Memory faults and protection from them.

[Andrew Sporner]

Hi,

A thought had occured to me a week ago. A few
years ago I was listening to a guy at HP talk
about making memory more reliable in the HP/UX
by "fencing" out bad pages. In other words
any page of memory that generated an ECC fault
would be put away somewhere and would become
unavailable.

Has this been talked about here before?

Thanks!

Andy