X server hard crash when playing youtube videos (Nvidia GPU)

[chconnor]

Howdy -- every now and then (maybe once every couple days) my X server dies completely. Always happens when I start to play a video... IIRC it's only via youtube, but it might happen other times (I'll start paying more attention now.) The desktop session dies and I get back to the login prompt.

After logging back in, Xorg.0.log.old has:

[ 1041.827] (EE)
[ 1041.827] (EE) Backtrace:
[ 1041.829] (EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x139) [0x56301af806e9]
[ 1041.829] (EE) 1: /lib/x86_64-linux-gnu/libc.so.6 (__sigaction+0x50) [0x7f8765159520]
[ 1041.829] (EE) 2: /usr/lib/xorg/Xorg (NewCurrentScreen+0x1b9) [0x56301ae151e9]
[ 1041.830] (EE) 3: /usr/lib/xorg/Xorg (MaybeDeliverEventsToClient+0x4d5) [0x56301ae16c85]
[ 1041.830] (EE) 4: /usr/lib/xorg/Xorg (MaybeDeliverEventsToClient+0x9d3) [0x56301ae17183]
[ 1041.830] (EE) 5: /usr/lib/xorg/Xorg (WindowsRestructured+0x163) [0x56301ae18183]
[ 1041.830] (EE) 6: /usr/lib/xorg/Xorg (InitProximityClassDeviceStruct+0x1fdd) [0x56301aef6a7d]
[ 1041.830] (EE) 7: /usr/lib/xorg/Xorg (XkbHandleActions+0x1dc) [0x56301af20e4c]
[ 1041.830] (EE) 8: /usr/lib/xorg/Xorg (XkbRemoveResourceClient+0x7c1) [0x56301af19f21]
[ 1041.830] (EE) 9: /usr/lib/xorg/Xorg (XkbRemoveResourceClient+0x9be) [0x56301af1a11e]
[ 1041.830] (EE) 10: /usr/lib/xorg/Xorg (TimerSet+0x170) [0x56301af79c60]
[ 1041.831] (EE) 11: /usr/lib/xorg/Xorg (WaitForSomething+0x258) [0x56301af79ee8]
[ 1041.831] (EE) 12: /usr/lib/xorg/Xorg (SendErrorToClient+0x117) [0x56301ae0a257]
[ 1041.831] (EE) 13: /usr/lib/xorg/Xorg (InitFonts+0x3c4) [0x56301ae0e524]
[ 1041.831] (EE) 14: /lib/x86_64-linux-gnu/libc.so.6 (__libc_init_first+0x90) [0x7f8765140d90]
[ 1041.831] (EE) 15: /lib/x86_64-linux-gnu/libc.so.6 (__libc_start_main+0x80) [0x7f8765140e40]
[ 1041.831] (EE) 16: /usr/lib/xorg/Xorg (_start+0x25) [0x56301adf75f5]
[ 1041.831] (EE)
[ 1041.831] (EE) Segmentation fault at address 0x7fff72e48000
[ 1041.831] (EE)
Fatal server error:
[ 1041.831] (EE) Caught signal 11 (Segmentation fault). Server aborting
[ 1041.831] (EE)
[ 1041.831] (EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
[ 1041.831] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[ 1041.831] (EE)
[ 1046.920] (WW) NVIDIA: Wait for channel idle timed out.
[ 1047.141] (EE) Server terminated with error (1). Closing log file.​

System is an i7-4770K with a GTX 1060. (Nothing in the system is stressed or near limits; happens during casual usage.) Using nvidia driver 470.141.03.

Should I report this as an X bug? An nvidia bug?

[Snowhog]

[ 1041.831] (EE) Segmentation fault at address 0x7fff72e48000

Segmentation faults are always memory addresses conflicts. Could be bad programming on the part of an application, or a RAM chip beginning to fail.

[chconnor]

Thanks -- meaning the X server isn't necessarily to blame? (It is always the X server that dies...)

[chconnor]

I've collected some more info about this bug and want to report it, but... I can't for the life of me figure out where the appropriate place would be. Any tips on that? Thanks!

[oshunluvr]

I've collected some more info about this bug and want to report it, but... I can't for the life of me figure out where the appropriate place would be. Any tips on that? Thanks!

For info about bug reporting: https://community.kde.org/Get_Involved/Issue_Reporting

To go right to bug reporting: https://bugs.kde.org

[chconnor]

Ah, thanks. I had interpreted "Have you found a problem with KDE Software?" to mean KDE-specific software, and that upstream bugs should be filed upstream. Also the bug report form there offers Applications, Plasma, KDE Neon, Websites, Frameworks and Libraries, Developer tools, Miscellaneous, defined as "KDE software that doesn't fit into any of the above groups", and "I don't know"... I'll file it under "I don't know", I guess? I also don't see any X server bugs filed there (only searched a little though) which made me hesitate...

[claydoh]

You might need to report it on Ubuntu's Launchpad, since this seems to be involving the graphics server.

The easiest way is to report this immediately after logging in after the crash:

Hit alt-space or alt-f2 to open krunner, and enter ubuntu-bug xorg
If for some reason this doesn't work in krunner, enter this in Konsole.
The correct item to report the bug against can be changed or added to later on if necessary. A simple GUI walks you through a bunch of things, and gather logs and other info to send to Launchpad.

The long-winded general process is described here: https://help.ubuntu.com/community/ReportingBugs It is overly detailed for your purposes, but is a good reference.

Gather as much descriptive info as you can - which browser and version you are playing Youtube on, if you have tried other browsers, or had this happen during playback in other applications, or any other instances.

[chconnor]

Thanks! -- I'll do that next time it crashes (and will update this post with a link to the bug.)

[axelmasok]

The crash is infrequent but at least you know a trigger. Have you thought about running a Live copy from DVD/USB and giving it a flogging with browser and You tube videos? The other difference is the live instance won't be running the Nvidia driver. This test might tease out a memory or other hardware issue.

[chconnor]

Thanks, I might give that a shot... the crash is pretty infrequent, though, on the order of once a week. Sometimes it'll happen twice a day... For whatever it's worth, I did run the BIOS memory test and have also confirmed that GPU memory usage is not significant (some threads online about an X server crash at the same method call mentioned that it was a common failure point that was actually triggered due to running out of GPU memory).

[chconnor]

Bug posted: https://bugs.launchpad.net/ubuntu/+s...g/+bug/1995766