Crash (probably of X11) after an image is opened with gimp

Hello,
I just noticed that opening any image with gimp results in a immediate crash, probably of X11, that brings me back to login. I’ve been using gimp for years on this machine, I guess this may be related to the update from Kubuntu 19.10 to Kubuntu 20.04, which also updated my nVidia drivers I think.
Crash seems deterministic. As soon as the image is opened, I get the crash.
I have no idea what is causing the issue, but I tried to switch to Intel HD from the nVidia settings, and I could not reproduce the crash anymore.
This is the log (I reproduced the issue twice immediately before creating the logs): nvidia-bug-report.log (349.8 KB)

Any idea what could be causing this?

There’s actually no crash of Xorg visible in the logs, neither any errors from the nvidia driver.

I don’t actually know if there is X11 crash. What I see is that I’m back to the login and the session is closed. I don’t think that can be a simple gimp crash, right?
I see these logs in the archive:

[ 7493.948] (EE)
[ 7493.948] (EE) Backtrace:
[ 7493.954] (EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x139) [0x560b1fd89889]
[ 7493.954] (EE) 1: /lib/x86_64-linux-gnu/libpthread.so.0 (funlockfile+0x50) [0x7ff0c53cff8f]
[ 7493.955] (EE) 2: /lib/x86_64-linux-gnu/libc.so.6 (gsignal+0xc7) [0x7ff0c5212ed7]
[ 7493.955] (EE) 3: /lib/x86_64-linux-gnu/libc.so.6 (abort+0x121) [0x7ff0c51f4535]
[ 7493.955] (EE) unw_get_proc_name failed: no unwind info found [-10]
[ 7493.955] (EE) 4: /lib/x86_64-linux-gnu/libc.so.6 (?+0x0) [0x7ff0c51f4400]
[ 7493.955] (EE) 5: /lib/x86_64-linux-gnu/libc.so.6 (__assert_fail+0x42) [0x7ff0c5204012]
[ 7493.955] (EE) 6: /usr/lib/xorg/Xorg (dixRegisterPrivateKey+0x237) [0x560b1fc49017]
[ 7493.955] (EE) 7: /usr/lib/xorg/modules/libglamoregl.so (glamor_init+0xc9) [0x7ff0c01b0759]
[ 7493.956] (EE) unw_get_proc_name failed: no unwind info found [-10]
[ 7493.956] (EE) 8: /usr/lib/xorg/modules/drivers/modesetting_drv.so (?+0x0) [0x7ff0c0805a50]
[ 7493.956] (EE) unw_get_proc_name failed: no unwind info found [-10]
[ 7493.956] (EE) 9: /usr/lib/xorg/modules/drivers/modesetting_drv.so (?+0x0) [0x7ff0c07fd970]
[ 7493.956] (EE) 10: /usr/lib/xorg/Xorg (AddGPUScreen+0xf0) [0x560b1fc2ad90]
[ 7493.956] (EE) 11: /usr/lib/xorg/Xorg (xf86PlatformMatchDriver+0x9fc) [0x560b1fc8593c]
[ 7493.956] (EE) 12: /usr/lib/xorg/Xorg (xf86PlatformDeviceCheckBusID+0x201) [0x560b1fc8aac1]
[ 7493.956] (EE) 13: /usr/lib/xorg/Xorg (config_fini+0xf79) [0x560b1fc87a49]
[ 7493.956] (EE) 14: /usr/lib/xorg/Xorg (config_fini+0x15cb) [0x560b1fc88d2b]
[ 7493.956] (EE) 15: /usr/lib/xorg/Xorg (config_init+0x9) [0x560b1fc86239]
[ 7493.956] (EE) 16: /usr/lib/xorg/Xorg (InitInput+0xc7) [0x560b1fc69c37]
[ 7493.956] (EE) 17: /usr/lib/xorg/Xorg (InitFonts+0x36f) [0x560b1fc2e97f]
[ 7493.957] (EE) 18: /lib/x86_64-linux-gnu/libc.so.6 (__libc_start_main+0xeb) [0x7ff0c51f5b6b]
[ 7493.957] (EE) 19: /usr/lib/xorg/Xorg (_start+0x2a) [0x560b1fc1867a]

I’m not a X11 expert, so I don’t frequently read X11 logs, can this be related?
Do you have any advice to collect some info about this? I have no idea where the issue can be, I’m asking here only because switching to Intel, everything works as expected.
Regards.

That backtrace was from a log from back last year and a crash in the modesetting driver. Completely unrelated.
Taking your other problem with scaled vmbox without 3D accel into account, I’m leaning towards your system memory being defective. The nvidia driver is especially touchy with that. So please check system memory by removing modules and trying to reproduce the crash.

This is a Dell laptop, and warranty is active here. I cannot therefore open it, remove things, etc… I could call Dell and tell them there is a hardware issue, but I really have no proof there is a hardware issue here. I checked the system with the Dell diagnostic tools, and nothing was found defective. I understand that tool is not providing any guarantee though.

I tested with nVidia driver version 390.132. I cannot reproduce any issue here. Gimp works properly and VirtualBox scales without crashing the system. Can this info be of help?

Maybe. Please check if prime sync is involved by disabling it when using the 440 driver:
xrandr --output eDP-1-1 --set “PRIME Synchronization” “0”

Hello,
both crashes remain after executing that command.

I’m out of ideas, looks like some machine specific regression in the driver. Better take the last nvidia-bug-report.log and mail it with a description to linux-bugs[at]nvidia.com

Hello,
ok. You are referring to nvidia-bug-report_2.log (346.9 KB) right?
Do you think I should report both these crashes together or separately?
Are there specific guidelines to report the bug?
Thanks for your help.

Take a look at https://gitlab.freedesktop.org/xorg/xserver/-/issues/980

Actually, you should start reporting the vbox issue since the logs for the gimp issue don’t show any crash at all. Guidelines - freestyle.

Hello,
I reported the other issue and I’m providing info, thanks for your help.

This issue with gimp seems to be instead even more severe. It seems there are many others with the same issue: https://gitlab.gnome.org/GNOME/gimp/-/issues/4317. I checked with gdb and I can confirm xorg is crashing. I’m still missing some symbols but seems to be cursor related for me as well. I have no idea where the bug actually lies, but do you know if nVidia is investigating this somehow?
Thank you for your assistance.

Good question, since this cursor bug comes in many flavours and is investigated by distro devs, I can only hope that they’ve contacted nvidia.

Found this post when experiencing the same problem. Would crash to login screen whenever mouse cursor moved into canvas area of Gimp, every time; plus sometimes on Firefox displayed webpages, when images that were zoomed and mouse entered image area.

Am running Mint Cinnamon 20 [Kernel: 5.4.0-42-generic x86_64 bits: 64 compiler: gcc v: 9.3.0 Desktop: Cinnamon 4.6.7; Dell XPS 9550 with 3840x2160 HiDPI display.
wm: muffin dm: LightDM Distro: Linux Mint 20 Ulyana base: Ubuntu 20.04 focal ]
Was using Nvidia prime: version 440.100-ubuntu0.20.04.1,

Would not crash with Intel graphics chosen, nor nouveau.
Used driver manager to choose older nvidia-driver-390, version 390.138-0ubuntu0.20.04.1;
******** Crashes fixed! ***************** for now… Hope next driver, (460 when added to repositories), will not have same issue.

(Similar posting found at - https://gitlab.gnome.org/GNOME/gimp/-/issues/4317. Others reported same condition with Libreoffice, but I hadn’t seen it there, though don’t recall using lots of enlarged images there.)

Can you please try driver 450.57 or newer? One of the listed fixes in that release is
Fixed a bug that could cause the X server to crash when using large cursor images with PRIME display offloading.

Hello,
I installed 450.57 and both gimp and virtualbox do not seem to crash anymore.

1 Like

Great, thanks for confirming!

Agree. Linux mint instructions to add repository were found at: https://easylinuxtipsproject.blogspot.com/p/nvidia.html
https://launchpad.net/~graphics-drivers/+archive/ubuntu/ppa

Thanks for your answer.