1

I've got kind of a huge problem in my Ubuntu 16.04.

Yesterday I turned the computer on, and was working in my stuff when a message where "some important updates will be installed" appeared. Innocently, I selected the "yes" option.

Then the problems started to appear. Mendeley desktop failed to start, so I ran it from terminal an got this kind of error:

Failed to create OpenGL context for format QSurfaceFormat

I didn't pay too much attention to it, thinking it was a problem of the program itself.

But today I couldn't boot the computer. It normally starts with the BIOS screen (The Ultimate Force is its name), then a gray screen 3 seconds, then a quick list of processes, most of them developed as a green [OK] (thus some got a red [FAILED]) and then boots.

This time after that screen, it appeared another short list where the first 2 lines were the following:

Stopping NVIDIA Persistence Daemon...
[ OK ] Stopped NVIDIA Persistence Daemon.

Then screen turns black. But I could get to tty1 using Ctrl+Shift+F1.

As of it, I looked for information in forums like this one, and tried a solution of the following link: Ubuntu 16.04 nvidia drivers don't work from which I followed the next codelines:

sudo apt-get purge nvidia-*
sudo apt-get install intel-microcode
sudo apt-get install intel-gpu-tools

Reboot

sudo apt-get install nvidia-yyy

Where yyy were any of 396, 390, 384, or 375. Nothing happened, still same issue.

But when yyy was any of 304 (nvidia-current version seems to have stopped in 304) or 340, error changed and the process list I mentioned before, starts saying something like this:

[ FAILED ] Failed to Start Snappy Daemon
A start job is running for ...

Where ... didn't show well what it it was exactly doing, but it were multiple things including, of course, Snappy Daemon (which truly I don't have idea what it is). That kept looping with some weird number code and didn't let me get to any tty. So I was forced to restart manually the computer and follow these instructions: How do I boot into a root shell? where from I could uninstall those old NVIDIA drivers and reinstall the newer ones (installed 384, the one I had before the issues) to, at least, access the tty easily.

Viewing no advance, I tried the next instruction:

sudo apt-get install nvidia*

Which told me that couldn't proceed because "I held broken packages", but something called my attention: ALL the nvidia drivers were informed to have conflict with something called xorg-driver-binary, while the newer had conflict with nvidia-smi and nvidia-persistenced, from which my logic is that somewhere there was the specific problem that I had at the beginning.

Then maybe I did something really dumb in my desperation. Seeing another 2 conflicts, that I show here:

nvidia-libopencl1-384 : Conflicts : libopencl
nvidia-opencl-icd-384 : Conflicts : nvidia-opencl-icd

I thought it was a good idea to install manually those components:

sudo apt-get install nvidia-libopencl1-384
sudo apt-get install nvidia-opencl-icd-384

Called my attention that they were not installed before. So I did it, rebooted, and the error changed. This time after the quick process list screen, it appeared another short list where the last first and last lines are the following:

Starting Thermal Service Daemon...
...
[ OK ] Started Update UTMP about System Runlevel Changes.

So i followed 16.04 - system hangs at boot time, installing xserver-xorg-video-intel, with no results.

Important data of output of inxi -b are more or less the following:

  • Kernel: 4.13.0-39-generic x86_64
  • Machine model SABERTOOTH 990 FX R2.0
  • Graphic Card: NVIDIA GM204 (GeForce GTX 970)
  • Graphic Display Server: X.org 1.18.4
  • Graphic Driver: FAILED: nouveau

So then i did the following:

sudo apt-get install *nouveau*

And error persisted, with the difference that inxi -b or inxi -Gx didn't tell anymore that nouveau driver failed.

So after all that, I think somewhere there is the main problem (Nouveau as driver instead of the nvidia installed ones), but I don't know how to change that.

Also, I can't turn off Secure Boot mode as other posts suggest because ASUS UEFI BIOS does not have that option (just has "Windows", or "Other Systems"). Neither I can post the real output of some commands since I don't know how to copy them from the damaged computer to this one in which i'm typing this.

I'm giving up for now, but I'm worried because I have important files in this computer and I wouldn't want to do a 16.04 fresh install, since I read that wasn't a sure way to solve the problem.

muru
  • 197,895
  • 55
  • 485
  • 740

1 Answers1

0

Do You have cuda toolkit package installed? Purge it completely as well as any nvidia drivers and install from https://launchpad.net/~graphics-drivers/+archive/ubuntu/ppa version 390.59, it's long supported driver with cuda precompiled in it so no extra cuda toolkit installation is needed....and You showed some conflicts with files which are included in the likes of nVidia 390.59 driver...if this won't work do reverse action and switch to nVidia 396.24 + additionally instal cuda toolkit after that, maybe there will be no conflicts. If You have Intel + nVidia and still nVidia installed so You have e.g. nvidia prime package try to get into root and type in terminal mount -o rw,remount / (and optionally mount --all) to mount partition (or all partitions mentioned in /etc/fstab) in read/write mode and then prime-select intel and reboot. If You can get into ubuntu without root mode then just type sudo prime-select intel and reboot. After that You have safe use of ubuntu under Intel and can fix nVidia from that point. Prime select has: prime-select intel prime-select nvidia prime-select query use.

PawełG
  • 182