[TriLUG] Sunfire x4140 issues

[Zac Durham]

I had a pair of quad opties in a build that were in a series that had a well known TLB bug that was only avoidable if the motherboard manufacturer published a BIOS specifically to work around the bug. I wonder if your CPUs carry this same bug.

Zac



On Jun 26, 2013, at 11:47 AM, Brian McCullough <bdmc at buadh-brath.com> wrote:

> Folks,
> 
> Once again, I have been fighting what appears to be a hardware issue for
> longer than I should.
> 
> I recently received a new ( old ) Sunfire x4140 for a project.  It has
> 2 four-core Opterons and 32 GB of RAM.
> 
> I think that I am seeing two issues, although they may be related.
> 
> The first, and most obvious, is "Hypertransport sync flood error,"
> which, apparently, is triggered by the hardware during the OS boot
> process.  The kernel and ramdisk get loaded, and the rest of the startup
> gets rolling, and then suddenly the machine reboots.
> 
> The second, on the odd occasion when the machine actually completes the
> boot and runs ( runs fine ), is that the "svm" flag is not set in
> /proc/cpuinfo.
> 
> I have tried setting the SVM Enable/Disable setting both ways in the
> BIOS.  No difference.
> 
> We tried removing half of the memory to try and address the
> Hypertransport issue.  No change. ( vendor suggestion )
> 
> I got the ".pkg" files for three different versions of the BIOS (
> apparently ), and tried installing them through the ILOM Service
> Processor Update process.  Only the newest seemed to make any
> difference, and that didn't seem to have any effect on the issues.
> 
> The vendor suggested that I try "re-flashing" the BIOS with the
> "bootable USB" technique.  I don't seem to be able to create a bootable
> USB stick that the machine accepts.  ( I don't know whether I am
> creating a bootable USB stick that is usable -- I think it needs to be
> DOS. )  I found a "fw....iso" file in one of the "BIOS update" packages,
> burned it to a CD and was able to get into that, but the tool in there
> seemed to only re-flash the HD controller card, not the BIOS.
> 
> If you have any bright ideas, the vendor ( in California ) and I are
> running out of things to do, or perhaps just the ability to do them.
> 
> 
> Thanks,
> Brian
> 
> -- 
> This message was sent to: Zac Durham <zacdurham at gmail.com>
> To unsubscribe, send a blank message to trilug-leave at trilug.org from that address.
> TriLUG mailing list : http://www.trilug.org/mailman/listinfo/trilug
> Unsubscribe or edit options on the web    : http://www.trilug.org/mailman/options/trilug/zacdurham%40gmail.com
> Welcome to TriLUG: http://trilug.org/welcome