Xorg SIGSEGV in nvidia_drv.so during FreePixmap — driver 595.71.05 (open), RTX 5080 - repeatable

Graphics / Linux Linux
, , ,
1

Xorg reliably crashes with SIGSEGV in the FreePixmap path, with the immediate caller inside nvidia_drv.so (offset +0x235894). Happens on the open-kernel driver 595.71.05 with an RTX 5080 on Fedora 44.

Reproduce:

Running “High on Life” video game via Steam, using Steam Linux Runtime. Installed via Flatpak.
Happens regardless of Proton / Wine version used also.

Environment:

  • GPU: NVIDIA GeForce RTX 5080
  • Driver: 595.71.05, open kernel modules
  • Kernel: Linux fedora 7.0.7-200.fc44.x86_64 #1 SMP PREEMPT_DYNAMIC Thu May 14 17:19:46 UTC 2026 x86_64 GNU/Linux
  • Xorg: xorg-x11-server 21.1.22-1.fc44
  • Distro: Fedora 44
  • Display manager: LightDM
  • Desktop / WM: xfce4

Crash signature:

From coredumpctl info:

Signal: 11 (SEGV)
#0  0x000000000053ee20 n/a (n/a + 0x0)
#1  0x00007fa64dc35894 n/a (/usr/lib64/xorg/modules/drivers/nvidia_drv.so + 0x235894)

From gdb (xorg-server has debug symbols, nvidia_drv.so does not):

#0  OsSigHandler (signo=11, sip=..., unused=...) at ../os/osinit.c:114
#1  <signal handler called>
#2  FreePixmap (pPixmap=0x3347a580) at ../dix/pixmap.c:130
#3  0x00007fa64e01d4f9 in ?? ()  [in nvidia_drv.so, ~2 MiB past the offset above]

The crash is in pixmap teardown, called from nvidia_drv.so. pPixmap may be stale by the time FreePixmap is entered — I wasn’t able to inspect it because gdb couldn’t load shared library symbols from the core (the full backtrace is in the attached file).

gdb reported ‘No shared libraries loaded’ for this core, so frame #3’s address is unresolved; the coredumpctl output above places a frame inside nvidia_drv.so at +0x235894, which may or may not correspond to the same frame.

Note: above summary has been largely generated via AI as I am well outside my experience with all this stuff. Hopefully it has everything you need.

coredumpctl-info.txt (4.7 KB)

nvidia-bug-report.log.gz (914.0 KB)

xorg-backtrace.txt (3.8 KB)

xorg.core.gz (10.9 MB)

2

Following up on this as it seems consistent regardless of Wine / Proton version used. Is repeatable and happens if game is in full-screen or windowed mode.

Is there any more information needed for NVIDIA engineers to look at this?

I am adding an addition backtrace which unwinds further.

Same nvidia_drv.so offset, the X-server entry point is damageDestroyPixmap and crash site is FreePixmap - possibly triggered by display HPD events.

The only change from the previously uploaded report is that this one is replicated on the latest kernel (i.e. 7.0.9-204.fc44.x86_64 - which was released after my original post ).

repeated-crash-core-and-nvdebug-files.zip (14.6 MB)

3

Thanks @nvidiage1

I have filed a bug 6205955 internally for tracking purpose.

Team will attempt to duplicate issue locally and get back if required any additional information.

4

@nvidiage1 I tried to replicate the issue inhouse with 5080+ Fedora but do not see repro for the issue,

  1. Does the Xorg crash occur at a specific location, checkpoint, or gameplay event in High on Life? if so, can you provide the saved game file, I can try with it.Does it occur with any other game ?

  2. Which Proton version was used when the crash occurred?

  3. Is the game running in fullscreen, borderless, or windowed mode when the issue occurs?

  4. Are any specific graphics settings required ?

  5. Does the issue occur during gameplay only, or can it also occur while alt-tabbing, opening the Steam overlay, loading screen, changing graphics settings, or exiting the game?

  6. Have you observed the issue on any GPU other than the RTX 5080?

  7. Have you observed any XID failure anytime when the issue occurs?

~Thanks

5

Happens when loading the game. The game / desktop crashes and I’m dropped back at the lightdm login screen when trying to run the game. Doesn’t seem to happen with any other games.

Any / all versions tested: ProtonGE10-34, Proton-Hotfix, Proton-Experimental, Proton 11, Proton 10.

Happens regardless of Full Screen or Windowed mode.

Not that I can tell.

When starting the game.

No - seems related specifically to NVIDIA drivers. The issue happens with 595.80 and 610.43.02 drivers also.

Nope.

6

Thanks for the reply