Thanks for the info,
I did some digging into Curtis's references and found all sorts of stuff:
This web site has the info
Including kernel patches which do system calls, vfs extensions etc, they do
not have quite all the interfaces with which to squeeze the xfs API through,
specifically flags for link following etc.
We should probably have a meeting of the minds on this one.
> Andi Kleen wrote:
> > On Thu, Jul 27, 2000 at 02:19:25PM -0500, The Unknown User-ID wrote:
> > > I agree - except there is no 'approved' interface for extended attributes
> > > yet. This is as much a way for us to exercise the code in the kernel and
> > > proof of concept as anything else. If the final version of an extended
> > > attribute interface is different then we will change to match it.
> > The problem is not really the interface, just the syscall slot number.
> > I doubt it will be a problem to get a sysxfs() reserved, you could
> > hook the interface onto that and still switch over to the official interfac
> > later.
> There may be no 'approved' interface for extended attributes, but there is
> another group of developers who have (roughly) adopted the XFS Extended
> Attribute interface. Consensus may be forming on an API.
> Andreas Gruenbacher (a.gruenbacher@xxxxxxxxxxx) has been working on using
> Extended Attributes to support ACLs in Linux and has based his work on
> the XFS EA API. James Buster (bitbug@xxxxxxx) has participated as well.
> Curtis Anderson - Storage Group Leader canderson@xxxxxxxxxxxxx