2011-03-16 21:32:52

by Daniel Walsh

Subject: [refpolicy] I think refpolicy should replace all corenet.*all_nodes with corenet.*generic_node()

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

Similarly corenet.*all_if with corenet.*generic_if()

If I add a label to an interface I want no domains to be able to use it,
except unconfined domains and domains I specify in a policy module.

Since we have only one generic node and one generic interface, I think
these are bugs.

I have just changed Fedora F15 to match this assumption.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/

iEYEARECAAYFAk2BLIQACgkQrlYvE4MpobOSlQCcCoi6gQQ+FaMgrbdeAGa93Zqc
zkQAoMdMUkZ080563tJWSbE5Eor29EjB
=tadE
-----END PGP SIGNATURE-----

2011-03-17 12:16:33

by cpebenito

Subject: [refpolicy] I think refpolicy should replace all corenet.*all_nodes with corenet.*generic_node()

On 03/16/11 17:32, Daniel J Walsh wrote:
> Similarly corenet.*all_if with corenet.*generic_if()
>
> If I add a label to an interface I want no domains to be able to use it,
> except unconfined domains and domains I specify in a policy module.
>
> Since we have only one generic node and one generic interface, I think
> these are bugs.
>
> I have just changed Fedora F15 to match this assumption.

Yes, though the one exception is the perms on kernel_t. Last year I
made a pass through the policy to do this for nodes and netifs, but I
guess I missed a few.

--
Chris PeBenito
Tresys Technology, LLC
http://www.tresys.com | oss.tresys.com