[Openchrome-users] Fw: serious problem: possible memory corruption

[Robert P. Thille]

On Apr 7, 2008, at 10:46 AM, Hindrik Hettema wrote:
> This looks more like the "VIA board deadlock" problem analyzed
> thoroughly amongst others at the VIA-ARENA forum. DMA-wise something
> seems wrong with these VIA-boards (bot J7F2 and J7F4). Some people  
> blame
> the temperature, but not so.
>
> I myself had one "hang" with not even X-windows running, but that is
> very rare. Many interactions on X increase the chance.
>
> I gave up, as no remedy except disabling totally DMA seemed to help  
> (and
> thinking about changing the computer already for a very long time  
> now).

It definitely sees to be a kernel bug.  I tried every Fedora since 5,  
several Ubuntu versions, and one or two Suse.  All were unstable on  
my EN15000G w/1GB RAM.  Memtest 86+ would run for days reporting no  
errors and NetBSD-Current was completely stable.

What I did to kill the machine was boot without X and run my 'hammer'  
script which would dd some data from /dev/urandom, make a bunch of  
copies of it, and then MD5 the copies.  Pushing a lot of data thru  
the disks and IO subsystem was enough to lock the system hard within  
about 30-60 minutes.  NetBSD could handle the test just fine.

Then I got Fedora Beta 9 earlier this week and my system seems stable  
with that.  I ran several of my 'Hammer' scripts in parallel and one  
in a loop for about 36 hours.

So, I'm not sure what the bug was, or when it was fixed, but it seems  
that the current Fedora 9 kernel is stable for me.

Robert

--
HELP ME FIGHT CANCER!           DONATE HERE: http://tinyurl.com/ynq7uk
--
Robert Thille                7575 Meadowlark Dr.; Sebastopol, CA 95472
Home: 707.824.9753    Office/VOIP: 707.780.1560     Cell: 707.217.7544
Robert.Thille at rangat.org   YIM:rthille   http://www.rangat.org/rthille