Bug 1024913

Summary:	nvidia: X server locks then crashes while scrolling with touchpad
Product:	[openSUSE] openSUSE Distribution	Reporter:	Thomas Mitterfellner <thomas>
Component:	X11 3rd Party Driver	Assignee:	E-mail List <xorg-maintainer-bugs>
Status:	RESOLVED FIXED	QA Contact:	Stefan Dirsch <sndirsch>
Severity:	Normal
Priority:	P3 - Medium	CC:	mstaudt
Version:	Leap 42.2
Target Milestone:	---
Hardware:	Other
OS:	Other
Whiteboard:
Found By:	---	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	Xorg log

Description Thomas Mitterfellner 2017-02-11 13:14:32 UTC

Created attachment 713772 [details]
Xorg log

I had two lockups/crashes of the X server lately. In both instances, I was browsing, more precisely, scrolling, using two-finger scrolling on a synaptics touchpad, the session suddenly locked. I could only move the mouse pointer, but the session would not accept any input, neither clicks nor keyboard input (ALT+Tab, CTRL+ALT+F1 etc.). Finally, after some time, ~1-2 minutes, the session would crash, dropping me back to the login screen.
In the backtrace, synaptics_drv.so is mentioned, which is why I'm wildly guessing the touchpad driver to be the culprit.

I also found a very similar bug report (ubuntu): https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1529932

Unfortunately, I couldn't find a way to reproduce this.
I'm attaching the Xorg log showing some overflow events and backtraces; is there a simple way to get a more meaningful backtrace?

Comment 1 Thomas Mitterfellner 2017-02-11 22:09:59 UTC

I had another lock-up (it's getting more frequent it seems), but without a crash (I had to shut down eventually). Moreover, this time, I wasn't scrolling but ALT-Tabbing. The Task switcher of plasma5 was still showing half-transparently but would not go away fully. A skype session was still going on, I could hear and talk to the participant, but no keyboard/pointer interaction with the X Server was possible.

Comment 2 Stefan Dirsch 2017-02-13 10:39:29 UTC

This is a graphics driver issue wrongly pointing to the input driver. NVIDIA 375.26. There is not much you/we can do, apart from tying the next driver available by NVIDIA or switching to nouveau, modesetting or even fbdev X driver.

Comment 3 Thomas Mitterfellner 2017-02-13 12:00:30 UTC

I suspected that too, but in the linked bug report it seems to be an AMD graphics card and my girlfriend also had such a lock-up and she has Intel graphics. Either way, I installed the debug symbols for the associated packages, so if that crash happens next time, I'd get a better backtrace.

Comment 4 Max Staudt 2017-02-17 16:23:15 UTC

It would be especially helpful if you could provide us with a backtrace on the Intel machine, wich debug symbols installed (since those are unavailable for NVIDIA).

Comment 5 Max Staudt 2017-03-07 12:02:23 UTC

Any update?

Comment 6 Thomas Mitterfellner 2017-03-07 14:48:13 UTC

Neither of us experienced this crash lately (which is good!), but the debug symbols are installed, so should it happen again, we would get a useful backtrace hopefully.

Comment 7 Max Staudt 2017-03-07 16:01:19 UTC

Awesome! Sometimes, magic happens.

Let's close this ticket (assuming the issue is fixed) so it's off the table.
If the crash happens again and you have a backtrace, simply reopen it and we'll have a look.