I know this is a couple of months old now, but I'm not seeing anybody address this yet.
I'm definitely interested in in a resolve for this as well. I've installed CentOS 6_2 Linux Kernel 2.6.32-220.7.1el6.x86_64 on a SUPERMICRO X7DVL-E and get the same message, but no halt yet.
Prior i had installed CentOS 5_7 Linux Kernel: 2.6.18-274.el5.x86_64.
The resolve was to add:
to the Kernel line in grub
This no longer has any affect in the new kernel.
This has to do with the memory hole setting in the bios. Its not something that can be disabled in bios you can only change value. It is a mtrr missmatch compensation for 32 bit OS with 4 gig or more of memory installed from what I understand.
I think this what is causing message in var/log/messages
mcelog: failed to prefill DIMM database from DMI data
At the moment, it does not seem to have any detrimental affect, but I would hate to see this become a problem down the road.
Hi. I just ran into this problem yesterday with a HP ProLiant server.
The solution for CentOS 6 is to append "erst_disable" to the kernel command line.
This will generate a "ERST: Error Record Serialization Table (ERST) support is disabled." in dmesg but, theoretically, should fix any freezing problem.