Bug 1173748

Summary:	System hangups in dual monitor use, likely Plasma/Xorg related
Product:	[openSUSE] openSUSE Tumbleweed	Reporter:	Peter Sütterlin <P.Suetterlin>
Component:	Kernel	Assignee:	openSUSE Kernel Bugs <kernel-bugs>
Status:	RESOLVED FIXED	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None	CC:	emailej, fabian, jslaby, mbenes, mitch, tiwai
Version:	Current
Target Milestone:	---
Hardware:	Other
OS:	Other
Whiteboard:
Found By:	---	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	Log output during attempt to login

Description Peter Sütterlin 2020-07-06 09:46:39 UTC

I am experiencing repeated issues with my laptop (running current TW, 20200701) when sitting in the dock with a second monitor connected.

It usually starts with the second monitor going dark, then getting signal again (and the monitor displays the current input as info, so indeed the signal seems to have been gone?).  Sometimes, it does stay black until I move the cursor in there and/or go to a workspace that has a window on that screen.

Sometimes, in that situation where the monitor is black, moving the cursor to that screen doesn't help, but short after that the whole computer freezes dead.  It also isn't pingable via network anymore, only 5s power button switchoff helps.

I've seen it several times already over the last months, sometimes not for a longer time.  With the current release I had it 2 times already, plus a coredump of kwin after a restart (the first coredump since many months)

Unfortunately I never found anything in the logfiles after a lockup.  And the system is configured not to write coredumps, so also nothing from the kwin crash :(

In short, I don't really have a clue if that is software or hardware, or how to narrow it down, as I have so far not found a way to produce the issue.
I *think* it is only happening after the monitors went to DPMS suspend, and with no window on the second monitor.  

Hardware is a Lenovo T460p, Skylake with HD530 graphics, and Xorg based KDE/Plasma.

Maybe someone has an idea what to check/look for.

Comment 1 Peter Sütterlin 2020-08-10 09:24:16 UTC

This morning I had another crash.  The computer had been running overnight, both monitors were in DPMS suspend.  I moved the mouse, the laptop screen switched on, but not the second monitor.  I moved the mouse cursor to the second display and back (which sometimes activated it), but it stayed dark.  Then I wanted to switch workspace using the mouse scroll wheel, but the machine was dead.

Contrary to previous occurences, this time I had an error message in the log:

Aug 10 07:16:17 woodstock.pitnet kernel: i915 0000:00:02.0: [drm] *ERROR* CPU pipe B FIFO underrun
Aug 10 07:16:17 woodstock.pitnet org_kde_powerdevil[2951]: powerdevil: set screen brightness value:  852
Aug 10 07:16:17 woodstock.pitnet dbus-daemon[787]: [system] Activating service name='org.kde.powerdevil.backlighthelper' requested by ':1.31' (uid=1000 pid=2951 comm="/usr/lib64/libexec/org_kde_powerdevil ") (using servicehelper)
Aug 10 07:16:17 woodstock.pitnet dbus-daemon[787]: [system] Successfully activated service 'org.kde.powerdevil.backlighthelper'
Aug 10 07:16:17 woodstock.pitnet org_kde_powerdevil[2951]: powerdevil: Udev device changed "/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1/intel_backlight" "/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1/intel_backlight"

Currently TW 20200804, kernel 5.7.11-1-default

Comment 2 Fabian Vogt 2020-08-11 10:47:58 UTC

That's somewhat caused by the intel GPU (driver). While some issues can be caused by userspace, the complete freeze most likely isn't, so reassigning to kernel.

Comment 3 e j 2020-08-12 18:35:23 UTC

Just to confirm that I was able to install kernel-default-5.6.14-1.6.x86_64.rpm and observed normal behavior with no other changes. After installing and booting with this kernel everything is working as expected. DM greeter login screen behaves normally and I am able to initiate Plasma (Wayland) session. Seems to confirm something got broken in 5.7, 5.8 kernel.

Comment 4 Miroslav Beneš 2020-12-23 12:15:30 UTC

Peter, is this still a problem with the latest kernels in TW?

I recall there were some issues with i915 driver and a docking station. Takashi will remember more, I am sure.

Comment 5 Peter Sütterlin 2020-12-23 12:26:39 UTC

(In reply to Miroslav Beneš from comment #4)
> Peter, is this still a problem with the latest kernels in TW?
> 
> I recall there were some issues with i915 driver and a docking station.
> Takashi will remember more, I am sure.

I have not seen crashes in the last months (I typically update the system every 7-10 days).  ISTR one or two cases where the monitor did not turn on, but instead of moving the mouse pointer to that monitor and trying to click I just took the laptop out of the dock and re-inserted it.  But that is also quite some time back.

I do now and then still see short glitches, where the 2nd monitor goes black for a few seconds.  That is possibly the issues you refer to.  

HTH,

  Peter

Comment 6 Mitch Frazier 2022-01-16 20:54:52 UTC

I'm using tumbleweed from 2022-01-14 and I am seeing what sounds like a similar issue although it may not be the same issue.  The issue as I see it never gets me past the login screen, plasma/X11 crashes before the desktop even comes up.  If I disconnect my second monitor I can login without issue.  Also, using Wayland works with both monitors.

See attached system log output.

Comment 7 Mitch Frazier 2022-01-16 20:55:54 UTC

Created attachment 855320 [details]
Log output during attempt to login

Comment 8 Mitch Frazier 2022-01-16 21:04:32 UTC

Just reconnected the second monitor, as soon as I run xrandr plasma/X11 crashes.

Comment 9 Takashi Iwai 2022-01-24 15:34:27 UTC

Only judging from the log, this looks like a different problem.

Comment 10 Jiri Slaby 2023-01-27 06:33:49 UTC

Hopefully the original problem is fixed?

Comment 11 Peter Sütterlin 2023-01-27 08:08:36 UTC

(In reply to Jiri Slaby from comment #10)
> Hopefully the original problem is fixed?

I haven't seen it happen at all last year.  So yes, it's fixed.