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

[Roger Larsson <roger.larsson@norran.net>]

On torsdagen den 4 april 2002 21.26, Robert Love wrote:
> On Thu, 2002-04-04 at 14:14, Linus Torvalds wrote:
> > The answer is that preempt_schedule() illegally sets
> >
> > 	current->state = TASK_RUNNING;
> >
> > without asking the process whether that's OK. The SMP code never does
> > anything like that.
>
> Well Ingo added that ;)
>
> We used to just set a flag in the preempt_count that marked the task as
> preempted and made sure on its next trip into schedule it ran again.
>

How about doing:

asmlinkage void preempt_schedule(void)
{
	unsigned long saved_state;

	if (unlikely(preempt_get_count()))
		return;

	preempt_disable(); /* or use an atomic operation */
	saved_state = current->state;
	current->state = TASK_RUNNING;
	preempt_enable_no_resched(); /* we are scheduling anyway... */
	schedule();
	current->state = saved_state;
}

It is unlikely to get preemption between schedule() and the
setting since schedule it self checks - the window is small.
And when it hits if will correctly restore the correct value.

Note this code does not need to solve the FLAG problem.

	 current->state |= FLAG
	* PREEMPT *
		current->state |= FLAG
		schedule()
		current->state &= ~FLAG
	schedule() with flag disabled


/RogerL

-- 
Roger Larsson
Skellefteå
Sweden