Kernel 5.15.0-86 causes LightDM to disappear after 1s

[Serge Stroobandt]

This is to report a critical bug introduced with Linux kernel 5.15.0-86-generic and newer.
It does not occur with 5.15.0-84-generic and previous kernels.

I am running Xubuntu LTS 22.04.03 on several desktops, but this bug only manifests on this specific system, which is a fairly popular HP t630 Thin Client featuring an AMD Embedded G-Series GX-420GI system on chip with Radeon R7E integrated graphics. Here is the neofetch output:

       :ydddddddddddddddddddddddy: OS: Xubuntu 22.04.3 LTS x86_64
    -yddddddddddddddddddddhdddddddy- Host: HP t630 Thin Client
   odddddddddddyshdddddddh`dddd+ydddo Kernel: 5.15.0-84-generic
 `yddddddhshdd- ydddddd+`ddh.:dddddy` Uptime: 1 day, 1 hour, 16 mins
 sddddddy /d. :dddddd-:dy`-ddddddds Packages: 3354 (dpkg), 22 (flatpak)
:ddddddds /+ .dddddd`yy`:ddddddddd: Shell: bash 5.1.16
sdddddddd` . .-:/+ssdyodddddddddds Resolution: 1600x900
ddddddddy `:ohddddddddd DE: Xfce
dddddddd. +dddddddd WM: Xfwm4
sddddddy ydddddds WM Theme: Numix
:dddddd+ .oddddddd: Theme: Greybird [GTK2/3]
 sdddddo ./ydddddddds Icons: Faenza-Darker [GTK2], elementary-xfce-darker [GTK3]
 `yddddd. `:ohddddddddddy` Terminal: xfce4-terminal
   oddddh/` `.:+shdddddddddddddo Terminal Font: DejaVu Sans Mono 12
    -ydddddhyssyhdddddddddddddddddy- CPU: AMD Embedded G-Series GX-420GI Radeon R7E (4) @ 2.000GHz
      :yddddddddddddddddddddddddy: GPU: AMD ATI Radeon R5/R6/R7 Graphics
        .+yddddddddddddddddddy+. Memory: 2786MiB / 31044MiB

Symptoms:
The systems boots as usual, XFCE's LightDM Greeter appears for about one second, then the screen turns completely black and the keyboard becomes unresponsive. This implies that one cannot switch to a console screen using Ctrl+Alt+F1 etc. to resolve any problems.

I was lucky to have SSH access from another system, to change GRUB settings and reboot with the older 5.15.0-84 kernel, otherwise this system could have been a loss. More importantly, this proofs the system did not crash.

dmesg logs were compared between kernels using meld, but no noticeable differences were detected.
However, both working (84) and non-working (86,88) kernel dmesg logs show similar ACPI errors; here for the 84 kernel, the 88 kernel dmesg log is attached to this bug report:

[ 3.584496] acpi PNP0C14:01: duplicate WMI GUID 5FB7F034-2C63-45E9-BE91-3D44E2C707E4 (first instance was on PNP0C14:00)
[ 3.584510] acpi PNP0C14:01: duplicate WMI GUID 05901221-D566-11D1-B2F0-00A0C9062910 (first instance was on PNP0C14:00)

...

[ 6.920766] ACPI BIOS Error (bug): Attempt to CreateField of length zero (20210730/dsopcode-133)

[ 6.921268]
               Initialized Local Variables for Method [HWMC]:
[ 6.921272] Local0: 000000006835d165 <Obj> Integer 0000000000000004
[ 6.921293] Local1: 000000000d18c184 <Obj> Buffer(12) 00 00 00 00 00 00 00 00
[ 6.921317] Local5: 00000000a100fffd <Obj> Integer 0000000000000000

[ 6.921332] Initialized Arguments for Method [HWMC]: (2 arguments defined for method invocation)
[ 6.921335] Arg0: 00000000ca1a283d <Obj> Integer 0000000000000002
[ 6.921346] Arg1: 000000007cb48472 <Obj> Buffer(144) 53 45 43 55 01 00 00 00

[ 6.921379] ACPI Error: Aborting method \HWMC due to previous error (AE_AML_OPERAND_VALUE) (20210730/psparse-529)
[ 6.921491] ACPI Error: Aborting method \_SB.WMID.WMAA due to previous error (AE_AML_OPERAND_VALUE) (20210730/psparse-529)
[ 6.992738] input: HP WMI hotkeys as /devices/virtual/input/input13
[ 7.012049] ACPI BIOS Error (bug): Attempt to CreateField of length zero (20210730/dsopcode-133)

[ 7.012149]
               Initialized Local Variables for Method [HWMC]:
[ 7.012152] Local0: 00000000a6c78dff <Obj> Integer 0000000000000004
[ 7.012167] Local1: 000000004991298c <Obj> Buffer(12) 00 00 00 00 00 00 00 00
[ 7.012189] Local5: 00000000c8c4ed33 <Obj> Integer 0000000000000000

[ 7.012202] Initialized Arguments for Method [HWMC]: (2 arguments defined for method invocation)
[ 7.012205] Arg0: 00000000ca1a283d <Obj> Integer 0000000000000002
[ 7.012215] Arg1: 00000000be9a5486 <Obj> Buffer(144) 53 45 43 55 01 00 00 00

[ 7.012239] ACPI Error: Aborting method \HWMC due to previous error (AE_AML_OPERAND_VALUE) (20210730/psparse-529)
[ 7.012332] ACPI Error: Aborting method \_SB.WMID.WMAA due to previous error (AE_AML_OPERAND_VALUE) (20210730/psparse-529)

...

[27661.868977] amdgpu 0000:00:01.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0000 address=0x1070f6720 flags=0x0070]
[27661.868996] amdgpu 0000:00:01.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0000 address=0x1070f6730 flags=0x0070]
[27661.869006] amdgpu 0000:00:01.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0000 address=0x1070f6690 flags=0x0070]
[27661.869014] amdgpu 0000:00:01.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0000 address=0x1070f6680 flags=0x0070]

...

[27663.177627] hid-generic 0003:260D:1019.0002: offset (3) exceeds report_count (3)
[27663.177642] hid-generic 0003:260D:1019.0002: offset (4) exceeds report_count (3)

I am happy to provide any other log files to get to the bottom of this kernel regression.

ProblemType: Bug
DistroRelease: Ubuntu 22.04
Package: linux-image-5.15.0-86-generic 5.15.0-86.96
ProcVersionSignature: Ubuntu 5.15.0-84.93-generic 5.15.116
Uname: Linux 5.15.0-84-generic x86_64
ApportVersion: 2.20.11-0ubuntu82.5
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC1: serge 1046 F.... pulseaudio
 /dev/snd/pcmC1D0p: serge 1046 F...m pulseaudio
 /dev/snd/controlC0: serge 1046 F.... pulseaudio
CasperMD5CheckResult: unknown
CurrentDesktop: XFCE
Date: Fri Nov 10 13:23:37 2023
InstallationDate: Installed on 2020-12-06 (1068 days ago)
InstallationMedia: Xubuntu 20.04.1 LTS "Focal Fossa" - Release amd64 (20200731)
IwConfig:
 lo no wireless extensions.

 enp1s0 no wireless extensions.
MachineType: HP HP t630 Thin Client
ProcFB: 0 amdgpudrmfb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-5.15.0-84-generic root=UUID=135cac5b-904e-4d19-80c1-c3ca0b0ac1c8 ro quiet splash vt.handoff=7
RebootRequiredPkgs: Error: path contained symlinks.
RelatedPackageVersions:
 linux-restricted-modules-5.15.0-84-generic N/A
 linux-backports-modules-5.15.0-84-generic N/A
 linux-firmware 20220329.git681281e4-0ubuntu3.21
RfKill:

SourcePackage: linux
UpgradeStatus: Upgraded to jammy on 2022-08-19 (447 days ago)
dmi.bios.date: 03/11/2019
dmi.bios.release: 5.11
dmi.bios.vendor: AMI
dmi.bios.version: M40 v01.10
dmi.board.name: 8158
dmi.board.vendor: HP
dmi.board.version: A01
dmi.chassis.asset.tag: 8CN90200WG
dmi.chassis.type: 3
dmi.chassis.vendor: HP
dmi.modalias: dmi:bvnAMI:bvrM40v01.10:bd03/11/2019:br5.11:svnHP:pnHPt630ThinClient:pvr:rvnHP:rn8158:rvrA01:cvnHP:ct3:cvr:sku3JJ22ES#ABH:
dmi.product.family: 103C_53302C G=D
dmi.product.name: HP t630 Thin Client
dmi.product.sku: 3JJ22ES#ABH
dmi.sys.vendor: HP

[Serge Stroobandt (serge-stroobandt)]

Booting 5.15.0-88-generic with acpi=off does not resolve the issue.

[Daniel Letzeisen (dtl131)]

Can you boot completely with 'nomodeset'?

[Serge Stroobandt (serge-stroobandt)]

Yes! I tried with kernel 5.15.0-91-generic now, and the system boots with LigthDM remaining on screen. Thanks!

[Serge Stroobandt (serge-stroobandt)]

With kernel 5.15.0-84-generic and previous, setting nomodeset was not necessary.

[Serge Stroobandt (serge-stroobandt)]

Interestingly, the information about nomodeset cannot be found in the [Ubuntu Wiki about kernel debugging](https://wiki.ubuntu.com/Kernel/Debugging).

I had to resort to [AskUbuntu](https://askubuntu.com/questions/207175/what-does-nomodeset-do), which refers to an [Ubuntu Forum post](https://ubuntuforums.org/showthread.php?t=1613132) for more information.

So, this may be merely a documentation issue. Thanks again.

[Serge Stroobandt (serge-stroobandt)]

By setting nomodeset in grub, I may be able to now run kernel 5.15.0-91-generic, however this resulted in the following new issues:
- No ability to resume (graphics?) from suspend
- No graphic hardware acceleration available