I don't really mind the patch, but this patch doesn't actually do what
it says it does.

It factors out start_group_exit_locked() - which all looks good.

But then it also creates that new start_group_exit() function and
makes the declaration for it, and nothing actually uses it. Yet.

I'd do that second part later when you actually introduce the use in
the next patch (5/9).

Hmm?

           Linus