Re: [Contiki-developers] MSP430 watchdog mode (was: Re: MSP430 watchdog_start()/watchdog_stop() in TimerA vectors)

[Peter B. <bi...@ac...>]

+1 : A watchdog that's neutered to the point it doesn't do its job is a
waste of code space and a loss of a valuable MCU feature.

Whether it's possible for a generic platform like Contiki to support a
watchdog policy other than "turn it off on start and leave the application
to decide how it should be used" is another question.  Promising more than
that requires a policy that all the infrastructure routines invoke a reset
operation at a certain interval during periods when user code isn't running
(e.g., in the 6lowpan stack), or a maximum runtime guarantee on entrypoints
to infrastructure code.

Peter

On Thu, Sep 29, 2011 at 3:05 AM, Oskar Nordquist <
osk...@cr...> wrote:

> Hello,
>
> While what you are saying is true, the watchdog will not work in ISR:s
> for MSP430 platforms. The reason for this is that the watchdog is
> configured in timer interval mode which generates interrupts. This means
> watchdog will not reset CPU when hanged in an interrupt vector unless
> nested interrupts are enabled.
>
> >From what I can see, the timer interval mode is only used to print a
> stack trace on the TMote Sky. We propose that the watchdog should be in
> watchdog mode by default.
>
> Best regards,
> Oskar
>
> On Mon, 2011-09-26 at 14:15 +0200, Joakim Eriksson wrote:
> > Hello,
> >
> > This is handled by the counter in watchdog.c. So the call to
> > start does not actually start it if it already is started and
> > vice versa (with stop).
> >
> > Best regards,
> > -- Joakim Eriksson, SICS
> >
> >
> > Oskar Nordquist skrev 2011-09-26 11:20:
> > > Hi,
> > > This question is really directed directly to Adam, so excuse me for
> > > CC:ing you directly. I just think it is important.
> > >
> > > In cpu/msp430/{clock,rtimer_arch}.c, the TimerA vectors now starts and
> > > stops the watchdog. I understand from Sky and Z1 platforms that this is
> > > because you want the watchdog to be enabled during TimerA interrupts
> > > during LPM (as watchdog is disabled in LPM).
> > >
> > > My question is: What happens when TimerA interrupts occurs when *not* i
> > > LPM? Would this not result in the watchdog being inactive upon return?
> > >
> > > Best regards,
> > > Oskar
> > >
> > >
> > >
> ------------------------------------------------------------------------------
> > > All the data continuously generated in your IT infrastructure contains
> a
> > > definitive record of customers, application performance, security
> > > threats, fraudulent activity and more. Splunk takes this data and makes
> > > sense of it. Business sense. IT sense. Common sense.
> > > http://p.sf.net/sfu/splunk-d2dcopy1
> > > _______________________________________________
> > > Contiki-developers mailing list
> > > Con...@li...
> > > https://lists.sourceforge.net/lists/listinfo/contiki-developers
> >
> >
> >
> ------------------------------------------------------------------------------
> > All the data continuously generated in your IT infrastructure contains a
> > definitive record of customers, application performance, security
> > threats, fraudulent activity and more. Splunk takes this data and makes
> > sense of it. Business sense. IT sense. Common sense.
> > http://p.sf.net/sfu/splunk-d2dcopy1
> > _______________________________________________
> > Contiki-developers mailing list
> > Con...@li...
> > https://lists.sourceforge.net/lists/listinfo/contiki-developers
> >
>
>
>
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> Contiki-developers mailing list
> Con...@li...
> https://lists.sourceforge.net/lists/listinfo/contiki-developers
>