LinuxSucks claims that anti-cheat kernel modules are safe, by comparing them to other drivers and defending gaming companies who use them
It's only 2 years since a buggy crowdstrike windows kernel driver crashed over 8.5 million computers.
- Anti-cheat modules aren't the same as a Network driver, or storage driver which are needed for a specific purpose. If every app on your computer used a kernel module, people would clearly be unimpressed... And for some reason, some people want to give some of the worst companies a hall-pass
- The more kernel modules, the bigger the risk to the stability of the system
- The vast majority of gamers are not cheaters
- There is a trend towards trying to use User-space drivers/daemons wherever possible, unless there is a need for performance or other needs unable to be fulfilled. Anti-cheat isn't trying to address a performance issue. It's taking a clunky approach at just grabbing access to your entire computer because cheaters take the same approach. This is often to reduce the amount of actual moderation required
- There are a lot of general strategies which are effective at detecting cheating, by recording keystrokes (like in trackmania) or even by placing random obstacles in paths (Wow), and watching who starts running into them.
- These modules are generally closed source, so there is no real way to audit what they are actually doing. This could include reading personal information and returning it.
- If the anti-cheat kernel module isn't maintained, there is a real risk it won't run correctly on future OS's (whether Linux, MacOS or Windows). MacOS in particular has had issues with kernel module breakage in the past during even minor updates. So, long term maintainability is reduced
- Anti-cheat isn't a GPU driver. GPU's are needed by EVERYTHING on the computer, and the idea is to have the minimum amount in the kernel module to make everything work. Anti-cheat modules are generally only used by 1 or 2 games
Whether or not you agree with them or disagree, there is a clear risk to games installing lots of kernel modules on a computer.
The good news is that there aren't many games which rely on kernel modules to run.
1 Comments
TootSweet@lemmy.world · 1 pts · 12d
Yeah, this one makes it clear how little Madthumbs understands.
I personally, on my main box, build my own kernel and have kernel module loading completely disabled. The functionality built into my kernel is the functionality that my kernel has and I don't allow any module to be loaded ever.
I mean, Madthumbs knows that "Reddit does evil shit" is the whole reason Lemmy exists, right? So yeah. When a company does evil shit, it's noble to boycott. Plain and simple.
Is he referring to Linux lacking a "rootkit system call"? (I mean, Linux doesn't lack a "rootkit system call". Modules are a thing, and lots of modules manage to keep working with the Linux ABI just fine.) What's "standardized signing" exactly? Maybe he's talking about "Trusted Computing" or something? Is he talking about something that Windows does have?
I mean, I can't speak for "Linux advocates", but I don't. Not that I wouldn't trust VirtualBox or ZFS drivers/modules because they are FOSS. But yeah. Proprietary graphics drivers, especially when there are FOSS alternatives, are a no-go in my book.
I don't know what he's talking about here. Wine invokes Linux kernel interrupts, but you know what else does that? Literally every Linux executable anyone has ever executed. Any basic "hello world" application that anyone writes in C or x86 assembly does kernel interrupts. If there's anything more "kernel-adjacent" in Wine than that, I'm not aware of it. (I suppose it could be said to somewhat "impersonate" the Windows kernel to Windows executables... but that doesn't seem relevant to Madthumbs' point. :shrug: ) Again, not that Wine doing "kernel-adjacent" things would necessarily be a deal breaker for me given that it's FOSS. But also, I don't have Wine installed right now.
On to something from this post.
Yeah, any client-side anti-cheat is already going about it the wrong way. Nobody hardens their banking webapp with kernel modules or rootkits. They close vulnerabilities by fixing the server. Plus, I'm not sure it would be possible to do FOSS client-side anti-cheat. (At least I can't think of a way to make that work given the 5 minutes I've spent thinking about it.) Client-side anti-cheat is necessarily antagonistic to players and FOSS is kindof designed to prevent software from doing adversarial actions to users, so it makes sense that the two would be incompatible.
Maybe the world would benefit from some authoritative source "writing the book" on how to properly do server-side anti-cheat. 🤔