Re: [PATCH] BUG(): sched.c: Line 944

[Ingo Molnar]

On Tue, 17 Sep 2002, Linus Torvalds wrote:

> 	void schedule(void)
> 	{
> 		BUG_ON(in_atomic());
> 		BUG_ON(current->state & TASK_ZOMBIE);
> 		do_schedule();
> 	}
> 
> 	void exit_schedule(void)
> 	{
> 		BUG_ON(!(current->state & TASK_ZOMBIE));
> 		do_schedule();
> 	}

i'd rather do it the other way around - ie. do a:

	if (unlikely(current->state == TASK_ZOMBIE)) {
		... exit checks ...
	} else {
		... normal checks ...
	}

this test is cheaper than a function call. I'd say the ZOMBIE check alone
is not significant enough to introduce a function split in the hotpath.  
This check is also easier to remove later on.

TASK_ZOMBIE is only ever allowed to be set for exiting tasks. [and if
TASK_ZOMBIE handling is broken then it will trigger the normal atomic
tests so it's not like we are silently ignoring those errors.]

	Ingo