Kernel 5.8.0-34-generic CRASHED

[farlander762]

Ubuntu 20.04. Due to a spotty Internet connection I've been fighting for two months (thanks Comcast), I got a bad DL of the latest Kernel via standard Ubuntu updater. When attempting to boot it goes through the disk check with the spinning icon and then to a black screen with blinking cursor. No command line and the screen flickers about every 10 seconds like it's trying to restart the graphics. Won't start in 5.8.0-34's "safe mode" or "debug mode" or whatever the Grub menu option is either. It did post an error during the update about the Nvidia graphics driver and I did let the bug reporter do it's thing, but I'm just not advanced enough to know what happened.

I can boot into the previous 5.4.0-59-generic just fine. I think what I need to do is remove the newer kernel and then let it update again but am open to most anything that fixes it. Comcast ran a new temporary cable yesterday so the connection is now stable. I do have the option to fully R&R Ubuntu since my personal stuff is on a separate drive and am competent to do that. Can someone walk me through removing the affected kernel first?

Please help!!

[CelticWarrior]

I would try, before anything else

sudo apt update && sudo apt full-upgrade

[CatKiller]

I got a bad DL of the latest Kernel via standard Ubuntu updater.

Unlikely. The packages are checksummed and cryptographically signed. If the package was corrupted by the Internet connection the package manager would either download it again or stop and complain.

It did post an error during the update about the Nvidia graphics driver

How did you install the driver?

When it's installed through the package manager it will automatically reconfigure itself on kernel upgrades so that it keeps working. Other methods of installing the driver lack that facility.

[farlander762]

How did you install the driver?

I didn't do anything with the Nvidia driver. Whatever was done was through the updater. If you're asking about on my original install, that was last year sometime and I don't remember. If not a bad DL, I don't know what to say. It was a standard kernel update and I wasn't paying it too much attention. The system was otherwise idle during the updates because kernels are big important stuff and I knew I was about to restart anyway.

Going to give the "apt" advice above a quick whirl. UPDATE: No change after running

sudo apt update && sudo apt full-upgrade

Thanks!

[CatKiller]

If you're asking about on my original install, that was last year sometime and I don't remember.

Yes, that's what I mean.

Installed the Linux way - through the package manager, optionally after adding a PPA with a newer version - the shim that goes between the open source kernel and the proprietary driver module gets automatically rebuilt on kernel upgrades through a process called DKMS. That stops the driver breaking on kernel upgrades, since otherwise they would only work with one specific kernel version.

Nvidia have a .run file on their website, which is intended for package maintainers rather than end users, but ex-Windows users insist on trying to use it because they feel it's sufficiently close to what they're used to doing on Windows. In many cases, that file will not enable DKMS, which means the driver breaks on kernel upgrades.

A broken graphics driver followed by frustrated forced power downs would give exactly the symptoms you've described.

If that is what you've done - erroneously installed a driver from a random website - you can boot into the kernel that still has a working graphics driver, uninstall it using the script that you used to install it, and then install the driver through the package manager or the built in automatic tool. That will create working shims for your installed kernels and create a new one for each new kernel that you install in the future.

[farlander762]

Gotcha, I would not have gone off reservation to get a driver. Most likely it was the one recommended by Ubuntu (I actually select Nvidia due to their Linux support).

BTW, I went back in my browser history and found the bug report. https://bugs.launchpad.net/ubuntu/+s...0/+bug/1885344 Looks like this is a real bug and this same issue crops up on fresh installs with this driver/kernel combo.

Thanks for your help, I'll see what they do going forward and keep booting to the older kernel for now.

[grahammechanical]

My 20.04 got a kernel update a couple of hours ago. It was an update to kernel 5.4. An earlier update was to the 18.04 Hardware Enablement Stack. If there is an issue with the 5.8 kernel and the 20.04 Hardware Enablement Stack then under the phased updates system the 5.8 kernel will not be made available to more users until the issue is fixed.

I suggest that the original poster continue loading the 5.4 kernel. At some point there will an update to the 5.8 kernel and when that comes through you can try loading with it again.

Regards

[farlander762]

5.8.0-36-generic was pushed down this morning. Same issue.

Just really happy I still have the older kernel available.

[Erlander]

I was experiencing the same problem but have fixed it by changing from the NVidea driver to the X_org driver.

[farlander762]

I can't boot into those two kernels at all to be able to do so. Not even to CLI.