POSIX ACLs are not even compiled into the kernel.
--Matt
> -----Original Message-----
> From: Timothy Shimmin [mailto:tes@xxxxxxxxxxxxxxxxxxxxxxx]
> Sent: Tuesday, January 08, 2002 9:45 PM
> To: Tad Dolphay
> Cc: "ZINKEVICIUS,MATT (HP-Loveland,ex1)"; 'Steve Lord';
> linux-xfs@xxxxxxxxxxx
> Subject: Re: link() messes up file attributes
>
>
> On Tue, Jan 08, 2002 at 08:43:56PM -0600, Tad Dolphay wrote:
> > >
> > > Okay here is the current situation...
> > >
> > > Without your patch:
> > > over NFSv2: open new file with 666 permission creates
> file with 644
> > > permissions
> > > over NFSv3: open new file works fine, but hard-linking
> changes file perm to
> > > 644
> > >
> > > With your path:
> > > over NFSv2: (same as above)
> > > over NFSv3: Works fine
> > >
> > > So you patch does have some affect.
> > >
> > I vaguely remember a similar problem reported before. If I
> remember correctly,
> > it had something to do with the default acls on the
> exported directory. Could
> > you do a "chacl -l ." on the exported directory to check
> the default acls
> > and then possibly do "chacl -d u::rwx,g::rwx,o::rwx ." to
> set the default
> > acls. Then see if the problem goes away.
> >
> Yes, I was wondering the same thing.
> This is filed in pv#843903 (sgi bug#).
> The umask of nfsd is being applied when it shouldn't be
> (if we find we don't have a default ACL then we apply the umask).
> Giving a default mininum ACL with all permissions on as
> Tad suggests (-d u::rwx,g::rwx,o::rwx)
> would stop the umask from being applied.
> (This, of course, is not a fix, but would be good to check :)
>
> Are ACLs enabled ?
>
>
> (The fix for pv#843903 was being deferred 'til we had more complete
> integration with Andreas G's kernel EA/ACL patch)
>
> --Tim
>
>
> > Tad
> > > --Matt
> > >
> > > > -----Original Message-----
> > > > From: Steve Lord [mailto:lord@xxxxxxx]
> > > > Sent: Tuesday, January 08, 2002 2:57 PM
> > > > To: ZINKEVICIUS,MATT " "(HP-Loveland,ex1)
> > > > Cc: linux-xfs@xxxxxxxxxxx
> > > > Subject: RE: link() messes up file attributes
> > > >
> > > >
> > > > On Tue, 2002-01-08 at 15:49, ZINKEVICIUS,MATT
> (HP-Loveland,ex1) wrote:
> > > > > Nevermind, the fix didn't help. Occured again today :-(
> > > >
> > > > Right, I am not totally surprised, the difference
> between the two
> > > > code paths should be a noop. This was just the only
> recent change
> > > > related to xfs.
> > > >
> > > > I suppose you have HP clients on the other end of the wire here?
> > > >
> > > > Since this is an intermittent problem, how can you be
> totally sure
> > > > it is actually XFS which is at fault?
> > > >
> > > > Steve
> > > >
> > > > --
> > > >
> > > > Steve Lord voice:
> +1-651-683-3511
> > > > Principal Engineer, Filesystem Software email:
> lord@xxxxxxx
> > > >
> > >
> >
>
|