Re: [PATCH] selinux-testsuite: Review and rework sctp tests

[Richard Haines <richard_c_haines@btinternet.com>]

On Wed, 2020-06-03 at 16:50 +0200, Ondrej Mosnacek wrote:
> Hi Richard,
> 
> On Wed, Jun 3, 2020 at 1:01 PM Richard Haines
> <richard_c_haines@btinternet.com> wrote:
> > The main changes have been to sctp_asconf_params_client.c and
> > sctp_asconf_params_server.c to make them more reliable for running
> > the
> > client and server on different systems.
> > 
> > Updated common code in sctp_common.c
> > 
> > Removed obsolete code that was used to test permissions that never
> > made it
> > to the final commit.
> > 
> > Also reviewed and updated tests to check policy rules for denying
> > access to all possible net/sctp kernel hooks as described in the
> > kernel
> > Documentation/security/SCTP.rst
> 
> This looks promising, but would it be possible to split the patch
> into
> a series of smaller changes? It is hard to review such a large patch
> that changes a lot of things at once...

I'll split into three patches 1) Remove obsolete 2) Update asconf/test
3) Update remaining tests

> 
> > Signed-off-by: Richard Haines <richard_c_haines@btinternet.com>
> > ---
> >  policy/test_sctp.te                    |  98 ++++--
> >  tests/sctp/.gitignore                  |   3 -
> >  tests/sctp/Makefile                    |   2 +-
> >  tests/sctp/sctp_asconf_params_client.c | 320 ++++++++-----------
> >  tests/sctp/sctp_asconf_params_server.c | 269 +++++++++-------
> >  tests/sctp/sctp_client.c               |   4 +-
> >  tests/sctp/sctp_common.c               | 189 ++++++++++-
> >  tests/sctp/sctp_common.h               |  12 +-
> >  tests/sctp/sctp_connectx.c             |  33 +-
> >  tests/sctp/sctp_peeloff_server.c       |  42 +--
> >  tests/sctp/sctp_server.c               |   4 +-
> >  tests/sctp/sctp_set_params.c           | 205 ------------
> >  tests/sctp/sctp_set_peer_addr.c        | 415 -------------------
> > ------
> >  tests/sctp/sctp_set_pri_addr.c         | 135 --------
> >  tests/sctp/test                        |  95 +++++-
> >  15 files changed, 688 insertions(+), 1138 deletions(-)
> >  delete mode 100644 tests/sctp/sctp_set_params.c
> >  delete mode 100644 tests/sctp/sctp_set_peer_addr.c
> >  delete mode 100644 tests/sctp/sctp_set_pri_addr.c
> > 
> > diff --git a/policy/test_sctp.te b/policy/test_sctp.te
> > index 3b16db1..380e442 100644
> > --- a/policy/test_sctp.te
> > +++ b/policy/test_sctp.te
> > @@ -60,6 +60,12 @@ corenet_sctp_bind_all_nodes(test_sctp_server_t)
> >  corenet_inout_generic_node(test_sctp_server_t)
> >  corenet_inout_generic_if(test_sctp_server_t)
> > 
> > +# Don't allow ports < 1024
> > +# neverallow test_sctp_server_t self:capability net_bind_service
> > +# neverallow test_sctp_server_t reserved_port_t:sctp_socket
> > name_bind;
> > +dontaudit test_sctp_server_t self:netlink_route_socket { create
> > getattr bind };
> > +dontaudit test_sctp_server_t self:udp_socket { getattr connect };
> 
> Could you clarify what these dontaudit rules are for?
> 
> That's the only thing that caught my eye after a quick glance. I'm
> still hoping you will be able to split up the patch into more
> digestible chunks, so I'm not going to look at it deeper just yet. If
> that proves impossible or too difficult, I'll try to review the patch
> as-is, but please understand that it may take a long time then...
> 
> Thanks,
>