Message-ID: <87wm2ha7ms.ffs@tglx>
Date: Sat, 20 Dec 2025 09:18:35 +0100
From: Thomas Gleixner <tglx@...utronix.de>
To: paulmck@...nel.org
Cc: LKML <linux-kernel@...r.kernel.org>, Daniel J Blueman
 <daniel@...ra.org>, John Stultz <jstultz@...gle.com>, Waiman Long
 <longman@...hat.com>, Peter Zijlstra <peterz@...radead.org>, Dave Hansen
 <dave.hansen@...ux.intel.com>, Tony Luck <tony.luck@...el.com>, Borislav
 Petkov <bp@...en8.de>, Stephen Boyd <sboyd@...nel.org>, Scott Hamilton
 <scott.hamilton@...den.com>
Subject: Re: clocksource: Reduce watchdog readout delay limit to prevent
 false positives

On Fri, Dec 19 2025 at 16:18, Paul E. McKenney wrote:
> On Fri, Dec 19, 2025 at 11:13:05AM +0100, Thomas Gleixner wrote:
>> On Wed, Dec 17 2025 at 16:48, Paul E. McKenney wrote:
>> > On Wed, Dec 17, 2025 at 06:21:05PM +0100, Thomas Gleixner wrote:
>> >> Assume TSC is the clocksource and HPET is the watchdog and both have a
>> >> uncertainty margin of 250us (default). The watchdog readout does:
>> >> 
>> >>   1) wdnow = read(HPET);
>> >>   2) csnow = read(TSC);
>> >>   3) wdend = read(HPET);
>> >     4) wd_end2 = read(HPET);
>> 
>> That's completely irrelevant for the problem at hand.
>
> It is exactly one of the problems at hand.  If we didn't have HPET readout
> performance issues, we could dispense with that piece of the puzzle.

It's part of the "make it work" hack, but unrelated to the fact that the
readout valid window is larger than the interval valid window.

> By the way, did this issue really occur on real hardware?  Or are we
> having a strictly theoretical discussion?

It has been reported by Daniel and it happens on real hardware.

>> >> The valid window for the delta between #1 and #3 is calculated by the
>> >> uncertainty margins of the watchdog and the clocksource:
>> >> 
>> >>    m = 2 * watchdog.uncertainty_margin + cs.uncertainty margin;
>> >> 
>> >> which results in 750us for the TSC/HPET case.
>> >
>> > Yes, because this interval includes two watchdog reads (#1 and #3 above)
>> > and one clocksource read (#2 above).  We therefore need to allow two
>> > watchdog uncertainties and one clocksource uncertainty.
>> 
>> That's a made up and broken theory based on ill defined heuristics.
>
> If we want better heuristics, we either restrict ourselves to timer
> hardware that has reasonable access latencies or we add something like
> an ->access_time field to struct clocksource.
>
> Otherwise, you will just be cutting off one end of the heuristic "blanket"
> and sewing on to another.
>
> Accounting more directly for vCPU preemption would also help a lot, as
> that was the root cause of much (but by no means all) of the problem on
> Meta's fleet.

The goal is to get rid of the heuristics completely.

>> The uncertainty of a clocksource is defined by:
>> 
>>     1) The frequency it runs with, which affects the accuracy of the
>>        time readout.
>> 
>>     2) The access time
>> 
>> Your implementation defines that the uncertainty is at least 50us,
>> with a default of 125us and an upper limit of 1ms. That's just made up
>> numbers pulled of of thin air which have nothing to do with reality.
>
> These were set using test data from real hardware.
>
>> Declaring that TSC access time of up to 1ms and at least 50us is
>> hillarious.
>
> Yes, in theory we could instead have an ->access_time that is used to
> reject bad reads, so that ->uncertainty_margin would only be used to
> detect timer skew.  The discussion of such an addition didn't go far last
> time around, but maybe now is a better time.  Though I suspect that the
> issues around determining what ->access_time should be set to are still
> with us.  But I would love to be proven wrong.
>
>> Adding up these made up margins and then double the watchdog margin to
>> validate the readout is just a hack to make it "work for me" and thereby
>> breaking the whole machinery for existing problematic systems. See below.
>
> Nope.
>
> The readout #4 above into wd_end2 wasn't something my employer needed.
> For our use cases, #1, #2, and #3 sufficed.  It was instead needed by
> other users running memory-bandwidth-heavy workloads on multi-socket
> systems.  This resulted in HPET access times that were astoundingly large.
>
> One might naively hope that the HPETs in a given system could synchronize
> themselves so that each CPU could read the HPET nearest it instead of
> everyone reading CPU 0's HPET.  Or that some other means would allow
> reasonable access times.  Hey, I can dream, can't I?

Latest at that point it should have been clear that heuristics aren't
going to cut it.

>> >> The actual interval comparison uses a smaller margin:
>> >> 
>> >>    m = watchdog.uncertainty_margin + cs.uncertainty margin;
>> >> 
>> >> which results in 500us for the TSC/HPET case.
>> >
>> > This is the (wd_seq_delay > md) comparison, righr?  If so, the reason
>> > for this is because it is measuring only a pair of watchdog reads (#3
>> > and #4).  There is no clocksource read on the latency recheck, so we do
>> > not include the cs->uncertainty_margin value, only the pair of watchdog
>> > uncertainty values.
>> 
>> No. This is the check which does:
>> 
>> 	int64_t md = 2 * watchdog->uncertainty_margin;
>>         ...
>> 
>>                 *wdnow = watchdog->read(watchdog);
>> 		*csnow = cs->read(cs);
>> 		wd_end = watchdog->read(watchdog);
>>                 ...
>>                 
>> 		wd_delay = cycles_to_nsec_safe(watchdog, *wdnow, wd_end);
>> 		if (wd_delay <= md + cs->uncertainty_margin) {
>>                           ...
>>                           return WR_READ_SUCCESS;
>>                 }
>> 
>> It has nothing to do with the wd_seq_delay check.
>
> Eh?  Here wd_delay is twice watchdog uncertainty plus clocksource
> uncertainty.

Correct. As implemented by Paul McKenney, no?

>>                                              Let me draw you a picture:
>> 
>>     H  = HPET read
>>     T  = TSC read
>>     RW = read valid window
>>     IW = interval valid window
>> 
>>     RW-------------------------------------------RW
>>     IW-------------------------IW
>> 
>>     HT                                       H          <- Read 1 valid
>>     H                                  TH               <- Read 2 valid
>>      |---------------------------------|                <- Interval too large
>> 
>> Q: How is increasing the uncertainty values fixing the underlying math
>>    problem of RW > IW?
>> 
>> A: Not at all. It just papers over it and makes the false positive case
>>    more unlikely by further weakening the accuracy.
>
> Yep.  Given the current lack of something like ->access_time to go along
> with the current ->uncertainty_margin, there are limits to what we can do.

Well, we can at least limit the damage by making RW and IW consistent, no?

> Again, was the report from a real workload running on real hardware, or
> are we simply having a theoretical discussion of what might happen?

It's real hardware and again even in theory it's entirely clear that
having a readout validation window which is larger than the interval
validation window leads exactly to the problem reported by
Daniel. No?

> OK, that is at least a real problem, even if on antique hardware.
> A problem that surprises me, but I don't have that sort of hardware,
> so I will take your word for it.  Though you would need what, a 62.5
> millisecond SMI for the old watchdog to catch it, right?
>
> (Yes, yes, I have seen far longer SMIs.  Don't get me started...)

It's a SMI plagued horrorshow where BIOS writers did value add!

>> We really need to go back to the drawing board and rethink this
>> machinery from scratch.
>> 
>> There are two main aspects to the watchdog:
>> 
>>   1) Detect general frequency skew by comparing it against a known
>>      (assumed to be) stable clocksource
>> 
>>      This addresses the problems on older machines where the BIOS
>>      fiddles with the CPU frequency behind the kernels back.
>
> Fair enough.
>
>>      That's a non-issue on modern CPUs because the TSC is constant
>>      frequency.
>
> Give or take the occasional messed-up motherboard or firmware, where
> the frequency is constant---but wrong.  :-/

We still can enforce the frequency check even for constant frequency
systems.

>>   2) Detect inter CPU skew
>> 
>>      This addresses the problems where
>> 
>>      A) the BIOS fiddles with the TSC behind the kernels back (see
>>         above) and the modification cannot be detected by the TSC_ADJUST
>>         MSR due to its non-existance
>> 
>>         Not an issue on modern CPUs
>> 
>>      B) the sockets are not synchronized and the TSC frequency on them
>>         drifts apart
>> 
>>      C) hardware failure (something you mentioned back then when you
>>         hacked this uncertainty magic up)
>
> 3) Deal with insanely slow readout times for some clocksources.
>    Which your smp_call_function() approach describe below should handle.

That's an implementation detail, but not part of what the watchdog
should actually detect, i.e. frequency change and inter CPU skew.

>> As I just validated on that old machine the whole "scalability" hackery
>> broke #2A and #2B unless the modifications or drifts are massive. But if
>> they are not it leaves both user space and kernel with inconsistent
>> time.
>
> Agreed.  Imperfect though the current setup might be, there were reasons
> why we changed it.
>
>> That means we really have to look at this in two ways:
>> 
>>   I)  On old hardware where TSC is not constant frequency, the validation
>>       against HPET is required.
>> 
>>   II) On new hardware with constant frequency TSC and TSC_ADJUST_MSR the
>>       HPET comparison is not really useful anymore as it does not detect
>>       low drift issues between unsynchronized sockets.
>> 
>> What can be done instead?
>> 
>>   - Make CPU0 the watchdog supervisor, which runs the timer and
>>     orchestrates the machinery.
>
> Fine on x86, but last I knew some systems can run without CPU 0.
> Maybe the boot CPU?  Though that can be offlined on some systems.
> Handoff?  For NO_HZ_FULL people, confine to the housekeeping CPUs?

Sure, that's a detail where you assign it to.

> Or is x86 the only system with more than one timer?

The watchdog is only enabled on x86 and MIPS, but none of the MIPS
clocksources sets the MUST_VERIFY flag. PARISC sets the MUST_VERIFY
flag, but does not enable the watchdog. So it's effectively x86.

It seems that everything else either has a sane timer or at least
believes so.

>>   - Limit the HPET comparison to CPUs which lack constant frequency,
>>     which makes the whole legacy I/O issue on large systems go away.
>
> I thought that was already the default, and that you need to use the
> tsc=watchdog kernel boot parameter to override that default in order to
> use a known-stable TSC as the watchdog for HPET and/or ACPI PM timer.

Correct.

>>     Run this comparision only on CPU0
>> 
>>   - Instead of moving the timer around CPUs, utilize a SMP function call
>>     similar to what the TSC synchronization mechanism does, i.e.
>> 
>>     CPU0         CPUN
>>     sync()       sync()
>>     t1 = rdtsc()    
>>     handover()
>>                  t2 = rdtsc()
>>                  handover()
>>     t3 = rdtsc()
>>     ....
>> 
>>     Each step validates that tN <= TN+1, which detects all issues of #2
>>     above for both old and contemporary machines.
>> 
>>     Of course this mechanism is also affected by interconnect
>>     congestion, but the only side effect of interconnect congestion is
>>     that the handover becomes slow, which means that the accuracy of
>>     detecting time going backwards suffers temporarily.
>
> Is this going to be OK for real-time workloads?  I guess that you have at
> least as good a chance as I would to convince them.  The more aggressive
> NO_HZ_FULL people might have concerns.  Especially these guys:
>
> https://arxiv.org/abs/2509.03855

I'm sure that the wishful thinking departement which ignores the system
immanent latencies and demonstrates great results on a system which is
tuned to run only the demonstrator will go apeshit.

> I have similar code in the current clocksource watchdog, but it doesn't
> run until clocksource skew is detected, in other words, only after
> something has likely gone sideways.

Of course we can exclude the nohz full CPUs, but that's again an
implementation detail and does not change the underlying design.

>> With that all these made up uncertainty heuristics go completely away or
>> can be turned into something which actually reflects the reality of the
>> universe, i.e. the accuracy of the clocks and their access time.
>> 
>> I doubt it's needed because the original implementation worked just fine
>> and it's only relevant for the actual HPET/TSC comparison case. That is
>> then limited to museum pieces which are not really affected by the
>> scalability issues of todays machines. Especially not as the HPET/TSC
>> check is limited to CPU0, which is the one closest to the legacy
>> peripherals which contain the HPET or in real old machines the ACPI_PM
>> timer.
>
> Give or take firmware/hardware bugs that cause the TSC to at a constant
> but incorrect rate.  Maybe interact with userspace facilities such as
> NTP?

Either we enforce HPET/ACPI_PM comparison on the supervisor CPU or let
user space deal with it. If NTP detects time going sideways, but being
consistent between CPUs, then it can switch the clocksource to HPET or
ACPI_PM already today.

>> I'll go and utilize my copious spare time to implement this, but don't
>> expect it to materialize before 2026 :)
>
> The current implementation is working for us, so your schedule is our
> schedule.  ;-)

Thanks for confirming that this is based on "works for me" :)

Thanks,

        tglx

Powered by blists - more mailing lists