memory.c - bad pmd - x86_64

[Ivan Adzhubey]

On Thursday 02 June 2005 11:29 am, Peter J. Stieber wrote:
> CF = Christoph Franke
> CF>>>>> Ok :-) But could you tell us if you already based this one upon
> CF>>>>> 2.6.11.11, which contains some x86_64 fixes?
>
> DJ = Dave Jones
> DJ>>>> Yes, it does include 2.6.11.11
>
> CF>>> Great. Ok, another good news is: so far I am
> CF>>> unable to reproduce the
> CF>>> problem any longer!
> CF>>>
> CF>>> /home/christoph$ cat davej
> CF>>> ./notes/people_to_hug_before_I_die.txt
>
> IA = Ivan Adzhubey
> IA>> Just rebooted fileserver under -30 build, looks
> IA>> like running clean, 2 compute nodes also have not
> IA>> seen errors since the new kernel installed, quite
> IA>> promising I would say. I'll report if I see any errors.
>
> PS = Peter J. Stieber
> PS> Dave,
> PS>
> PS> I only see -31 in you web content. I must have
> PS> missed a version.
> PS>
> PS> Was there a fix specifically added to address the
> PS> x86_64 bad pmd issue, or is it simply masked due
> PS> to other changes?
> PS>
> PS> I'll load 2.6.11-1.31smp and report back.
>
> I have been running 2.6.11-1.31_FC3smp for over a day and loading the
> system in a manner that has caused the bad pmd problem in the past, but
> haven't seen the problem :-))

I can confirm, 3 boxes here now running 48 hours under heavy to regular load, 
no more pmd errors.

> Dave, I'm still wondering if a specific patch was added to this version
> to address the bad pmd problem?
>
> Thank for all your efforts,
> Pete

-- 
Ivan