xfs
[Top] [All Lists]

Re: [patch] security. namespace

To: Nathan Scott <nathans@xxxxxxx>
Subject: Re: [patch] security. namespace
From: Chris PeBenito <pebenito@xxxxxxxxxx>
Date: Wed, 03 Dec 2003 00:20:51 -0600
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <20031203052510.GE1302@frodo>
References: <1070301662.7842.11.camel@xxxxxxxxxxxxxxxxxx> <20031203052510.GE1302@frodo>
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Tue, 2003-12-02 at 23:25, Nathan Scott wrote:
> On Mon, Dec 01, 2003 at 12:01:02PM -0600, Chris PeBenito wrote:
> > Here is a patch against -test10 that adds an option for the security.
> > namespace (controlled by a configure option), which is used by SELinux
> > to store it's security labels.
> 
> What is the permissions model for the security attribute(s)?
> 
> It seems from the patch that nothing is enforced, e.g. for
> setting a new value it seems anyone can do it (which can't
> be right, can it?)...

Actually in terms of SELinux, this is correct, see the very simple ext3
handler (fs/ext3/xattr_security.c), and you'll notice there's not any
capability checks.  I'm not familiar with lsm hooks/internals, but I
would assume that the set is checked pretty early, way before it gets
down to the fs.  I did turn on SELinux auditing for the set (normally
only denials are audited) to make sure it was getting checked, and it
was.

-- 
Chris PeBenito
<pebenito@xxxxxxxxxx>
Developer,
Hardened Gentoo Linux
Embedded Gentoo Linux
 
Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE6AF9243
Key fingerprint = B0E6 877A 883F A57A 8E6A  CB00 BC8E E42D E6AF 9243

Attachment: signature.asc
Description: This is a digitally signed message part

<Prev in Thread] Current Thread [Next in Thread>