The ptr::write is needed because the stack vector contains bytes and the code wants to do a pointer cast and write a u64. It could be done in safe Rust with u64

The ptr::write is needed because the stack vector contains bytes and the code wants to do a pointer cast and write a u64. It could be done in safe Rust with u64::to_ne_bytes and a copy_from_slice or something like that, but since all of this code is extravagantly unsafe anyway, I think it's reasonably clear to Just Do It :)