Re: [PATCH 7/7] sched: Use WARN_ONCE for the might_sleep() TASK_RUNNING test

[Peter Zijlstra <peterz@infradead.org>]

On Fri, Oct 31, 2014 at 11:42:37PM +0100, Oleg Nesterov wrote:
> On 10/31, Peter Zijlstra wrote:
> >
> > In some cases this can trigger a true flood of output.
> >
> > Requested-by: Ingo Molnar <mingo@kernel.org>
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> > ---
> >  kernel/sched/core.c |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -7301,7 +7301,7 @@ void __might_sleep(const char *file, int
> >  	 * since we will exit with TASK_RUNNING make sure we enter with it,
> >  	 * otherwise we will destroy state.
> >  	 */
> > -	if (WARN(current->state != TASK_RUNNING,
> > +	if (WARN_ONCE(current->state != TASK_RUNNING,
> 
> Agreed, but sorry for off-topic, can't resist.
> 
> Sometimes I hate WARN_ONCE() because you can't reproduce the problem
> once again without reboot.

Yes, and the fact that you can only see the first fail, even if more are
present.

> Perhaps WARN_ON_RATELIMIT() should be used more often (not sure about
> this particular case). Or, perhaps, we can add a special section for
> these "__warned" variables and add, say, sysctl which clears that
> section ?

Yeah, maybe, /debug/warn_once/file/line/enable or whatnot. For now I'll
continue removing ONCEs whenever I feel like it though ;-)