Re: [PATCH v3 2/2] x86/tsc: skip tsc watchdog checking for qualified platforms

[Paul E. McKenney]

On Tue, Nov 30, 2021 at 09:39:32PM +0100, Thomas Gleixner wrote:
> 
> Can you folks please trim your replies? Finding content in the middle of
> quoted nonsense becomes harder with every mail in this thread.
> 
> On Tue, Nov 30 2021 at 08:28, Paul E. McKenney wrote:
> > On Tue, Nov 30, 2021 at 11:02:56PM +0800, Feng Tang wrote:
> >> For this case, I don't have access to the HW and only have the
> >> dmesg log, from which it seems the watchdog timer has been postponed
> >> a very long time from running.
> >
> > Thank you for the analysis!
> >
> > One approach to handle this situation would be to avoid checking for
> > clock skew if the time since the last watchdog read was more than (say)
> > twice the desired watchdog spacing.  This does leave open the question of
> > exactly which clocksource to use to measure the time between successive
> > clocksource reads.  My thought is to check this only once upon entry to
> > the handler and to use the designated-good clocksource.
> >
> > Does that make sense, or would something else work better?
> 
> Seriously. Jiffies is not usable as watchdog simply because lost ticks
> cannot be compensated and you cannot use TSC to bridge them because you
> are not trusting TSC. This is simply a circulus vitiosus.

OK, HPET or nothing, then.

> We really need to remove the watchdog requirement for modern hardware.
> Let me stare at those patches and get them merged.

You are more trusting of modern hardware than I am, but for all I know,
maybe rightfully so.  ;-)

							Thanx, Paul