Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 29 Sep 2010 10:35:13 -0700
From:      Jeremy Chadwick <freebsd@jdc.parodius.com>
To:        Chuck Swiger <cswiger@mac.com>
Cc:        freebsd-stable@freebsd.org, Vitaly Magerya <vmagerya@gmail.com>
Subject:   Re: resume slow on Thinkpad T42 FreeBSD 8-STABLE
Message-ID:  <20100929173513.GA95222@icarus.home.lan>
In-Reply-To: <2B9D8374-AA0A-4F2C-9681-5216204859F8@mac.com>
References:  <20100224165203.GA10423@zod.isi.edu> <20100927170317.I90633@sola.nimnet.asn.au> <4CA0E892.4010204@gmail.com> <201009271621.17669.jkim@FreeBSD.org> <4CA2488D.7000101@gmail.com> <04FA16F2-26AD-425D-9E4A-2A923219B73E@mac.com> <4CA35E64.1040101@gmail.com> <0FDB4144-8BE4-4BA5-B911-8652E07D60C2@mac.com> <20100929170757.GA94672@icarus.home.lan> <2B9D8374-AA0A-4F2C-9681-5216204859F8@mac.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Sep 29, 2010 at 10:16:13AM -0700, Chuck Swiger wrote:
> On Sep 29, 2010, at 10:07 AM, Jeremy Chadwick wrote:
> > On Wed, Sep 29, 2010 at 09:57:53AM -0700, Chuck Swiger wrote:
> >> 
> >> I doubt repeated coincidences.  :-)  Is prime95 testing running stable after waking from sleep?
> > 
> > He's not running Prime95 (native Win32 app), he's running
> > ports/math/mprime under FreeBSD natively.  I don't know if this
> > application stresses hardware to the same degree Prime95 does; I've used
> > Prime95 many times to burn in new workstations.
> 
> It's doing the same math operations; something like "mprime -t" is the same as the Win32 test mode per the docs:
> 
>      -t              Run the torture test.  Same as Options/Torture Test.
> 
> > The Thinkpad hardware he's on is """old""" (note the quotes), so I
> > wouldn't be surprised if the CPU (Intel Pentium M) happens to induce a
> > strange/odd MCA event as a result of going in/out of sleep state.  It
> > could be a general system bug of some sort as well (one which has no
> > repercussions).
> 
> That sounds reasonable to me, but I'm wary of uncorrected errors which seem to be reproducible to specific circumstances.
> 
> > Look at it this way: if his L1 cache was going bad, his system would be
> > freaking out doing literally anything (booting the kernel for example);
> > I'm under the impression Pentium M CPUs do not have ECC L1 cache.
> 
> Sure, if the MCA report is reflecting a legitimate problem, and it was happening more often than every few minutes, and it happened after a cold reboot rather than after wakeup from sleep....  :-)
> 
> I place more faith in ~17 hours of Prime95/mprime working OK to validate that the hardware is not obviously broken.

Oh, absolutely.  If anything my statement was indirectly agreeing with
your recommended test (sans being unsure how mprime behaved).  :-)

I wonder if there's CPU errata or something along those lines which
might explain the behaviour.

-- 
| Jeremy Chadwick                                   jdc@parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.              PGP: 4BD6C0CB |




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100929173513.GA95222>