"Unfortunately, last time I tried to write tests for Unix sockets, I started triggering kernel panics in Windows." It seems like a Windows specific issue! A kernel panic should never occur, even when provided with invalid data.
@dropbear I think you could use the code and turn it into a test for theses patches?
I think it's best to ignore Windows <-> Unix communications tests for now as I don't think `winetest` allow running tests in both Windows and Unix space at the same time, that would require `winetest` supporting `wsl` when testing on Windows10+, and I think `wsl` support is out of scope of this PR.