Re: [PATCH 4/4] watchdog: configure nmi watchdog period based on watchdog_thresh

[Ingo Molnar <mingo@elte.hu>]

* Don Zickus <dzickus@redhat.com> wrote:

> On Tue, May 17, 2011 at 09:16:42AM +0200, Ingo Molnar wrote:
> > Hm, our tolerance for the two thresholds is not just human but technical: hard 
> > lockup warnings should indeed be triggered after just a few seconds, soft 
> > lockups can have false positives under extreme conditions.
> > 
> > So we generally want a higher threshold for soft lockups than for hard lockups.
> > 
> > So how about we couple the thresholds with a factor: we make the soft threshold 
> > twice the amount of time the hard threshold is? Then we could change the 
> > upstream default as well i think: lets change the NMI timeout to 10 seconds 
> > (and thus have the soft threshold at 20 seconds). Is 20 seconds short enough 
> > for most users to not hit reset?
> 
> Making softlockup twice as long as hardlockup seems to make sense.
> Setting the hardlockup to 10 seconds can be ok, but then you get into
> power savings issues.  For example, I have the timers setup to trigger 5
> times a period (I know it probably should be 2 times), so at 10 seconds
> that means the timers are firing every 2 seconds.  That shows up on
> powertop :-(.  Though I was flirting with the idea of trying to slow down
> or stop the timer when the cpu goes into deeper c-states.  But that is a
> different problem.
> 
> > 
> > We might want to change another aspect of the NMI watchdog: right now it tries 
> > to abort the offending task - which is really nasty if there was a spuriously 
> > long irqs-off section somewhere in the kernel. How about we just print a 
> > warning instead?
> 
> I dont understand this.  IIRC NMI watchdog will either printk or panic on
> a hardlockup.  What do you mean by 'aborting' the task?

Oh, simple dementia on my side. We used to attempt a do_exit() in some long 
gone version of that code :-)

Thanks,

	Ingo