A fault.memory.dimm is declared in two different cases:
(1) A pattern of CEs from a DIMM has been detected that indicates that the
DIMM is likely to become the cause of a UE. The DIMM should be replaced
as soon as possible.
(2) The number of CEs from that DIMM has caused the page retirement limit
to be reached. There is no danger of a latent UE. The DIMM should be
replaced at the customer's convenience.
The messages and KAs related to these two cases need to be different.
However, use of the same fault event means the same message and KA
is used for the two cases. A unique event needs to be used for
each case.
More in comments.