It's not NFS-related, it happens on a local disk too. It's "ar" giving a
NULL pointer as a name.
>The address listed for the exception is in getname, which puzzles me as
>getname seems to go to some length to ensure valid addresses.
Well, getname() still only makes sure that it's a user-space address, it
doesn't verify that it is a _valid_ user space address. The exception
mechanism does that part.
>Is this a known problem, or has anyone looked into the cause? It doesn't
>seem to do any harm, but it seems that either ar or the kernel might
>need a small patch-up.
It definitely looks like "ar" - it's been doing this for quite a while.
Linus