Cannot boot with thunderbolt dock connected before kernel loads [groovy] [focal with groovy hwe kernel]

[Luis Alberto Pabón]

Dell XPS 9650 with a TB16 thunderbolt dock. 2 external displays through the dock (HDMI and VGA). My system uses graphical boot via plymouth. Dell logo remains on screen while ubuntu loads into gdm3 (when it is booting normally). I'm using LUKS.

This is a regression from focal and kernel 5.4.

If I boot the laptop with the thunderbolt dock connected, the system will get past the Dell logo, go to a black screen and remain there. The screen is powered up as I can see some light on the edges. It won't boot up or respond to any keys.

After force-shutdown via power key long press, on the next boot the grub menu will appear. The system will boot normally then (with the dock still connected).

If I boot the system with the dock disconnected, it will correctly go into disk encryption password prompt. If at that point I connect the dock and wait without entering the password, the system will correctly set up the dock: the displays lit up and the mouse and keyboard plugged into it become active. The system boots normally.

I'm trying to bisect in which kernel version this happened. I'll put results in a comment.

ProblemType: Bug
DistroRelease: Ubuntu 20.10
Package: linux-image-generic 5.8.0.25.30
ProcVersionSignature: Ubuntu 5.8.0-25.26-generic 5.8.14
Uname: Linux 5.8.0-25-generic x86_64
ApportVersion: 2.20.11-0ubuntu50
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC1: luis 3332 F.... pulseaudio
 /dev/snd/controlC0: luis 3332 F.... pulseaudio
CasperMD5CheckResult: skip
Date: Fri Oct 23 16:27:30 2020
InstallationDate: Installed on 2019-08-17 (432 days ago)
InstallationMedia: Ubuntu 19.04 "Disco Dingo" - Release amd64 (20190416)
MachineType: Dell Inc. XPS 15 9560
ProcEnviron:
 LANGUAGE=en_GB:en
 TERM=xterm-256color
 PATH=(custom, no user)
 LANG=en_GB.UTF-8
 SHELL=/bin/bash
ProcFB: 0 i915drmfb
ProcKernelCmdLine: BOOT_IMAGE=/vmlinuz-5.8.0-25-generic root=/dev/mapper/vglinux-root ro quiet splash
PulseList: Error: command ['pacmd', 'list'] failed with exit code 1: No PulseAudio daemon running, or not running as session daemon.
RelatedPackageVersions:
 linux-restricted-modules-5.8.0-25-generic N/A
 linux-backports-modules-5.8.0-25-generic N/A
 linux-firmware 1.190
SourcePackage: linux
UpgradeStatus: Upgraded to groovy on 2020-10-09 (13 days ago)
dmi.bios.date: 11/17/2019
dmi.bios.release: 1.18
dmi.bios.vendor: Dell Inc.
dmi.bios.version: 1.18.0
dmi.board.name: 05FFDN
dmi.board.vendor: Dell Inc.
dmi.board.version: A00
dmi.chassis.type: 10
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.:bvr1.18.0:bd11/17/2019:br1.18:svnDellInc.:pnXPS159560:pvr:rvnDellInc.:rn05FFDN:rvrA00:cvnDellInc.:ct10:cvr:
dmi.product.family: XPS
dmi.product.name: XPS 15 9560
dmi.product.sku: 07BE
dmi.sys.vendor: Dell Inc.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This change was made by a bot.

[Alex Murray (alexmurray)]

FWIW I am seeing this also with my Lenovo X1 Carbon 6th when docked to my Thinkpad Ultra Docking Station (that has 2 external monitors connected to it). This booted fine on focal but after upgrading to Groovy (or doing a fresh install) the boot hangs as soon as the plymouth splash screen tries to display. Setting nospash in /etc/default/grub (GRUB_CMDLINE_LINUX_DEFAULT) works around this for now by disabling the splash screen.

[Johnny Bratholt (keruberne)]

I got the same issue on Thinkpad X1 Carbon 6th & ThinkPad Thunderbolt 3 Dock Gen2

I have seen this issue on other distros like Fedora and Manjaro aswell when using luks and it looks very much that is has something to do with plymouth and kernel version 5.6 and up.
I could replicate the issue on focal with kernel 5.8 and 5.9

[Luis Alberto Pabón (copong)]

After going through a lot of kernels from the mainline PPA I've concluded the same, issue happens with kernel 5.7+.

To complicate matters, if there's any output at all on the screen before the dell logo screen with the LUKS password entry is supposed to come back, it will work regardless of the kernel.

[Luis Alberto Pabón (copong)]

I had to do a clean-reinstall of 20.04 as there were other issues (my brother scan not working among others) that were deal breakers to me. I noticed the 5.8.0 kernel is available for install via the metapackage linux-generic-hwe-20.04-edge (installs linux-image-5.8.0-25-generic and related packages)

To cut a long story short, the error occurs in focal with the groovy kernel as well.

[Luis Alberto Pabón (copong)]

So it looks like focal will inherit this problem when HWE starts rolling out.

[Alex Murray (alexmurray)]

There is a fix for this which will be in the OEM kernel - https://bugs.launchpad.net/hwe-next/+bug/1902469 - and I am trying to see if this can be merged for groovy as well - https://lists.ubuntu.com/archives/kernel-team/2020-November/114483.html

[Luis Alberto Pabón (copong)]

Excellent stuff that is. Just to clarify, will this come in the default 5.4.0 kernel for Ubuntu 20.04.1 or the hwe kernel as the link to the fix suggests?

[Luis Alberto Pabón (copong)]

I realised what an idiotic question that was as I submitted the form. Please disregard 😬

[Johnny Bratholt (keruberne)]

Yesterday i updated my system, i noticed there was updates for plymouth and kernelupdate
After that i am back to the beginning again.
The workaroung Setting nospash in /etc/default/grub (GRUB_CMDLINE_LINUX_DEFAULT) does not work for me anymore.
I now have to open the screen on the notebook to type in my password for LUKS