http://bugs.winehq.org/show_bug.cgi?id=59333 --- Comment #16 from Hoshino Lina <lina@lina.yt> --- My understanding is IL2CPP still uses the Mono runtime and bdwgc (it just replaces the managed bytecode with native code) but I could be wrong about how all the pieces fit together in different configurations. See this random IL2CPP build tree I found: https://git.noc.ruhr-uni-bochum.de/krekem24/ANN_Visualisation_in_AR_Builds/-... It's not just any Unity app though, it's specifically Unity apps that use managed threads that perform system calls, and it's a tight race that triggers it. uOSC (what VNyan uses) has a managed thread with a blatant poll loop. It's likely other managed libraries also cause the issue. VRChat doesn't use that though, it uses this: https://github.com/stella3d/OscCore While it does use a receive thread, it uses blocking reads, which means the bug can only trigger *each time an OSC message is received*. If you don't use OSC or use it sparingly chances are ~zero. Meanwhile uOSC: https://github.com/hecomi/uOSC/blob/3c80df7ee3ce7c58cb33f5820921a3b192e22c04... ... polls 1000 times per second, no matter what. So I think this still affects ~every Unity app that uses threads, IL2CPP or not. It's just that the race is so tight, that you're only likely to hit it if the Unity app does Very Silly Things Indeed, like uOSC does. To put things into perspective: If you do use OSC with VRChat but only get messages once per frame at ~60Hz, then you're 16 times less likely to hit the bug than on VTube Studio. If it takes 5 hours to hit the bug on average on VTS, it would take 80 hours on average to hit the bug on VRChat. It's all probabilistic, you can still get unlucky and hit it in 5 minutes, but you get the idea. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.