NVIDIA drivers - again!

[stevepusser]

Headers to non-backport? What kernel are you using?

If you're using the 4.18 kernel from backports, then you need to install nvidia-driver from backports. The stock version in the main Deban repos will error out when you try to build it against a newer kernel.

I think that would be rather evident if you look at the end of the output of "apt-get install --reinstall nvidia-kernel-dkms" with the standard Nvidia version and the 4.18 kernel. That's why I asked if you had seen the message about blacklisting the nouveau driver you get with a successful build...never answered, either.

And to the world...please stop running inxi as root. Just stop.

[jlarysz]

Thanks for the hint on firmware errors. I think that issue is cleared up.

[jlarysz]

Thanks stevepusser - I went to backport headers because that seemed to worked. I'll roll back again and see what happens with and nvidia-driver build.

And here's what happens with the standard linux headers:

Code: Select all

root@BlueLight:~# apt-get install nvidia-driver
Reading package lists... Done
Building dependency tree
Reading state information... Done
Some packages could not be installed. This may mean that you have
requested an impossible situation or if you are using the unstable
distribution that some required packages have not yet been created
or been moved out of Incoming.
The following information may help to resolve the situation:

The following packages have unmet dependencies:
 nvidia-driver : Depends: nvidia-driver-libs (= 384.130-1) but 390.87-2~bpo9+1 is to be installed or
                          nvidia-driver-libs-nonglvnd (= 384.130-1) but it is not going to be installed
                 Depends: nvidia-driver-bin (= 384.130-1) but 390.87-2~bpo9+1 is to be installed
                 Depends: xserver-xorg-video-nvidia (= 384.130-1) but 390.87-2~bpo9+1 is to be installed
                 Depends: nvidia-vdpau-driver (= 384.130-1) but 390.87-2~bpo9+1 is to be installed
                 Depends: nvidia-alternative (= 384.130-1) but 390.87-2~bpo9+1 is to be installed
                 Depends: nvidia-kernel-dkms (= 384.130-1) but 390.87-2~bpo9+1 is to be installed or
                          nvidia-kernel-384.130
E: Unable to correct problems, you have held broken packages.
root@BlueLight:~#

[bw123]

I don't have the list of unresolved libraries, and I'm reluctant to roll back the headers to non-backport just yet. Here's the list of repositories I am using:

There's no "roll back" to it, you can have more than one ver linux-headers pkg installed. You're really having trouble, but you're not answering questions, like what kernel are you using/building on? What sources.list produced that list of unresolved "libraries" or whatever?

Why didn't you show stevepusser the cmd he showed 'apt-get install --reinstall nvidia-kernel-dkms' ?

There are a boatload of threads here on the forum about nvidia proprietary. It's a difficult hardware, so maybe searching the forum a little will help you out? People solve it occasionally, so look for nvidia+[solved] maybe?

[jlarysz]

I'm trying to answer questions as I figure out what the answers are.

The kernel, as given by uname -a is:

Code: Select all

Linux BlueLight 4.9.0-8-686-pae #1 SMP Debian 4.9.130-2 (2018-10-27) i686 GNU/Linux

The source list is already listed above. Here it is again:

Code: Select all

# deb cdrom:[Debian GNU/Linux 9.5.0 _Stretch_ - Official i386 NETINST 20180714-11:10]/ stretch main

#deb cdrom:[Debian GNU/Linux 9.5.0 _Stretch_ - Official i386 NETINST 20180714-11:10]/ stretch main

deb http://ftp.us.debian.org/debian/ stretch main contrib non-free
deb-src http://ftp.us.debian.org/debian/ stretch main contrib non-free

deb http://security.debian.org/debian-security stretch/updates main contrib non-free
deb-src http://security.debian.org/debian-security stretch/updates main contrib non-free

# stretch-updates, previously known as 'volatile'
deb http://ftp.us.debian.org/debian/ stretch-updates main contrib non-free
deb-src http://ftp.us.debian.org/debian/ stretch-updates main contrib non-free

# stretch-backports
# deb http://httpredir.debian.org/debian stretch-backports main contrib non-free

Here's the result of reinstalling kdms. Since I've flushed away the nvidia drivers so that I can rebuild them, and the rebuild isn't working using non-backport headers, I don't think that it is installed at all.

Code: Select all

root@BlueLight:~# apt-get install --reinstall nvidia-kernel-dkms
Reading package lists... Done
Building dependency tree
Reading state information... Done
Reinstallation of nvidia-kernel-dkms is not possible, it cannot be downloaded.
The following packages were automatically installed and are no longer required:
  libegl-nvidia0 libgles-nvidia2 libnvidia-cfg1 libnvidia-egl-wayland1 libnvidia-eglcore libopengl0 libvulkan1 linux-headers-4.18.0-0.bpo.1-686-pae
  linux-headers-4.18.0-0.bpo.1-common linux-kbuild-4.18 nvidia-driver-bin nvidia-driver-libs nvidia-egl-common nvidia-egl-icd nvidia-egl-wayland-common
  nvidia-egl-wayland-icd nvidia-persistenced nvidia-vulkan-common nvidia-vulkan-icd
Use 'apt autoremove' to remove them.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.

[bw123]

...
Reinstallation of nvidia-kernel-dkms is not possible, it cannot be downloaded.
...

This happens when you have a newer ver installed than is available. Since you've installed a ver from bpo, and now commented out backports in sources.list, you can't reinstall, and you can't get any needed dependencies. It looks like you have marked pkgs to be installed form backports, but before that could happen, you commented it out and apt updated the available pkgs.

There's a 'backing out in case of failure' on the nvidia wiki. It might help you get it all removed, then start over. I would try the stable ver first, spend a little time on it, and if it absolutely will not work, then I'd remove it, and try the backport version. When using a backported video driver, I'd probably use the newest backported kernel too. That's one possible thing to try now, since you have the headers, and the dkms, why not reenable backports repo, install the 4.18 backported kernel and see?

Hang in there, and when you get it working, let people know, thanks.

[jlarysz]

OK, I finally have Nvidia drivers working. Thank you everyone for your contributions.

Here's a summary as I see things.

1: I have a "standard" most-recent Debian distribution for Windows. I cannot use backport headers.

2: The most recent driver, "nvidia-driver", does not work yet - several required libraries haven't been copied to the standard distribution at this time. I have to use the older "nvidia-legacy-340xx-driver" package

3: The documentation in https://wiki.debian.org/NvidiaGraphicsDrivers is wrong in instructing the reader to use backports in several places when the "standard" distribution is involved.

4: The "xorg.conf" file generated by "nvidia-config" maybe be imperfect and even necessary, but it seems to works OK.

I rat holed a few times, but the experience was valuable and I learned a few things on the way.

[milomak]

looking at the nvidia site, this card should not be using the legacy drivers but the main driver

you have should enable stretch-backport, remove all nvidia packages and install from fresh

[sunrat]

1: I have a "standard" most-recent Debian distribution for Windows.

Is this a WSL install from Windows Store? It's not a "standard" Debian installation in that case. Very important information you should have mentioned in your first post. AFAIK it's designed to run a command line interface, not a graphical one so Nvidia drivers would not be needed or supported.

[jibberjabber]

I am confused, so is this a MS windows based on Debian system, or a Debian system based on MS windows ?