> problem Ulrich mentioned - it splits up ->user_tid into ->set_child_tid
> and ->clear_child_tid pointers. This way the clearing and setting
> functionality is cleanly separated.
How about making ->set_child_tid a parameter for schedule_tail, or even
directly using it in the ret_from_fork assembly?
It doesn't make much sense to have a variable in task_struct which is
used only at task creation.