[Ocfs2-devel] [PATCH] ocfs2: Prevent mount with "-o acl" if acl isn't supported.

[Joel Becker]

On Thu, Jun 11, 2009 at 08:43:49AM -0700, Joel Becker wrote:
> On Thu, Jun 11, 2009 at 05:03:55AM +0800, Tao Ma wrote:
> > Currently, if we don't enable CONFIG_OCFS2_FS_POSIX_ACL, ocfs2
> > will not support acl. That is OK.
> > 
> > But in the following case:
> > 1. kernel isn't built to support acl.
> > 2. mount -t ocfs2 -o acl /dev/sdx /mnt/ocfs2.
> > It sucesses. And what's more, if I do "mount|grep sdx" will get
> > /dev/sdx on /mnt/ocfs2 type ocfs2 (rw,acl,heartbeat=none)
> > 
> > So a normal user will just think acl now is ok for the volume and
> > he will surely be puzzled when he tries setfacl and fails with
> > "Operation not supported".
> > 
> > The good thing is that the kernel already printk some useful
> > information, so this patch just tries to prevent the user from
> > mounting the volume.
> 
> 	Hmm, everyone else seems to do it the way we have been.  Trying
> to think about the surprise you describe vs the surprise of ocfs2 being
> different than other linux filesystms.

	Ok, I bounced this off of various fs developers, and there was
significant support for leaving it as-is.  Basically, it would suck for
every filesystem in fstab to fail to mount just because you changed your
kernel.
	Chris and Cristoph pointed out that btrfs and xfs always assume
'-o acl,user_xattr' if the feature is compiled in.  They are always on.
btrfs only disables acls with '-o noacl'.  Might be worth considering.

Joel

-- 

"I am working for the time when unqualified blacks, browns, and
 women join the unqualified men in running our overnment."
	- Sissy Farenthold

Joel Becker
Principal Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127