Re: [PATCH 2/2 v4] softlockup: check all tasks in hung_task

[Mandeep Singh Baines <msb@google.com>]

Andrew Morton (akpm@linux-foundation.org) wrote:
> On Thu, 5 Feb 2009 15:34:53 +0100 Ingo Molnar <mingo@elte.hu> wrote:
> 
> > 
> > Subject: [PATCH] softlockup: check all tasks in hung_task
> > 
> > Impact: extend the scope of hung-task checks
> > 
> 
> A nanonit:
> 
> > +static const int hung_task_batching = 1024;
> 
> static const definitions look pretty but they're a bit misleading.
> 
> >  static void check_hung_uninterruptible_tasks(unsigned long timeout)
> >  {
> > +	int batch_count = hung_task_batching;
> >  	int max_count = sysctl_hung_task_check_count;
> >  	unsigned long now = get_timestamp();
> >  	struct task_struct *g, *t;
> > @@ -131,6 +159,13 @@ static void check_hung_uninterruptible_tasks(unsigned long timeout)
> >  	do_each_thread(g, t) {
> >  		if (!--max_count)
> >  			goto unlock;
> > +		if (!--batch_count) {
> > +			batch_count = hung_task_batching;
> > +			rcu_lock_break(g, t);
> > +			/* Exit if t or g was unhashed during refresh. */
> > +			if (t->state == TASK_DEAD || g->state == TASK_DEAD)
> > +				goto unlock;
> > +		}
> >  		/* use "==" to skip the TASK_KILLABLE tasks waiting on NFS */
> >  		if (t->state == TASK_UNINTERRUPTIBLE)
> >  			check_hung_task(t, now, timeout);
> 
> The reader of this area of the code will expect that hung_task_batching
> is a variable.  It _looks_ like the value of that variable can be altered
> at any time by some other thread.  It _looks_ like this code will explode
> if someone has accidentally set hung_task_batching to zero, etc.
> 

The code would not break if hung_task_batching was exported out as a sysctl.

If hung_task_batching is set to zero at any time, the behavior will be
to process all tasks in one batch. This seems like a reasonable behavior
for the  zero case. It is also consistent with the behavior of
sysctl_hung_task_check_count.

Maybe a comment should be added by the declaration of both variables explaining
the zero behavior?

> But none of that is actually true, because hung_task_batching is, surprisingly,
> a compile-time constant.
> 
> All this misleadingness would be fixed if it were called
> HUNG_TASK_BATCHING.  But then it wouldn't be pretty.
>