From: Jan Kara Date: Sat, 17 Oct 2009 11:44:22 +0200 Subject: [Ocfs2-devel] [PATCH 1/3] ocfs2: Make ACLs always compiled into the kernel In-Reply-To: <20091015210731.GF3112@mail.oracle.com> References: <1255611245-27340-1-git-send-email-jack@suse.cz> <1255611245-27340-2-git-send-email-jack@suse.cz> <20091015130344.GA5677@lst.de> <20091015134200.GD5971@duck.suse.cz> <20091015210731.GF3112@mail.oracle.com> Message-ID: <20091017094422.GA31412@duck.suse.cz> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ocfs2-devel@oss.oracle.com On Thu 15-10-09 14:07:31, Joel Becker wrote: > On Thu, Oct 15, 2009 at 03:42:00PM +0200, Jan Kara wrote: > > On Thu 15-10-09 15:03:44, Christoph Hellwig wrote: > > > On Thu, Oct 15, 2009 at 02:54:03PM +0200, Jan Kara wrote: > > > > -#ifdef CONFIG_OCFS2_FS_POSIX_ACL > > > > if (opts & OCFS2_MOUNT_POSIX_ACL) > > > > seq_printf(s, ",acl"); > > > > else > > > > seq_printf(s, ",noacl"); > > > > -#endif > > > > > > It might be a good idea to always print acl here for backwards > > > compatiblity. > > I don't understand here - why would printing 'acl' even if acls are > > disabled be more backward compatible? > > I think he's saying "if you don't provide any acl option at all, > they will default to enabled but you will not print 'acl'. We will print 'acl' - what the code does is that ocfs2_check_set_options() sets OCFS2_MOUNT_POSIX_ACL in case 'noacl' hasn't been specified. > I think the correct solution is to start mount_op with > OCFS2_MOUNT_POSIX_ACL set. This correctly describes the behavior. Then > 'case Opt_acl' only does the check for xattrs and 'case Opt_noacl' does > the clearing of the option. This won't quite work because at the time parse_options is called during mount, we don't have filesystem features available yet. That's why I go through the hoops described above. Honza -- Jan Kara SUSE Labs, CR