Enabling policy capabilities

Enabling policy capabilities

[Stephen Smalley]

Where do we stand on actually enabling policy capabilities in policy so
that people can start using newer features that depend on them?

I've definitely seen patches adding permissions for the peer checks, so
is there anything preventing us from trying to enable
network_peer_controls in policy and seeing what breaks (after Fedora 9
at this point, I suppose - unfortunate that we didn't enable it sooner)?

I haven't seen patches adding permissions for open other than just to
define them, IIRC.  So enabling open_perms would be rather bad right now
except for unconfined domains.  As a possible strategy for gradual
roll-out of open perm, we could add open everywhere there is a read or
write granted, enable the open_perms capability, verify no breakage, and
then gradually remove open permission where we know it to be unneeded.

-- 
Stephen Smalley
National Security Agency


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: Enabling policy capabilities

[Paul Moore]

On Thursday 10 April 2008 9:38:39 am Stephen Smalley wrote:
> Where do we stand on actually enabling policy capabilities in policy
> so that people can start using newer features that depend on them?
>
> I've definitely seen patches adding permissions for the peer checks,
> so is there anything preventing us from trying to enable
> network_peer_controls in policy and seeing what breaks (after Fedora
> 9 at this point, I suppose - unfortunate that we didn't enable it
> sooner)?

I still owe Chris an updated set of patches for refpolicy to put all the 
right unlabeled checks in place for the new peer controls.  There have 
been lots of patches on the lists but none have been right, yet :)

Once I get the 2.6.26 patches straightened out I'm going to work on 
those.

-- 
paul moore
linux @ hp

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: Enabling policy capabilities

[Christopher J. PeBenito]

On Thu, 2008-04-10 at 10:01 -0400, Paul Moore wrote:
> On Thursday 10 April 2008 9:38:39 am Stephen Smalley wrote:
> > Where do we stand on actually enabling policy capabilities in policy
> > so that people can start using newer features that depend on them?
> >
> > I've definitely seen patches adding permissions for the peer checks,
> > so is there anything preventing us from trying to enable
> > network_peer_controls in policy and seeing what breaks (after Fedora
> > 9 at this point, I suppose - unfortunate that we didn't enable it
> > sooner)?
> 
> I still owe Chris an updated set of patches for refpolicy to put all the 
> right unlabeled checks in place for the new peer controls.  There have 
> been lots of patches on the lists but none have been right, yet :)
> 
> Once I get the 2.6.26 patches straightened out I'm going to work on 
> those.

I added a policy_capabilities file with the two existing caps commented
out.

-- 
Chris PeBenito
Tresys Technology, LLC
(410) 290-1411 x150



--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: Enabling policy capabilities

[Daniel J Walsh]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stephen Smalley wrote:
> Where do we stand on actually enabling policy capabilities in policy so
> that people can start using newer features that depend on them?
> 
> I've definitely seen patches adding permissions for the peer checks, so
> is there anything preventing us from trying to enable
> network_peer_controls in policy and seeing what breaks (after Fedora 9
> at this point, I suppose - unfortunate that we didn't enable it sooner)?
> 
> I haven't seen patches adding permissions for open other than just to
> define them, IIRC.  So enabling open_perms would be rather bad right now
> except for unconfined domains.  As a possible strategy for gradual
> roll-out of open perm, we could add open everywhere there is a read or
> write granted, enable the open_perms capability, verify no breakage, and
> then gradually remove open permission where we know it to be unneeded.
> 
Open checks will be added in Fedora 10, along with turning on Xace.  We
are frozen in Fedora 9.  No new functionality.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkf+OlIACgkQrlYvE4MpobM+4gCcCHxrHzMnej50qajUUFTqMU3j
BMcAn3JWNm2zr6nl6QiyqZbWwfLSjQx1
=bMXX
-----END PGP SIGNATURE-----

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.