Error everytime I install a kernel

[Cavsfan]

I've tried this many, many times and always end up with an unbootable system.
If I try to install an antix kernel or any kernel with the Mx Package Installer I always get this: Error! Bad return status for module build on kernel: xxx

I mean everytime since the first time I installed Mx_Linux. The first time I was able to install a 4.22 antix kernel and booted with it regularly.

I installed Mx_Linux 18.3 3-4 times today and each time it would get the same error and I ended up with a flashing cursor in the top left.

So, this last time I used a terminal to install the kernel. It is just the header and the image; I would think it would pull some other dependencies in but, it does not.
Here is what the output looked like from terminal:

Code: Select all

cavsfan@mxlinux18:~
$ sudo apt install linux-headers-5.2.15-antix.1-amd64-smp linux-image-5.2.15-antix.1-amd64-smp
[sudo] password for cavsfan: 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following NEW packages will be installed:
  linux-headers-5.2.15-antix.1-amd64-smp linux-image-5.2.15-antix.1-amd64-smp
0 upgraded, 2 newly installed, 0 to remove and 0 not upgraded.
Need to get 60.3 MB of archives.
After this operation, 355 MB of additional disk space will be used.
Get:1 http://www.gtlib.gatech.edu/pub/MX-Linux/workspace/antix/stretch stretch/main amd64 linux-headers-5.2.15-antix.1-amd64-smp amd64 5.2.15-antix.1-amd64-smp-1 [11.4 MB]
Get:2 http://www.gtlib.gatech.edu/pub/MX-Linux/workspace/antix/stretch stretch/main amd64 linux-image-5.2.15-antix.1-amd64-smp amd64 5.2.15-antix.1-amd64-smp-1 [48.9 MB]
Fetched 60.3 MB in 17s (3,444 kB/s)                                                                                                                                                                                                                                                           
Selecting previously unselected package linux-headers-5.2.15-antix.1-amd64-smp.
(Reading database ... 257882 files and directories currently installed.)
Preparing to unpack .../linux-headers-5.2.15-antix.1-amd64-smp_5.2.15-antix.1-amd64-smp-1_amd64.deb ...
Unpacking linux-headers-5.2.15-antix.1-amd64-smp (5.2.15-antix.1-amd64-smp-1) ...
Selecting previously unselected package linux-image-5.2.15-antix.1-amd64-smp.
Preparing to unpack .../linux-image-5.2.15-antix.1-amd64-smp_5.2.15-antix.1-amd64-smp-1_amd64.deb ...
Unpacking linux-image-5.2.15-antix.1-amd64-smp (5.2.15-antix.1-amd64-smp-1) ...
Setting up linux-image-5.2.15-antix.1-amd64-smp (5.2.15-antix.1-amd64-smp-1) ...
Error! Bad return status for module build on kernel: 5.2.15-antix.1-amd64-smp (x86_64)
Consult /var/lib/dkms/broadcom-sta/6.30.223.271/build/make.log for more information.
Error! Bad return status for module build on kernel: 5.2.15-antix.1-amd64-smp (x86_64)
Consult /var/lib/dkms/nvidia-current/390.116/build/make.log for more information.
Error! Bad return status for module build on kernel: 5.2.15-antix.1-amd64-smp (x86_64)
Consult /var/lib/dkms/virtualbox-guest/5.2.24/build/make.log for more information.
update-initramfs: Generating /boot/initrd.img-5.2.15-antix.1-amd64-smp
Generating grub configuration file ...
Found linux image: /boot/vmlinuz-5.2.15-antix.1-amd64-smp
Found initrd image: /boot/initrd.img-5.2.15-antix.1-amd64-smp
Found linux image: /boot/vmlinuz-4.19.0-5-amd64
Found initrd image: /boot/initrd.img-4.19.0-5-amd64
....

I should also mention that I installed nvidia-driver 390.116 The 1st time I installed the kernel before the Nvidia driver and was able to boot into it but, I only had maybe 1280 x 1024 resolution which looked pretty bad on this 4k 3840x2160 monitor.
I could not find any way to change it.

If I install it via MX Package Installer it does the DKMS thing to both kernels and is unbootable.

But, since I installed it via CLI, it lets me get in through the 4.19.0-5-amd64 kernel.

So, is there some mystery to installing these kernels? It should be a piece of cake. I tried nomodeset on the kernel line and it was still unbootable.

[timkb4cq]

The 390.116 nvidia driver won't build on the 5.2 kernel. That's why it fails.
The 390.129 nvidia driver in the MX Test repo has been patched for the newer kernels.

[Stevo]

You also have build failures for the stock 5.2.24 version of virtualbox and the stock version of broadcom-sta-dkms.

The 5.2.32 version of VBox in the experimental repo will build on the 5.2 kernel, and the 6.0.12 version in the test repo has been tested on up to and including the 5.3 kernel, as has the broadcom-sta-dkms package there.
This is nothing that we're not already aware of on MX, but we don't add the antiX kernels to our repo. We note this gotcha in the announcements when we jump to a new major version of the Liquorix kernels, which we do package.

[Cavsfan]

The 390.116 nvidia driver won't build on the 5.2 kernel. That's why it fails.
The 390.129 nvidia driver in the MX Test repo has been patched for the newer kernels.

Thank you! I'll do that!
Much appreciation!!!

[Cavsfan]

You also have build failures for the stock 5.2.24 version of virtualbox and the stock version of broadcom-sta-dkms.

The 5.2.32 version of VBox in the experimental repo will build on the 5.2 kernel, and the 6.0.12 version in the test repo has been tested on up to and including the 5.3 kernel, as has the broadcom-sta-dkms package there.
This is nothing that we're not already aware of on MX, but we don't add the antiX kernels to our repo. We note this gotcha in the announcements when we jump to a new major version of the Liquorix kernels, which we do package.

I noticed it mentioned virtualbox but, I do not want to have virtualbox installed. I have no use for it and didn't know it automatically installs.

How do I get rid of that? I know I need some dkms module to I believe load the video driver in the kernel as it's built. If I get rid of virtualbox, is there another dkms package I could/should use?
Edit: I think I found the DKMS package that goes along with the 390.129 nvidia driver in the MX Test repo.

Trying them now. But, need to reboot first.

[Cavsfan]

I installed these via the Test repo on the MX Package installer:
(which was 390.129-1~mx17+1 - Nvidia meta package - 390xx legacy version)

Code: Select all

The following additional packages will be installed:
  libegl-nvidia-legacy-390xx0 libgl1-nvidia-legacy-390xx-glvnd-glx
  libglx-nvidia-legacy-390xx0 libnvidia-legacy-390xx-eglcore
  libnvidia-legacy-390xx-glcore libnvidia-legacy-390xx-ml1
  nvidia-legacy-390xx-alternative nvidia-legacy-390xx-driver-bin
  nvidia-legacy-390xx-driver-libs nvidia-legacy-390xx-egl-icd
  nvidia-legacy-390xx-kernel-support nvidia-legacy-390xx-vdpau-driver
  xserver-xorg-video-nvidia-legacy-390xx
Recommended packages:
  nvidia-settings-legacy-390xx libnvidia-legacy-390xx-cfg1 nvidia-persistenced
  nvidia-legacy-390xx-driver-libs-i386 libopengl0 | libopengl0-glvnd-nvidia
  libgles-nvidia-legacy-390xx1 libgles-nvidia-legacy-390xx2
  nvidia-legacy-390xx-vulkan-icd
The following NEW packages will be installed:
  libegl-nvidia-legacy-390xx0 libgl1-nvidia-legacy-390xx-glvnd-glx
  libglx-nvidia-legacy-390xx0 libnvidia-legacy-390xx-eglcore
  libnvidia-legacy-390xx-glcore libnvidia-legacy-390xx-ml1
  nvidia-legacy-390xx-alternative nvidia-legacy-390xx-driver
  nvidia-legacy-390xx-driver-bin nvidia-legacy-390xx-driver-libs
  nvidia-legacy-390xx-egl-icd nvidia-legacy-390xx-kernel-dkms
  nvidia-legacy-390xx-kernel-support nvidia-legacy-390xx-vdpau-driver
  xserver-xorg-video-nvidia-legacy-390xx
0 upgraded, 15 newly installed, 0 to remove and 239 not upgraded.
Need to get 27.7 MB of archives.
After this operation, 115 MB of additional disk space will be used.
Do you want to continue? [Y/n]

The 5.2 kernel still won't boot but, the 4.19 kernel does.
So, how do I go about removing vbox or whatever to get the 5.2 kernel to boot?

Code: Select all

$ inxi -G
Graphics:  Device-1: NVIDIA GM200 [GeForce GTX 980 Ti] driver: nvidia v: 390.116 
           Display: x11 server: X.Org 1.19.2 driver: nvidia resolution: 3840x2160~60Hz 
           OpenGL: renderer: GeForce GTX 980 Ti/PCIe/SSE2 v: 4.6.0 NVIDIA 390.116

Everything says I still have the 390.116 driver running. I'm getting lost.

[Stevo]

The virtualbox packages won't prevent a boot on a bare-metal install, but you can open Synaptic, hit the "installed" on the Status tab, search for virtualbox there, and remove any that are installed. I don't know if you need broadcom-sta-dkms, since you haven't posted your wi-fi device, but you can remove it also if you don't have a Broadcom one.

You can manually try a rebuild of the 390xx legacy driver to see if it does error out--first make sure the 390.129 version is installed:

Code: Select all

apt policy nvidia-legacy-390xx-kernel-dkms

if so, then

Code: Select all

sudo dpkg-reconfigure nvidia-legacy-390xx-kernel-dkms

[Cavsfan]

Sorry but, I re-installed again and I'm just going to stick with the default kernel 4.19.0-5-amd64.

I've tried the Antix 5.1.2 kernel Meltdown and Spectre patched in the MX Package Installer and that is where I got the error initially.
Isn't there a normal progression of kernels on MX Linux 5.1.2, 5.1.3. etc? I've been playing around with MX about a year now and I have not seen a kernel update come down in regular updates.

I turned off auto update notifications and update via CLI, just like every other installation I have.

I have a custom grub setup and 9 Linux systems installed. MX Linux and Fedora are the only 2 I've come across that does not create symlinks to the kernels.

I believe most systems once installed, there are no symlinks to the kernel until the 2nd kernel gets installed. MX doesn't automatically upgrade kernels and Fedora just does not create symlinks.
This got me what I needed on MX Linux:

Code: Select all

cd /boot
sudo ln -s -f initrd.img-4.19.0-5-amd64 initrd
sudo ln -s -f vmlinuz-4.19.0-5-amd64 vmlinuz

I created a script to run on Fedora when a new kernel is created that gives me those symlinks.

I get the biggest kick out of installing/fine tuning systems. Then once they are like I want them I look for other stuff to do.

Here is my grub screen on MX Linux 18:


It would be nice if one didn't have to jump through so many hoops just to install a newer kernel. But, as long as the default kernel is safe and secure, then I'm good with it.

[Stevo]

We've already diagnosed the issue and given the answer--you must upgrade your Nvidia driver to the legacy 390xx version from the test repo, 390.129, before you try and install a 5.2 kernel. It's no mystery why you can't boot if you don't do that.

[Cavsfan]

We've already diagnosed the issue and given the answer--you must upgrade your Nvidia driver to the legacy 390xx version from the test repo, 390.129, before you try and install a 5.2 kernel. It's no mystery why you can't boot if you don't do that.

I can do that but, I don't think it should be that hard. It should just work. Not saying I know everything but, I know some things and I don't find this as easy as it should be.

I'll give that a shot and report back.

I just purged 3 virtualbox packages that were installed and am rebooting and if all goes well then installing the video driver you mention and the 5.2 kernel after that.