On Mon, Aug 12, 2002 at 10:28:28AM +1000, Nathan Scott wrote:
> On Sat, Aug 10, 2002 at 11:45:17PM -0800, Ethan Benson wrote:
> > ...
> > ideally XFS would not unecessarily create this attribute when the
> > standard permission bits are sufficient,
>
> Yep, that would be ideal. It may happen eventually, but its
> not really high on the priority list at the moment.
there is already code to ensure that the inode mode feild is in sync
with the USER, GROUP, and OTHER ACEs how difficult would it be to
delete the attribute at that point if it contains no further ACEs ?
> > this would perhaps save some
> > small ammount of disk space as well. failing that is it possible to
> > fix acl_extended_file() to notice and ignore superflurious
> > posix_acl_access attributes that are doing nothing but duplicating the
> > standard mode bits?
>
> Probably best to ask Andreas that one.
i tend to think adding more weight to acl_extended_file() is probably
the wrong approach since its really XFS which is not handling default
acls the way it should.
fwiw i just checked a SunOS box, and it does not have this problem.
--
Ethan Benson
http://www.alaska.net/~erbenson/
pgpZSvOaJLWBo.pgp
Description: PGP signature
|