Comment # 22 on bug 1129428 from J. Daniel Schmidt

(In reply to J. Daniel Schmidt from comment #20)
>
> > Also check for a BIOS update.
> 
> Will check and report back soon.

There was a BIOS Update.

Details:
https://www.dell.com/support/home/de/de/debsdt1/drivers/driversdetails?driverid=m26kc&oscode=biosa&productcode=precision-t3620-workstation


After installing the BIOS Update the segfaulting and Oopsing continues as
before.
With a slight difference:
Firefox with taskset -c 3 Oopses within seconds
Firefox with taskset -c 7 can run for up to an hour just with segfaults in the
background until the system finally froze.

Maybe it was just bad luck, but it was reproducible 3 times.


(In reply to Vlastimil Babka from comment #21)
> 
> Hmm that should be two hyperthreads of a single core. It really looks like
> it might be faulty.

Yes, they are thread siblings. I now disabled these two threads to make the
system stable again:

echo 0 > /sys/devices/system/cpu/cpu3/online
echo 0 > /sys/devices/system/cpu/cpu7/online 


How can I find out if its the hardware that is faulty?
Bernhard Wiedemann told me about a bug where a spectre patch had an off by one
when counting the cpu numbers (I don't have details, Bernhard can you please
provide links or details?). This might also be a possibility, right?