Boot problem after (Kernel) upgrade, more info

[Keith]

On 9/19/21 7:02 AM, Grizzly via ubuntu-users wrote:
>
> As mentioned I all was well with 20.04.3 (5.8.0-59) the I ran dist-upgrade, and
> did not reboot (apparently) when I did boot again it would not work
>
> AFAICT it was an nvidia driver (or my first guess) various attempts later (all
> yesterday on slow internet) got me to a point where even 5.8.0-59 would not
> boot, in an effort to get back to (at least) 5.8.0-59 I went thru
>
> Advanced-recovery-drop to root shell then to try to fix (fix dkpg had not
> helped)
>
> ~$ apt install -f
> Reading package lists... Done
> Building dependency tree
> Reading state information... Done
> 0 to upgrade, 0 to newly install, 0 to remove and 0 not to upgrade.
> 1 not fully installed or removed.
> After this operation, 0 B of additional disk space will be used.
> Setting up nvidia-340 (340.108-0ubuntu5.20.04.2) ...
> dpkg: error: version '-' has bad syntax: revision number is empty
> dpkg: error: version '-' has bad syntax: revision number is empty
> dpkg: error: version '-' has bad syntax: revision number is empty
> update-initramfs: deferring update (trigger activated)
> INFO:Enable nvidia-340
> DEBUG:Parsing /usr/share/ubuntu-drivers-common/quirks/lenovo_thinkpad
> DEBUG:Parsing /usr/share/ubuntu-drivers-common/quirks/put_your_quirks_here
> DEBUG:Parsing /usr/share/ubuntu-drivers-common/quirks/dell_latitude
> Removing old nvidia-340-340.108 DKMS files...
>
> -------- Uninstall Beginning --------
> Module:  nvidia-340
> Version: 340.108
> Kernel:  5.8.0-59-generic (x86_64)
> -------------------------------------
>
> Status: Before uninstall, this module version was ACTIVE on this kernel.
>
> nvidia.ko:
>   - Uninstallation
>     - Deleting from: /lib/modules/5.8.0-59-generic/updates/dkms/
>   - Original module
>     - No original module was found for this module on this kernel.
>     - Use the dkms install command to reinstall any previous module version.
>
>
> nvidia_uvm.ko:
>   - Uninstallation
>     - Deleting from: /lib/modules/5.8.0-59-generic/updates/dkms/
>   - Original module
>     - No original module was found for this module on this kernel.
>     - Use the dkms install command to reinstall any previous module version.
>
> depmod........
>
> DKMS: uninstall completed.
>
> ------------------------------
> Deleting module version: 340.108
> completely from the DKMS tree.
> ------------------------------
> Done.
> Loading new nvidia-340-340.108 DKMS files...
> Building for 5.8.0-59-generic 5.11.0-34-generic
> Building for architecture x86_64
> Building initial module for 5.8.0-59-generic
> Done.
>
> nvidia.ko:
> Running module version sanity check.
>   - Original module
>     - No original module exists within this kernel
>   - Installation
>     - Installing to /lib/modules/5.8.0-59-generic/updates/dkms/
>
> nvidia_uvm.ko:
> Running module version sanity check.
>   - Original module
>     - No original module exists within this kernel
>   - Installation
>     - Installing to /lib/modules/5.8.0-59-generic/updates/dkms/
>
> depmod....
>
> DKMS: install completed.
> Building initial module for 5.11.0-34-generic
> ERROR: Cannot create report: [Errno 17] File exists:
> '/var/crash/nvidia-340.0.crash'
> Error! Bad return status for module build on kernel: 5.11.0-34-generic (x86_64)
> Consult /var/lib/dkms/nvidia-340/340.108/build/make.log for more information.
> dpkg: error processing package nvidia-340 (--configure):
>   installed nvidia-340 package post-installation script subprocess returned
> error exit status 10
> Processing triggers for libc-bin (2.31-0ubuntu9.2) ...
> Processing triggers for initramfs-tools (0.136ubuntu6.6) ...
> update-initramfs: Generating /boot/initrd.img-5.11.0-34-generic
> Errors were encountered while processing:
>   nvidia-340
> E: Sub-process /usr/bin/dpkg returned an error code (1)
>
> after that I could boot to 5.8.0-59 again, but not 5.11.0-34, trying the same
> process via 5.11.0-34 recovery I got a similar (almost identical response but
> for a failur to write to "nvidia crash report") but after that I could not boot
> to 5.11.0-34 or 5.8.0-59
>
> I repeated the first try and have 5.8.0-59 back
>
> To my mind it seems that 5.11.0-34 (and 5.11.0-27 before it) broke nvidia
> driver or are themselves broken as far as nvida driver is concerned
>
> any ideas, for now I'll stay on old(er) kernel
>

It sounds like your problem is this:

nvidia-340 kernel module failed to build (ERROR: Kernel configuration is
invalid)

https://bugs.launchpad.net/ubuntu/+source/nvidia-graphics-drivers-340/+bug/1938978

You have a few options:

1. Upgrade your card to a model Nvidia officially supports. It dropped
support for cards using the 340 driver at the end of 2019, and there
have been no updates for the 340 driver issued from Nvidia since then.
Ubuntu has already dropped its support in 21.04 and 21.10

2. Continue to use 5.8 kernel while waiting for Ubuntu devs to fix the
driver to build on 5.11 or newer kernels in 20.04. Given the HWE kernel
version is now 5.11, I'm not sure if the 5.8 generic version will get
anymore updates. In which case, your system could be vulnerable to any
new security exploits that have come out since the 5.8.x version your
running now was released.

3. Install a 3rd party 340 driver package from an external source. This
option has all the usual caveats that come from installing software from
untrusted and unsupported sources. In the bug report I linked above,
someone in the comments mentioned installing a package from a PPA to get
the driver working.

4. Uninstall proprietary Nvidia driver and use the open source nouveau
driver. Issues are possible performance degradation, loss of some driver
configurability, and monitor support options. Recommend to completely
purge all packages related to the proprietary Nvidia driver and
nvidia-settings program before rebooting to use nouveau driver in order
to avoid any driver conflicts.

5. Switch from the current hardware enablement stack kernel (HWE)
5.11.x, to the general availability kernel (GA) 5.4.x. If your hardware
and software needs do not require a newer version of the kernel to work
properly, switching to the 5.4.x kernel will solve the problem of having
to fix the 340 graphic driver every time a new HWE kernel version is
installed with a point release of the distribution. Plus, you'll
continue to get regular and security updates to the 5.4.x kernel till
20.04's EOL in April 2025.

HTH,

--
Keith