Re: [PATCH 0/9] V2: idle page tracking / working set estimation

From: Eric B Munson
Date: Fri Sep 30 2011 - 14:19:18 EST


On Thu, 29 Sep 2011, Eric B Munson wrote:

> On Thu, 29 Sep 2011 13:25:00 -0700, Michel Lespinasse wrote:
> >On Thu, Sep 29, 2011 at 9:43 AM, Eric B Munson <emunson@xxxxxxxxx>
> >wrote:
> >>I have been trying to test these patches since yesterday
> >>afternoon.  When my
> >>machine is idle, they behave fine.  I started looking at
> >>performance to make
> >>sure they were a big regression by testing kernel builds with
> >>the scanner
> >>disabled, and then enabled (set to 120 seconds).  The scanner
> >>disabled builds
> >>work fine, but with the scanner enabled the second time I build
> >>my kernel hangs
> >>my machine every time.  Unfortunately, I do not have any more
> >>information than
> >>that for you at the moment.  My next step is to try the same
> >>tests in qemu to
> >>see if I can get more state information when the kernel hangs.
> >
> >Could you please send me your .config file ? Also, did you apply the
> >patches on top of straight v3.0 and what is your machine like ?
> >
> >Thanks,
>
>
> My .config will come separately to you. I applied the patches to
> Linus' master branch as of yesterday. My machine is a single Xeon
> 5690 with 12G of ram (do you need more details than that?)
>
> Thanks,
> Eric

I am able to recreate on a second desktop I have here (same model CPU but a
different MB so I am fairly sure it isn't dying hardware). It looks to me like
a CPU softlocks and it stalls the process active there, so most recently that
was XOrg. The machine lets me login via ssh for a few minutes, but things like
ps and cat or /proc files will start to work and give some output but hang.
I cannot call reboot, nor can I sync the fs and reboot via SysRq. My next step
is to setup a netconsole to see if anything comes out in the syslog that I
cannot see.

Eric

Attachment: signature.asc
Description: Digital signature