No "nvidia_profiles" at uninstall

Hello,

I have install 390.25 driver on Mint 18.03 (Ubuntu 16.04) kernel 4.13.0-32 by pulling script

then I’ve install CUDA

sudo apt install nvidia-cuda-dev nvidia-cuda-toolkit

At next reboot the driver was not working.
As the CUDA uninstall did not give the driver back I decided to backup

nvidia-xconfig --restore-original-backup

But no backup
I decided to uninstall to do fresh installation

sudo apt-get purge nvidia*
Lecture des listes de paquets... Fait
Construction de l'arbre des dépendances       
Lecture des informations d'état... Fait
E: Impossible de trouver le paquet nvidia_profiles

(unable to find nvidia_profiles)
Over installation give

Unable to load the 'nvidia-drm' kernel module

How can I uninstall properly ?
nvidia-bug-report.log.gz (116 KB)
nvidia-bug-report.log.gz (108 KB)

First : use your distros driver manager to update. Add garaphics ppa.

Second: Try with 4.15 kernel , it will work. To get 4.15 kernel , install Ukuu.

Thanks for help VulkanGuy

The Mint’s driver manager for the 384.111 and 387.34 Nvidia driver on 4.11 then 4.13 kernel have let Nouveau driver work. That’s why I’ve use the pulling script, witch have give Nvidia 390.25 working on 4.13 kernel, before installing CUDA have break the installation.

Here is the new installation following your advice:

Kernel 4.15.4 installed with Ukuu (reboot)
Xenial’s graphic driver ppa added
nvidia-390 driver updated with distro driver manager
after reboot:

grep "nvidia" /var/log/Xorg.0.log
[     7.140] (==) Matched nvidia as autoconfigured driver 0
[     7.140] (==) Matched nvidia as autoconfigured driver 2
[     7.140] (II) LoadModule: "nvidia"
[     7.140] (WW) Warning, couldn't open module nvidia
[     7.140] (II) UnloadModule: "nvidia"
[     7.140] (II) Unloading nvidia
[     7.140] (EE) Failed to load module "nvidia" (module does not exist, 0)
[     7.144] (==) Matched nvidia as autoconfigured driver 0
[     7.144] (==) Matched nvidia as autoconfigured driver 2
[     7.144] (II) LoadModule: "nvidia"
[     7.144] (WW) Warning, couldn't open module nvidia
[     7.144] (II) UnloadModule: "nvidia"
[     7.144] (II) Unloading nvidia
[     7.144] (EE) Failed to load module "nvidia" (module does not exist, 0)

I’ve try to blacklist Nouveau in /etc/modprobe.d but it do not give the nvidia driver

Please run nvidia-bug-report.sh as root and attach the resulting tar.gz file to your post (paperclip icon on existing posts).

Looks like you installed the distro package over .run package over distro package. Not advisable.
Please check which package are installed now using
dpkg-query -l nvidia* |grep ii

The .run install can be uninstalled using --uninstall option
blahblah.run --uninstall

./NVIDIA-Linux-x86_64-390.25.run --uninstall

gived:
http://pix.toile-libre.org/?img=1518993754.png

also away from driver manager:
http://pix.toile-libre.org/?img=1518994021.png

Just a sort of control:

apt-get purge nvidia*
Lecture des listes de paquets... Fait
Construction de l'arbre des dépendances       
Lecture des informations d'état... Fait
E: Impossible de trouver le paquet nvidia-cuda-toolkit_9.0.176-2_amd64.deb
E: Couldn't find any package by glob 'nvidia-cuda-toolkit_9.0.176-2_amd64.deb'
E: Impossible de trouver de paquet correspondant à l'expression rationnelle « nvidia-cuda-toolkit_9.0.176-2_amd64.deb »

Huumm… :/

# apt-get clean
# apt-get autoclean
Lecture des listes de paquets... Fait
Construction de l'arbre des dépendances       
Lecture des informations d'état... Fait
# apt-get autoremove
Lecture des listes de paquets... Fait
Construction de l'arbre des dépendances       
Lecture des informations d'état... Fait
Les paquets suivants seront ENLEVÉS :
  lib32gcc1 xserver-xorg-legacy
0 mis à jour, 0 nouvellement installés, 2 à enlever et 1 non mis à jour.
Après cette opération, 392 ko d'espace disque seront libérés.
Souhaitez-vous continuer ? [O/n] o
(Lecture de la base de données... 511109 fichiers et répertoires déjà installés.)
Suppression de lib32gcc1 (1:6.0.1-0ubuntu1) ...
Suppression de xserver-xorg-legacy (2:1.18.4-0ubuntu0.7) ...
Traitement des actions différées (« triggers ») pour libc-bin (2.23-0ubuntu10) ...
Traitement des actions différées (« triggers ») pour man-db (2.7.5-1) ...
#

Am I right to look for some nvidia.conf yet ? Where can they be hidden ? :)

Sorry I forget

before #6:

# dpkg-query -l nvidia* |grep ii
dpkg-query: aucun paquet ne correspond à nvidia-bug-report.log.gz
dpkg-query: aucun paquet ne correspond à nvidia-bug-report.sh
dpkg-query: aucun paquet ne correspond à nvidia-cuda-mps-control
dpkg-query: aucun paquet ne correspond à nvidia-cuda-mps-server
dpkg-query: aucun paquet ne correspond à nvidia-debugdump
dpkg-query: aucun paquet ne correspond à nvidia-detector
dpkg-query: aucun paquet ne correspond à nvidia-xconfig
ii  nvidia-settings     390.25-0ubuntu0~gpu16.04.1 amd64        Tool for configuring the NVIDIA graphics driver

After #6:

# dpkg-query -l nvidia* |grep ii
dpkg-query: aucun paquet ne correspond à nvidia-cuda-toolkit_9.0.176-2_amd64.deb

More cleaning ?

No, I don’t think so. You probably issued the command being in a directory where the file
nvidia-cuda-toolkit_9.0.176-2_amd64.deb
is saved. dpkg takes local files matching the parameter nvidia* into account. That’s where you got those strange ‘missing’ messages from. Just reinstall the driver now from ppa, reboot, and attach a new nvidia-bug-report.

ok

Installation Nvidia-390 driver with graphic tool in LinuxMint 18.3 kernel 4.15 made: not working, Nouveau not blacklisted (sometime it is).
Until now graphic tool did not give wait results to me. The only successful installation happen with pullingscript on 4.13

make.log (17.4 KB)

I don’t know what’s going on with Mint and the driver packages, VulkanGuy had the same problem, the kernel driver doesn’t compile. Please attach the file
/var/lib/dkms/nvidia/390.25/build/make.log

I don’t really know what exactly you’re doing, but the logs just tell everything is fine. You’re not creating the logs when everything works, do you?
Just for info, I fetched my old notebook that’s running Mint, upgraded it to Mint 18, added the ppa, installed libcuda1-390, nvidia-390 everything went fine, driver installed.
Then installed Ukuu, and with every kernel, the nvidia 390 driver from ppa just installed fine.

First thanks for all what you make cause I’m conscient that reading sooo looongs logs is for sure time…
Second I’m far from the machine for this evening :/
Third like you can read me I’m absolutly amazed about what you say O_o

I have install like I did so much time with the linuxMint driver manager yesterday and reboot. It just appear to myself that I was still using Nouveau (just appearance, something with fluidity) but this is absolutely right that when I add here “make.log” I did no supplementary control before.

Could it be ?

Just I can say nothing until tomorrow evening (I cannot join this shutdown machine) but be sure I’m now in a hurry to confirm or not !

Sorry not yet :/

$ grep "drivers" /var/log/Xorg.0.log
[    13.152] (II) Loading /usr/lib/xorg/modules/drivers/nouveau_drv.so
[    13.153] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[    13.154] (II) Loading /usr/lib/xorg/modules/drivers/fbdev_drv.so
[    13.154] (II) Loading /usr/lib/xorg/modules/drivers/vesa_drv.so

grep -i nvidia /var/log/Xorg.0.log
would be more appropriate.

ok ça calme !

Done
So a job, really ! But nothing so good that learning ;)
Thank you for help