I'm not sure it's a new bug, just that if you look up that call you'll find if it's used with any path containing a "/", then it can be raced to check somewhere

I'm not sure it's a new bug, just that if you look up that call you'll find if it's used with any path containing a "/", then it can be raced to check somewhere other that the place the original author intended. That's why symlinks MUST die.