Repair after faulty kernel update

[PolderPenguin]

This kernel update definitely went wrong with two of my desktops. Booting with this kernel locks out my screen ("no image" and screen goes black) and also my keyboard. What in heaven went wrong?

[hnimmo]

This kernel update definitely went wrong with two of my desktops. Booting with this kernel locks out my screen ("no image" and screen goes black) and also my keyboard. What in heaven went wrong?

There is apparently a bug (see a previous post). Fortunately, this wonderful OpenSUSE system allows you to boot with the previous kernel!!

[Bill_L]

There is apparently a bug (see a previous post). Fortunately, this wonderful OpenSUSE system allows you to boot with the previous kernel!!

uh, no it doesn’t! I put a fresh install on my windows drive(MISTAKE!), and it only has the new kernel, NO older version to go back to. This is the last time I answer YES to update the repositories during an install!

I will have to check that desktop’s graphic, don’t recall, I shut that machine down so, nothing else messes up.

i have also disabled updates on our laptops. My wife will dump on my head if something messes up her machine! And hope for a quick fix!

I dug out my Leap 15.2, and will put it on the dedicated Leap drives, “EOL” be darned!

[EdD]

Boot with the working kernel, delete only the faulty one and lock the kernel.

I have the same black screen after updating to version 15.3...59.43 of Leap yesterday. My card also has a Radeon, the HD4530. I boot verbose and the first 3 or 4 lines I can see (too fast to count) and then the screen goes black. If I judge the then following activity according to the HD led, the booting itself goes on as normal, but that is of course no more than an unconfirmed guess. Starting up in grub with the version ...59.40 runs, as you can see now.

I am not familiar with deleting a kernel version and locking the present one; can you please tell how? How do we know that we can trust the next update and unlock again? I have a complete system backup with all the settings, but that is always a hassle to do that.

[purevw]

I can perhaps provide a little insight. It only affected two of my old MBR, non-UEFI machines. My UEFI machines were unaffected. When booting, the grub menu displays the newest (correct) kernel version, but when it actually begins boot, it tries to use the previous kernel. You can catch a glimpse of the message "booting kernel-default-(old-version) just before the failure. If the previous kernel is already removed, of course the machine can't boot.
I have always removed the previous kernel while updating a new one. I guess that policy bit me this time, as there was no older kernel to find.
I am not sure if this is a grub2 (not grub-efi) problem or something else. Perhaps grub2 changed the menu, but failed to target the new kernel?
After any kernel upgrade, I verify the installation with "mkinitrd -v" to make sure the kernel is there as multiple times in the past, Yast left me with no kernel in the system (a different bug). Again, when there is a kernel update available, I always mark the current kernel for removal and leave the newest upgrade marked to install, and verify under the "Installation Summary" tab of Yast before selecting "OK". Several times, Yast removed the current kernel and failed to install the upgrade.

My solution was to use the original installation media to "upgrade" the system. The only thing I allow it to upgrade is the kernel.

Perhaps in the future, I will wait to remove the existing kernel until after the next boot.

[purevw]

As an addition to my last post, I should state that all my machine use the same Radeon graphics card model, but only the old MBR machines fail to boot.

[mrmazda]

I dug out my Leap 15.2, and will put it on the dedicated Leap drives, “EOL” be darned!

That's a silly way forward. Boot any live media you have available. Chroot to installed system. Install a previous kernel 5.3.18-59.40.1 or 5.3.18-59.37.1. Remove the broken kernel.

Or, extract and install just the radeon module from a prior kernel rpm as explained in the bug report's comment #5.

[Bill_L]

That's a silly way forward. Boot any live media you have available. Chroot to installed system. Install a previous kernel 5.3.18-59.40.1 or 5.3.18-59.37.1. Remove the broken kernel.

Or, extract and install just the radeon module from a prior kernel rpm as explained in the bug report's comment #5.

Don’t know what all that means. I don’t have ‘live’ media, just the iso for installing 15.3 Have no idea about chroot, or how to install a previous kernel.
You apparently forgot whom you are communicating with. A NEW USER WITHOUT ALL THE EXPERIENCE MOST OF YOU HAVE!

What I did do is install from the Leap 15.3 USB stick, BUT I didn’t allow it to add the latest update repositories. And I have disabled updates so that new kernel doesn’t sneak up on me. Kernel is 5.3.28-5-57-default.

it will be that way until the problem kernel is fixed. I’m tired if redoing my system every other day! So this I can live with.
I also set Windows as the default OS, and a -1 in the time to select box.

I’m sure there will be updates to the status of that kernel that doesn’t like my graphic card.
If I can find kernel(s) 5.3.18-59.40.1 or 5.3.18-59.37.1, I will consider installing them.

[Bill_L]

As an addition to my last post, I should state that all my machine use the same Radeon graphics card model, but only the old MBR machines fail to boot.

DITTO for me, Radeon graphics, you can see ‘MY’ workaround above. Not pretty. BUT I do have a working Leap 15.3.

[Sauerland]

@Bill_L
Do you install with netinstall?



There is also an workaround to copy the Radeon Module to the non working kernel:
https://bugzilla.opensuse.org/show_b...?id=1195142#c5