For those using the upstream open-source NVIDIA Linux driver “Nouveau”, with a pending fix coming for Linux 6.17 and existing kernel releases it should be a much more stable and reliable experience.
Instability issues have been somewhat common for the Nouveau driver usage but up until recently were hard to track down and reproduce. Fortunately, thanks to the stellar work of Faith Ekstrand and David Airlie, they managed to now address this in large part. Faith Ekstrand explained it’s been a fencing bug where the context would randomly wedge and could lead to the app/game freezing and losing your X.Org Server / compositor in the process. After many bug reports over the past year, a fix is on the way.
The patches posted by David Airlie on Friday is the missing sauce to a more reliable Nouveau experience:
“Nouveau has code that when it gets an IRQ with no allowed handler it disables it to avoid storms.
However with nonstall interrupts, we often disable them from the drm driver, but still request their emission via the push submission.
Just don’t disable nonstall irqs ever in normal operation, the event handling code will filter them out, and the driver will just enable/disable them at load time.
This fixes timeouts we’ve been seeing on/off for a long time, but they became a lot more noticable on Blackwell.
This doesn’t fix all of them, there is a subsequent fence emission fix to fix the last few.”
The less than two dozen lines of new code should provide a more reliable Nouveau driver experience in terms of stability. The patches are marked for back-porting once it makes it into Linux 6.17. The patches were not part of last week’s DRM fixes for Linux 6.17 but will hopefully make it for this week’s collection of DRM fixes ahead of Linux 6.17-rc5 and then onto the stable branches.
