a take at cache coherency

Simon 'corecode' Schubert corecode at fs.ei.tum.de
Tue Jan 24 13:10:14 PST 2006

Csaba Henk wrote:
We need to linearize the cache entries anyways now, as we can't 
maintain a tree structure with doubly linked lists.
We don't need to linearize them. We can use the same tree layout as in
the fs hierarchy: pointers in one direction, list of neighbours in the
other direction.
okay, I see now.  This is a possibility as well.  So we'd have

ncp_shadowed and SLIST_HEAD(ncp_shadows) and thus also ncp_shadowlink

So, from this on I'll assume the latter: having the flat thing as an
actually I forgot about the need to have direct shadow associations and 
thought about a replacement.  You are right, we can't do this.

I'd still use a ncp_shadowroot to get O(1) locking.  And if you are 
concerned about data replication/keeping it in sync, we could even 
access most data (which is only used internally in the cache layer) 
directly from the shadowroot (== group head).

More information about the Kernel mailing list