This kind of problem is going to require some deep debugging of the surface kernel drivers. This isn’t going to be a simple or quick fix. Somebody is going to need to do some extensive debugging and analysis to chase down an issue like this. A solution to this problem could take a few hours, or it could take a few months of meticulous trial and error to narrow down the problem space and gather enough data to enable somebody to zero in on the problem.

If you want to dive into tracking down the problem yourself, I suggest starting with the kernel’s own docs on the driver architecture and debugging tools, etc.

How do you control the oomkiller? Short answer - You don’t. That’s by design. You can google for the lkml discussions about it.

What you can do is set cgroups around vsc to limit its resource consumption to improve your odds of avoiding the oomkiller’s attention entirely. Google up the cgroups docs for details.

What is there to fear? Seriously. Pretending like this has anything to do with fear is the most childishly narcissistic framing possible. Grow up.

Xorg is nearly dead and buried. Nobody actually cares about Xlibre. Notice how the only thing being mentioned is the sidenote that this fork is run by a racist troll whose been kicked for cause from several prominent OSS projects. Literally the only reason XLibre exists is because this individual needed to start his own project because he’s worn out his welcome in many others.

And, considering the geopolitical state of affairs at the moment. It’s pretty plainly obvious that the only sort of person who is “anti-DEI” are fascists. And the only people making excuses for the fascists are other fascists. So, thanks for letting us know who you are.

Found the useful idiot. You’re willfully ignoring the racist dogwhistles. That either makes you a supporter or a fellow traveler. Either way, you’re complicit and not innocent.