[Medion Akoya E3223] Latest kernel breaks LCD screen

[Hadmut Danisch]

Hi,

I just installed Lubuntu 22.04 on a brandnew notebook, and everything was working well until rebooting after update:

Both the boot splash and the X11 screens are jumping and appear instable afer a few seconds, as out of sync and running through. It absolutely looks like a hardware problem like a broken LCD connector cable, but it isn't: booting fresh from the Lubuntu usb stick and evernything works. I have a second machine, exactly same model and installation procedure, which was working well for some weeks and had exactly the same problem after upgrading.

disabling the boot splash did not help.

But booting into the older vmlinuz-5.15.0-25-generic , that came with the Lubuntu installation iso file, instead of the latest vmlinuz-5.15.0-41-generic solved the problem. With the older kernel, everything is stable and fine.

So between the 25 and the 41 revision of the kernel there came something in which reliably breaks the screen configuration on both my two notebooks of identical hardware.

Both the X11 login screen and the desktop do work normally for about five seconds and then start running through like one of these old days analog screens when out of sync or a cable not plugged in correctly.

ProblemType: Bug
DistroRelease: Ubuntu 22.04
Package: linux-image-5.15.0-41-generic 5.15.0-41.44
ProcVersionSignature: Ubuntu 5.15.0-25.25-generic 5.15.30
Uname: Linux 5.15.0-25-generic x86_64
ApportVersion: 2.20.11-0ubuntu82.1
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: hadmut 1857 F.... pulseaudio
CRDA: N/A
CasperMD5CheckResult: unknown
CurrentDesktop: LXQt
Date: Thu Jul 21 12:28:31 2022
InstallationDate: Installed on 2022-07-20 (0 days ago)
InstallationMedia: Lubuntu 22.04 LTS "Jammy Jellyfish" - Release amd64 (20220419)
Lsusb:
 Bus 002 Device 002: ID 0bda:8153 Realtek Semiconductor Corp. RTL8153 Gigabit Ethernet Adapter
 Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
 Bus 001 Device 003: ID 8087:0aaa Intel Corp. Bluetooth 9460/9560 Jefferson Peak (JfP)
 Bus 001 Device 002: ID 0c45:6321 Microdia HP Integrated Webcam
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
MachineType: MEDION E3223
ProcFB: 0 i915drmfb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-5.15.0-25-generic root=UUID=ab20c954-2973-48a4-a73c-491b5bdeddc2 ro quiet cryptdevice=UUID=abc35d02-7e06-4152-b214-01f17a4740bb:luks-abc35d02-7e06-4152-b214-01f17a4740bb root=/dev/mapper/luks-abc35d02-7e06-4152-b214-01f17a4740bb nosplash
RelatedPackageVersions:
 linux-restricted-modules-5.15.0-25-generic N/A
 linux-backports-modules-5.15.0-25-generic N/A
 linux-firmware 20220329.git681281e4-0ubuntu3.2
SourcePackage: linux
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 09/13/2021
dmi.bios.release: 5.13
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: GeminiLake_YS13GR_V3.0.2
dmi.board.asset.tag: Default string
dmi.board.name: YS13GR
dmi.board.vendor: MEDION
dmi.board.version: Default string
dmi.chassis.asset.tag: Default string
dmi.chassis.type: 31
dmi.chassis.vendor: MEDION
dmi.chassis.version: GD7387P
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrGeminiLake_YS13GR_V3.0.2:bd09/13/2021:br5.13:svnMEDION:pnE3223:pvrELAN:rvnMEDION:rnYS13GR:rvrDefaultstring:cvnMEDION:ct31:cvrGD7387P:skuML-24100630032549:
dmi.product.family: Akoya
dmi.product.name: E3223
dmi.product.sku: ML-241006 30032549
dmi.product.version: ELAN
dmi.sys.vendor: MEDION

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

This change was made by a bot.

[Kai-Heng Feng (kaihengfeng)]

There are many versions between 5.15.0-25-generic and 5.15.0-41-generic, like 5.15.0-33-generic. So can you please give those versions a try so we can narrow down the offending change?

[Hadmut Danisch (hadmut)]

Is there instructions how to install older than recent kernel images?

[Hadmut Danisch (hadmut)]

I'm still looking for a method to test different build versions of kernels.

I tried that mainline tool described here
https://davejansen.com/install-a-linux-kernel-of-choice-under-ubuntu/

but that offers only different Linux versions (i.e. those who differ in the first three numbers like 5.15.0 and 5.15.1, but not in the ubuntu build number such as 5.15.0-25 and 5.15.0-41.

I am willing to do debugging for you, but you need to hint me about how to get these versions.

[Hadmut Danisch (hadmut)]

OK, got it. I wasn't aware that the older versions are still in the regular linux repository with different version numbers.

[Hadmut Danisch (hadmut)]

5.15.0-25 good
5.15.0-41 bad

and all available versions between hanging at boot time, not enabling the network.

Last message on console for these versions is

[ OK ] Finished Record Runlevel Change in UTMP.

no visible error message, but I can login on console 1. I need to figure out why these versions are hanging.

[Hadmut Danisch (hadmut)]

Boot with versions betwenn 25 and 41 is not hanging, it is the sddm (and thus X11) that is not running for yet unknown reasons, and thus the NetworkManager is not getting active.

What did you change betwenn 25 and 33 and between 40 and 41?

[Hadmut Danisch (hadmut)]

OK, I got further.

The distortion depends on the presence of the linux-modules-extra-5.15.0-*-generic package. I first did omit these extra modules and found a completely different behaviour without.

With linux-modules-extra-5.15.0-*-generic:

Kernels up to including 5.15.0-39-generic are fine. The bug came in with 5.15.0-40-generic

Without linux-modules-extra-5.15.0-*-generic:

System can't active ethernet and sddm hangs for unknown reasons when booting normally. But when booting into maintenance mode in grub, and just activating the "networking" (although it can't activate ethernet without the module), the system then comes up into a stable and normal X11 session, just the mouse pad not working (missing module) even with the 5.15.0-40-generic kernel!

So it must be
- something that came in with the -40 version
- probably something that is in the linux-modules-extra package and not the linux-image or the linux-modules package.

[Kai-Heng Feng (kaihengfeng)]

Here are the versions that can be tested:
linux-modules-extra-5.15.0-25-generic
linux-modules-extra-5.15.0-27-generic
linux-modules-extra-5.15.0-30-generic
linux-modules-extra-5.15.0-33-generic
linux-modules-extra-5.15.0-35-generic
linux-modules-extra-5.15.0-37-generic
linux-modules-extra-5.15.0-39-generic
linux-modules-extra-5.15.0-40-generic
linux-modules-extra-5.15.0-41-generic

[Stefan Bader (smb)]

Guessing from the description of the symptoms and the dependency on linux-modules-extra I would suspect the problem to be in the i915 driver (which is responsible for accelerated graphics on Intel GPUs). That would also explain problems of SDM to come up. Safe mode would likely choose a much lower resolution on whatever fallback graphics driver the system comes up with.

But then breaking between -39 and -40 makes no sense since there is no change at all to that driver between those versions. Breaking between -40 and -41 would be more plausible since there are 4:

e53980320c13 drm/i915: update new TMDS clock setting defined by VBT
04d6fb5f9f53 drm/i915: Fix SEL_FETCH_PLANE_*(PIPE_B+) register addresses
bf7bdd3e16a4 drm/i915: Check EDID for HDR static metadata when choosing blc
082bac8947b5 drm/i915/display/psr: Unset enable_psr2_sel_fetch if other checks in intel_psr2_config_valid() fails

The first one to me would sound most likely but I don't have that much insight into that driver, so it is hard to make connections between the descriptions and the symptoms described.

[Hadmut Danisch (hadmut)]

I just verified: 5.15.0-40-generic definitely shows the symptoms.

[Kai-Heng Feng (kaihengfeng)]

So what's the last kernel version that doesn't have such issue?

[Hadmut Danisch (hadmut)]

5.15.0-39-generic

[Hadmut Danisch (hadmut)]

vmlinuz-5.15.0-43-generic still has the issue.

[Kai-Heng Feng (kaihengfeng)]

There's no i915 commit between Ubuntu-5.15.0-39.42 and Ubuntu-5.15.0-40.43.

Can you please attach dmesg under 5.15.0-39 and 5.15.0-40? Thanks!

[Kai-Heng Feng (kaihengfeng)]

Does kernel parameter 'intel_iommu=on' help?

[Hadmut Danisch (hadmut)]

Oh ... YES. :-)

At least at the relevant parts.

- X11 good with -40 and -43 :-)
- boot time text console good
- but console after X11 (when shutting down) still shows some jumping effect
- even after removing the nosplash option, I still get the text console instead of graphical splash at boot time

[Hadmut Danisch (hadmut)]

Interesting effect:

When the distortion is shown and I do plugin a second screen, then both screens are working well. Obviously, some sort of reconfiguration happens.

[Hadmut Danisch (hadmut)]

Unforunately, this did not fully fix the problem. Although the screen appears stable (unless shutting down), every now and then (around evern 20 Minutes) the screen shows a jerk.

[kolAflash (colaflash)]

Maybe related: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1958191#yui_3_10_3_1_1676292775910_1524

Maybe these kernel boot parameters can workaround the problem:
intel_idle.max_cstate=2 i915.enable_dc=0 i915.enable_psr=0
(if cstate 2 doesn't help, try limiting cstate to 1 or even 0)

[kolAflash (colaflash)]

P.S.
See also:
https://gitlab.freedesktop.org/drm/intel/-/issues/5569#note_1770070

And maybe try running this to fix the problem temporarily:
xrandr --output eDP-1 --off; sleep 3; xrandr --output eDP-1 --auto

[Hadmut Danisch (hadmut)]

Comment #23 indeed helped. Sorry for the delay, I was away from home. Not sure yet, which of the options is the one that actually helps.

thanks!