Black screen on boot with 5.8.0-59.66 - reason why + delete kernel

This topics Goal is:

to determine if there is an issue with the 5.8.0-59 kernel I should report on launchpad as Im do not know to that moment
to ask for guidance to alter the standard kernel to start from,
how to delete other kernels.
EDIT: To install the dependencies for Kernel 5.12.14 for further Tryouts

I run a HP ProBook 450 G6 with LTS 20.04.2, but since Software Updater installed the kernel 5.8.0-59.66 to that Computer, it does go into a fully black screen when LUKS is supposed to prompt for the password. I have waited half an hour to check if something would show up, but nothing changed.

Entering the password blindly and hitting return didn't show any effect either. - WRONG, see below

I can tell that the System and LUKS is not manually altered too much except of some swap file there is no software initialized befor the user login, like I am running regular Gnome, the hardware only got 1 iGPU without dedicated GPU, running on X11 Drivers, so definitely no dedicated GPU drivers like with nVidia cards. I also unplugged all USB devices from the system for troubleshooting, without effect.

With using the advanced start options and selecting an older kernel, like 5.8.0-55.62, everything works as expected even when some settings like Audio Device and Microphone Device have been altered.

Issues with kernel 5.8.0-59?

I first stop here before I consider to report an error, so the error report can be more rich of information as I am not sure how to handle such.

I am not too sure what logs I could setup or readout from a startup attempt, so may you advise me to determine the issue is on my side with software conflicts or the dev. side?

Removing kernels

Filtering of dpkg command should work for determining which kernel we want to delete. So i found this example https://phoenixnap.com/kb/ubuntu-remove-old-kernels

sudo dpkg ––list | egrep –i ––color ‘linux-image|linux-headers’

sudo dpkg ––list | egrep –i ––color ‘linux-image|linux-headers’ | wc –l

but my system seems to not have wc, nor does the --color parameter work for me as the CLI prompts.

I would still appreciate what this complicated method would have as benefit, but to simply get the job done now use a simpler command

dpkg --list | grep linux-image

This prompts smth. I could work with.

Now I got to delete some selection of that. The manual in the link from above, suggest as Method to Manually Remove Individual Kernels from Command-Line the following command that lists again the Kernels, but with their installation status:

dpkg -l | tail -n +6 | grep -E 'linux-image-[0-9]+' | grep -Fv $(uname -r)

It prompted a list to me, missing the kernel that is currently in use (5.8.0-55.62...). I guess that sounds not too wrong as the kernel is active and running (?).

Now the fun part: The kernel 5.8.0-59 I want to mark to delete, is already marked for "installed and eligible for remove" with the "ii" status indicator.

So let´s try the autoremove command first:

sudo apt-get ––purge autoremove

Checking with the previous dpkg list again, if the kernel is now status "rc" or still "ii" proofs that Kernel 5.8.0-59 still is existing and marked "ii". Mhhhh... That didn't work as it should.

So we try to do it manually:

sudo dpkg ––purge linux-image-5.8.0-59-generic

sudo dpkg ––purge remove linux-image-5.8.0-59-generic

but this prompts

dpkg: error: need an action option

My best guess would be that this error code was prompted, because there is no other kernel left in the system except the 5.8.0-55.62, but I don't know for sure.

I am stuck at that point: I don't know how to manually remove the a desired kernel.

I also am not sure how the list of kernels is made - is it just generated by the the images that are available, or is there some list to alter?

--- Chronological Updates:

Kernel 5.8.0-63.71

Update: The machine was updated to kernel 5.8.0-63.71 and the blank screen persisted.

Another user, confirming the issue

Another user reported the same issue on his two HP ProBook 450 G7 machines. Sadly, he posted it as answer to my question and got downvoted/deleted, so now his reply is gone! Thanks for this great Housekeeping @Mods, would have been great to help the user and copy/paste his answer to the comment section.

As mentioned I use some HP ProBook 450 G6, so its the previous Generation. The hardware is different, but both are Intel CPUs and iGPUs. HPs Software/Settings also may interfere as many HP machines are reported with similar issues.

Firmware (UEFI)

I have checked for the newest HP ProBook 450 G6 and Intel Firmwares and upgraded from 1.08.0 to the newest Firmware R71 1.16.0 to this date. This seems to not have any effect.

Tipp for Linux users in need to upgrade their Firmware: HP only provides Windows installers of the files. If you have no native Windows or VM or Wine you can not access these files. As soon as your VM/Wine install the files you can browse the install directory (Standard Path is in C:\SWSetup\XXXXXX), create a FAT32 USB Stick and just manualy create the Folderstructure "EFI\HP\BIOS\New" and place the .bin file in there, as this is everything the installer does. If you got not native Windows, routing the USB device to Wine/VM may be tricky so by this hint I hopefully saved you some time.

TPM

IRC #ubuntu User TJ- recommended to start Grub - Command Line and "rmmod tpm" to test if the issue is regarding due to TPM. Seems like this command didnt worked as TPM was still active: https://pastebin.com/c4UeXXsa

Deactivating TPM globaly in the UEFI Firmware was tried aswell: The log should be the right one, but it states TPM was activated. I will perform another log the next restart again: https://pastebin.com/fjK8TVVL

But I will need to redo that and check the log if TPM was successfully deactivated.

update-grub

update-grub was performed, without effect.

Booting Recovery Mode

Ive tried booting the recovery modes of the 5.8.0-59 and the -63 Kernel: Booting into recovery, an than continue to booting "fully", does work. The iGPU/screen is not operated by the regular driver by this (which may be the reason the screen did not turned off this time), as for example Night Light does not work with that and a quick look into the Gnome Settings About Page revealed some different GPU/driver

Booting with blind password entry - works

I tried booting the -59 Kernel again and entered my LUKS and User password blind.

Because I had no second monitor I tried to turn on some music app and it worked. So the computer was initializing the kernel successfully and booted successfully to most part, the kernel just has some issues regarding the screen/gpu as far as we know! (Other Users with similar hardware and bug reports, describe a working external screen port)

Here is the some journalctl -b log of that:

https://pastebin.com/uTAUG6XH

and for comparing to a "successful" boot with Screen/GPU by Kernel 5.8.0-55:

https://pastebin.com/6qzCSKPR

I havent compared the logs professionaly, but IRC #ubuntu User tomreyn noted this line:

gnome-shell[1530]: Failed to set CRTC gamma: drmModeCrtcSetGamma on CRTC 51 failed: Permission denied

Further professional investigation beyond my user input is needed.

Kernel 5.12.14

It was recommended in here https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1934647 to try the 5.12.14 kernel. I downloaded the files and dpkg installed them, but some dependencie was missing: I think the name of the respetory was lib6c and it needed an upgrade from 2.1 to 2.3 or something like that.

But upgrading with apt-get install -f, apt-get update && upgrade, apt --fix-broken install wasnt able to perform the upgrade. I performed an autoremove which was removing one of the 4 kernel files with the missing dependency upgrade. update-grub after that, yet, the other 3 kernels Booting up again with the new kernel obviously wasnt possible and prompted smth like the Kernel, Ramdisk or initd was missing which is understandable as he needs the autoremoved kernel file and the dependencies required of it. I now run into the same situatio as stated above: Removing the kernels manually prompts the same "action required" as stated above. I either need to find the reason for that error prompt, or I got to install the missing file and upgrade the dependency.

Diagnostics

I want to provide further diagnostic logs but am not sure of what: So please feel free to advise me what logs you like to see in the comments and maybe quote me the command. Thanks

I have an HP ProBook 450 G7 with Ubuntu 20.04 and I'm in the same situation.

Please, keep updating this entry if you find something.

Now I'm working with 5.8.0-55.

I have found that booting with the "nomodeset" parameter makes the laptop monitor work again but external monitors stop working.