[cfarm-users] AMD Ryzen issues was: Upgrade status and status on cfarm offline

[Laurent GUERBY]

On Sat, 2017-08-19 at 13:24 +0200, Torbjörn Granlund via cfarm-users
wrote:
>   It is a CPU bug. The only solution is to RMA the CPU. Just contact
> AMD.
>   They are aware of the issue. All Ryzen CPUs are potentially
> affected.
>   See: https://community.amd.com/thread/215773
> 
> There is a Ryzen CPU bugs which causes random segfaults, no doubt.
> 
> But not all segfaults on Ryzens are due to that bug.  Firmware
> fixable
> DRAM incompatibilities and other firmware fixable Ryzen bugs can also
> cause segfaults.
> 
> Therefore, upgrading the BIOS before RMAing is a quite good
> idea.  After
> all, you might already have a Ryzen with the segfault fix.

Hi,

We plan to upgrade the BIOS of gcc67/68 this monday: we're on P2.60
06/07/2017 and will flash P3.00	7/18/2017 that comes with a
newer AGESA version (Asrock AB350 Gaming K4).

If we have time monday we'll also look at the processors packaging to
get the batch number in case we need to contact AMD.

Our Ryzen notes are here including the crash information we got from
netconsole:

https://pad.tetaneutral.net/p/gcc67

Baptiste upgraded the kernel to 4.12 yesterday on gcc67.

We haven't found out wether there's a watchdog or not on these
platforms (/dev/watchdog isn't present).

Sincerely,

Laurent