Re: [PATCH clocksource 1/3] clocksource: Reject bogus watchdog clocksource measurements

From: Paul E. McKenney
Date: Mon Nov 21 2022 - 10:22:26 EST


On Mon, Nov 21, 2022 at 08:55:15AM +0800, Feng Tang wrote:
> On Thu, Nov 17, 2022 at 03:09:10PM -0800, Paul E. McKenney wrote:
> > On Thu, Nov 17, 2022 at 10:57:34PM +0100, Thomas Gleixner wrote:
> > > Paul!
> > >
> > > On Mon, Nov 14 2022 at 15:28, Paul E. McKenney wrote:
> > > >
> > > > + /* Check for bogus measurements. */
> > > > + wdi = jiffies_to_nsecs(WATCHDOG_INTERVAL);
> > > > + if (wd_nsec < (wdi >> 2)) {
> > > > + pr_warn("timekeeping watchdog on CPU%d: Watchdog clocksource '%s' advanced only %lld ns during %d-jiffy time interval, skipping watchdog check.\n", smp_processor_id(), watchdog->name, wd_nsec, WATCHDOG_INTERVAL);
> > > > + continue;
> > > > + }
> > > > + if (wd_nsec > (wdi << 2)) {
> > > > + pr_warn("timekeeping watchdog on CPU%d: Watchdog clocksource '%s' advanced an excessive %lld ns during %d-jiffy time interval, probable CPU overutilization, skipping watchdog check.\n", smp_processor_id(), watchdog->name, wd_nsec, WATCHDOG_INTERVAL);
> > > > + continue;
> > > > + }
> > >
> > > This is really getting ridiculous.
> >
> > I have absolutely no argument with this statement, and going back a
> > long time. ;-)
> >
> > But the set of systems that caused me to send this turned out to have
> > real divergence between HPET and TSC, and 40 milliseconds per second of
> > divergence at that. So not only do you hate this series, but it is also
> > the case that this series doesn't help with the problem at hand.
>
> The drift is about 4% which is quite big. It seems that this is
> either problem of HPET/TSC's hardware/firmware, or the problem of
> frequency calibration for HPET/TSC. TSC calibration is complex,
> as it could be done from different methods depending on hardware
> and firmware, could you share the kernel boot log related with
> tsc/hpet and clocksource?

Thank you for looking into this!

Knucklehead here failed to save away the dmesg. Let me see what
I can drum up.

> Also if your platform has acpi PM_TIMER, you may try "nohpet"
> to use PM_TIMER instead of HPET and check if there is also big
> drift between TSC and PM_TIMER.

Let me see about giving this a go, and again, thank you!

Thanx, Paul