Re: Where might memory be reported?

From: Andriy Gapon <avg_at_FreeBSD.org>
Date: Tue, 8 Jun 2021 12:26:45 +0300
On 08/06/2021 10:39, Chris wrote:
> On 2021-06-07 22:58, Dewayne Geraghty wrote:
>> This issue was raised in April '21 by Andriy Gapon title "stable/13, vm
>> page counts do not add up".
>>
>> This was mentioned
>> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=253281
>> though its probably more of an affirmation of a symptom. (reminder to
>> adjust status)
>>
>> More relevant is
>> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234559
>> where Mark Johnson advised (in Feb '19)
>> "I've thought about this some more and decided to close this bug WONTFIX
>> for now.  Basically, we started lazily dequeuing wired pages to avoid
>> the page queue synchronization cost incurred by frequent wirings.  The
>> result is that wired pages may temporarily be counted as
>> active/inactive/laundry despite not being reclaimable.  Maintaining the
>> previous behaviour wrt statistics would introduce synchronization
>> overhead that we've been trying to avoid in past and future work.  I
>> don't plan to change the current behaviour and there doesn't seem to be
>> a compelling reason to do so, at least based on this report.  If you or
>> anyone else would like to discuss this further or describe a use-case
>> that depends on the old behaviour, please feel free to re-open the bug."
> tl;dr
> I recognize this needs work. But it's a GREAT DEAL of work, and I currently
> have much shinier things on my to-do list. ;-)
> 
> No offense to Mark. He's doing a tremendous job. I just couldn't resist. :-)

Just to clarify, Mark was commenting on small discrepancies (deviations, 
fluctuations) that can happen because of lazy statistics.
The large and persistent (and growing) mismatch seems like a different problem.


-- 
Andriy Gapon
Received on Tue Jun 08 2021 - 09:26:45 UTC

Original text of this message