorg also may be a candidate log file location.
One might also grep around in /var/log/kern.log to check for kernel level complaints, and maybe /var/log/syslog for user space complaints.
Hope this is helpful.
Offline
Hello:
If the kernel is the issue, maybe you should just install the 4.17 kernel ...
I thought that the driver version would be the issue (340.102 in Devuan vs 340.107 in PCLOS), so i went about updating the driver the wrong way and that got me a damaged X.
... have to reinstall the nvidia driver, but that shouldn't be a problem.
Well ...
That was exactly the problem I was in: stuck in tty1 without much idea as to how to proceed, hence my OP.
I eventually figured it out.
groucho@devuan:~$ dpkg --list | grep -i nvidia
... gave me a screen with any and all the nvidia stuff in my Devuan installation.
Then I made sure that /etc/apt/sources.list would not have ascii contrib and ascii non-free enabled.
[root@devuan groucho]# cat /etc/apt/sources.list
## package repositories
# Changed - 20180619
# deb http://pkgmaster.devuan.org/merged/ ascii main
# deb http://pkgmaster.devuan.org/merged/ ascii-security main
# deb http://pkgmaster.devuan.org/merged/ ascii-updates main
# deb http://pkgmaster.devuan.org/merged/ ascii-backports main
deb http://deb.devuan.org/merged/ ascii main
deb http://deb.devuan.org/merged/ ascii-updates main
deb http://deb.devuan.org/merged/ ascii-security main
# deb http://deb.devuan.org/merged/ ascii-backports non-free contrib main
# deb http://deb.devuan.org/devuan/ ascii-proposed main
# added x nvidia non-free installation
# deb http://deb.devuan.org/merged/ ascii contrib
# deb http://deb.devuan.org/merged/ ascii non-free
Once that was done, I did some cleaning up:
[root@devuan groucho]# apt-get clean
[root@devuan groucho]# apt-get autoclean
Reading package lists... Done
Building dependency tree
Reading state information... Done
[root@devuan groucho]#
[root@devuan groucho]# apt-get purge
Reading package lists... Done
Building dependency tree
Reading state information... Done
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
[root@devuan groucho]
[root@devuan groucho]# apt-get check
Reading package lists... Done
Building dependency tree
Reading state information... Done
[root@devuan groucho]#
Then I went about uninstalling everything that had an nvidia label.
This is just one of the files:
[root@devuan groucho]# apt-get purge xserver-xorg-video-nvidia
Eventually, the list shown by dpkg --list | grep -i nvidia came up empty.
To be on the safe side, I did apt-get check again and verified that I had not broken anything else. 8^/
I then enabled http://deb.devuan.org/merged/ ascii contrib and http://deb.devuan.org/merged/ ascii non-free in etc/apt/sources.list.
Without that, I would not be able to find the non-free drivers I wanted to install.
As I did not know the exact name/version I would have available, I asked:
[root@devuan groucho]# apt list | grep -i nvidia-legacy-340 | more
In the list I saw that there was an updated version, albeit not 340.107 (340.106), but closer than the previous 340.102 version.
So I installed it via apt-get install and that was it. The only thing I found odd was that at some point of the installation it listed as recommended nvidia-legacy-340xx-driver-libs-i386 but it was not available in in the repositories.
I thought it was odd because of i386.
On reboot X started as usual but the artifacts I mention in my OP are still there, so I'll have to see if there's some setting in Xorg.conf that may be causing them.
As having http://deb.devuan.org/merged/ ascii contrib and http://deb.devuan.org/merged/ ascii non-free enabled in etc/apt/sources.list as a default option is not a good idea, I commented them out again.
This is how I recall the recovery process, hopefully I have not skipped anything important.
It may be useful for someone who ends up in the same situation.
The lesson I learned:
All -contrib and -non-free stuff should be updated by hand and individually, just like when they were installed.
Cheers,
A.
Last edited by Altoid (2018-08-14 20:57:40)
Offline
Pages: 1