Re: [BUG] Softlockup detected with linux-2.6.14-rt6

[Ingo Molnar <mingo@elte.hu>]

* john cooper <john.cooper@timesys.com> wrote:

> Ingo Molnar wrote:
> >* Luca Falavigna <dktrkranz@gmail.com> wrote:
> >>I found this softlockup bug involving arts daemon using a
> >>linux-2.6.14-rt6 kernel (with "Complete Preemption" and "Detect Soft
> >>Lockups" compiled in).
> >>This bug does not happen everytime: I was able to reproduce it only
> >>three times in a week. [...]
> >
> >
> >does this happen with -rt13 too? I have fixed a softlockup 
> >false-positive in it.
> 
> Just curious what the cause of the false positive was?

the fix is below - we didnt reset the 'light' counter in the else 
branch.

	Ingo

Index: linux/kernel/softlockup.c
===================================================================
--- linux.orig/kernel/softlockup.c
+++ linux/kernel/softlockup.c
@@ -90,7 +90,8 @@ void softlockup_tick(void)
 
 		wake_up_process(per_cpu(watchdog_task, this_cpu));
 		per_cpu(timeout, this_cpu) = jiffies + msecs_to_jiffies(1000);
-	}
+	} else
+		touch_light_softlockup_watchdog();
 
 	if (per_cpu(print_timestamp, this_cpu) == timestamp)
 		return;