Re: [PATCH] selinux: init target class when add avc callback

[Stephen Smalley <sds@tycho.nsa.gov>]

On Tue, 2012-03-06 at 19:41 -0500, Eric Paris wrote:
> On Tue, 2012-03-06 at 16:15 -0800, Andrew Morton wrote:
> > On Wed, 07 Mar 2012 07:59:30 +0800
> > Wanlong Gao <gaowanlong@cn.fujitsu.com> wrote:
> > 
> > > On 02/05/2012 09:53 AM, Wanlong Gao wrote:
> > > 
> > > > Target security class should be initialized when add avc callback.
> > > > Although tclass is userless in callbacks now, but it may be used
> > > > in the future .
> > > > 
> > > > Signed-off-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
> > > > ---
> > > >  security/selinux/avc.c |    1 +
> > > >  1 files changed, 1 insertions(+), 0 deletions(-)
> > > > 
> > > > diff --git a/security/selinux/avc.c b/security/selinux/avc.c
> > > > index dca1c22..27495e6 100644
> > > > --- a/security/selinux/avc.c
> > > > +++ b/security/selinux/avc.c
> > > > @@ -576,6 +576,7 @@ int avc_add_callback(int (*callback)(u32 event, u32 ssid, u32 tsid,
> > > >  	c->events = events;
> > > >  	c->ssid = ssid;
> > > >  	c->tsid = tsid;
> > > > +	c->tclass = tclass;
> > > >  	c->perms = perms;
> > > >  	c->next = avc_callbacks;
> > > >  	avc_callbacks = c;
> > 
> > Perhaps James can take a look at this?
> > 
> > avc_add_callback() looks a bit odd.  It uses GFP_ATOMIC, but that is
> > unnecessary because avc_add_callback() is only ever called from
> > module_init() code.  And if it isn't only ever called from
> > module_init() code then it needs some locking for that list.
> 
> I'm a bad maintainer.  I should have done something with this patch.
> Adding sds, the only other person who ever actually maintains this code,
> to the thread.
> 
> __initcall() functions aren't serialized?  I guess that would be bad and
> we would need a lock.  I wonder if there are other places I assumed
> __initcall() would be serialized (note that all of these call sites are
> built in and not modules if that makes a difference)
> 
> I'll probably just rip all of that ssid, tsid, tclass, perms, stuff out.
> If all these years noone uses callbacks for anything other than reset
> why do we have it at all.  Probably more simplification we can do around
> avc_update_node() too...
> 
> Stephen, thoughts on ripping stuff out?

Yes, you should be able to replace GFP_ATOMIC with GFP_KERNEL there and
rip out the other callback fields/arguments as they are presently
unused.  Legacy of the original Flask code, where there were other
avc_ss_* interfaces for revoke and friends.

-- 
Stephen Smalley
National Security Agency