Re: [PATCH] sysctl selinux: Don't look at table->de

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

On Mon, 2007-01-29 at 10:43 -0700, Eric W. Biederman wrote:
> Stephen Smalley <sds@tycho.nsa.gov> writes:
> 
> > On Sun, 2007-01-28 at 12:21 -0700, Eric W. Biederman wrote:
> >> With the sysctl cleanups sysctl is not really a part of proc
> >> it just shows up there, and any path based approach will not
> >> adequately describe the data as sysctl is essentially a
> >> union mount underneath the covers.  As designed this mechanism
> >> is viewer dependent so trying to be path based gets even worse.
> >> 
> >> However the permissions in sys_sysctl are currently immutable
> >> and going through proc does not change the permission checks
> >> when accessing sysctl.  So we might as well stick with the well
> >> defined sysctl sid, as that is what selinux uses when proc is
> >> not compiled in.
> >> 
> >> I.e.  I see no hope for salvaging the selinux_proc_get_sid call
> >> in selinux_sysctl so I'm removing it.
> >> 
> >> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
> >> ---
> >>  security/selinux/hooks.c |    8 ++------
> >>  1 files changed, 2 insertions(+), 6 deletions(-)
> >> 
> >> diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c
> >> index 7b38372..3a36057 100644
> >> --- a/security/selinux/hooks.c
> >> +++ b/security/selinux/hooks.c
> >> @@ -1438,12 +1438,8 @@ static int selinux_sysctl(ctl_table *table, int op)
> >>  
> >>  	tsec = current->security;
> >>  
> >> -	rc = selinux_proc_get_sid(table->de, (op == 001) ?
> >> -	                          SECCLASS_DIR : SECCLASS_FILE, &tsid);
> >> -	if (rc) {
> >> -		/* Default to the well-defined sysctl SID. */
> >> -		tsid = SECINITSID_SYSCTL;
> >> -	}
> >> +	/* Use the well-defined sysctl SID. */
> >> +	tsid = SECINITSID_SYSCTL;
> >>  
> >>  	/* The op values are "defined" in sysctl.c, thereby creating
> >>  	 * a bad coupling between this module and sysctl.c */
> >
> > NAK.  Mapping all sysctls to a single security label prevents any kind
> > of fine-grained security on sysctls, and current policies already make
> > use of the current distinctions to limit access to particular sets of
> > sysctls to particular processes.  As is, I'd expect breakage of current
> > systems running SELinux from this patch, because (confined) processes
> > that formerly only required access to specific sysctl labels will
> > suddenly run into denials on the generic fallback label.
> 
> Reasonable.  There is the issue that your code already had this code
> path for when /proc was compiled out.

True, but a system that disables proc is likely a system with a custom
policy anyway, and dependency on proc is fairly basic to selinux these
days (due to reliance on /proc/self/attr for process attribute
manipulation in place of the old selinux syscalls).  Possibly we should
just make selinux depend on proc and drop the #ifdef there.

> > If the ctl_table supplied more information about the functional purpose
> > and the security sensitivity of the sysctl, then we could leverage that
> > information instead, as long as we can at least derive the current
> > labelings from that information for compatibility.
> 
> What do information do you need to do need?  Do you need extra fields in sysctl?
> I am more than willing to help but I am not familiar enough with selinux
> to do a reasonable job on my own.

At present, we map the sysctls into functional groups (e.g. net, vm,
fs, ...) that parallel the sysctl hierarchy so that we can limit access
to only those programs/processes that need access for their purpose, and
further partition where it makes sense to do so.  We also separate out
particularly security sensitive ones like modprobe and hotplug.  So if
the ctl_table carried some indication of functional grouping and
security relevance (for some relatively small number of equivalence
classes), then we could map those to labels instead of the current
scheme.  And if we could have the ctl_table inherit the information from
its logical "parent" in the hierarchy by default, then it shouldn't
require too invasive a patch.

-- 
Stephen Smalley
National Security Agency