Re: Patch: linux-2.5.8-pre1/kernel/exit.c change caused BUG() at

[Linus Torvalds]

On 4 Apr 2002, Robert Love wrote:
> 
> Curiously, what is wrong with the way we did it in the original patch? 
> I.e. set a bit in preempt_count and use that to skip to pick_next_task
> in schedule.  This allows us to preempt any task of any state ...

Because that requires that every user of "set_task_state()" needs to know 
about preemption.

Btw, I think entry.S should just call preempt_schedule() instead, instead 
of knowing about these details.

		Linus