Re: [patch] Real-Time Preemption, -VP-2.6.9-rc4-mm1-U3

[Ingo Molnar]

* Rui Nuno Capela <rncbc@xxxxxxxxx> wrote:

> Ingo Molnar wrote:
> >
> > it seems that SMP + PREEMPT_TIMING is not stable though, somehow the
> > latency printk's cause a crash sooner or later. I'm still debugging this
> > problem. Without PREEMPT_TIMING the SMP kernel is stable.
> >
> 
> How true!
> 
> My first successful SMP/HT PREEMPT_REALTIME has been achieved, by just
> turning off PREEMPT_TIMING. So you won't get any latency trace dumps
> from here ;)

meanwhile i've mostly debugged the problem: it's an illegal recursion
into the timer code caused by PREEMPT_TIMING printks from within the
timer code calling do_poke_blanked_console() which in turn calls
del_timer() ...

this bug has been in the PREEMPT_TIMING code for almost forever, it's
just that under PREEMPT_REALTIME all the other latency sources are gone,
only the few places that still use raw spinlocks are remaining, amongst
them the timer code ...

> Actual .config.gz attached.
> 
> But now I have my two prime machines on full-RT-throttle, yeepee! :)

great! :) I strongly suspect this bug is the one Lee and Mark are seeing
too.

	Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/