linux-kernel - Re: [PATCH -v3 -mm] LSM: Add security= boot parameter

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 03 Mar 2008 10:54:02 -0500
From:	Stephen Smalley <sds@...ho.nsa.gov>
To:	"Ahmed S. Darwish" <darwish.07@...il.com>
Cc:	James Morris <jmorris@...ei.org>,
	Casey Schaufler <casey@...aufler-ca.com>,
	Adrian Bunk <bunk@...nel.org>,
	Chris Wright <chrisw@...s-sol.org>,
	Eric Paris <eparis@...isplace.org>,
	Alexey Dobriyan <adobriyan@...ru>,
	LKML <linux-kernel@...r.kernel.org>,
	LSM-ML <linux-security-module@...r.kernel.org>,
	Anrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH -v3 -mm] LSM: Add security= boot parameter


On Mon, 2008-03-03 at 17:35 +0200, Ahmed S. Darwish wrote:
> Hi James,
> 
> On Mon, Mar 03, 2008 at 07:29:22PM +1100, James Morris wrote:
> > On Sun, 2 Mar 2008, Ahmed S. Darwish wrote:
> > 
> > > Add the security= boot parameter. This is done to avoid LSM 
> > > registration clashes in case of more than one bult-in module.
> > > 
> > > User can choose a security module to enable at boot. If no 
> > > security= boot parameter is specified, only the first LSM 
> > > asking for registration will be loaded. An invalid security 
> > > module name will be treated as if no module has been chosen.
> > > 
> > > LSM modules must check now if they are allowed to register
> > > by calling security_module_enable(ops) first. Modify SELinux 
> > > and SMACK to do so.
> > 
> > I think this can be simplified by folding the logic into 
> > register_security(), rather than having a two-stage LSM registration 
> > process.
> > 
> > So, this function would now look like
> > 
> > 	int register_security(ops, *status);
> > 
> > and set *status to LSM_WAS_CHOSEN (or similar) if the module being 
> > registered was also chosen via the security= parameter.  If there is no 
> > value for the parameter, the first module to register is automatically 
> > chosen, to preserve existing behavior.
> > 
> > The calling code can then decide what to do, e.g. not panic if 
> > registration failed and the LSM was not chosen; panic on failure when it 
> > was chosen.
> > 
> 
> I liked to do it like that at first, but I faced two problems:
> 
> SElinux (As you already know ;)) does the security setup of the initial 
> task before calling register_security. Would it be safe to do this
> setup after registeration ?

I wouldn't recommend it - the hook functions presume that the initial
task security blob has been set up already, and that other dependencies
like the inode security cache and access vector cache have been created
and can be used.  We have to assume that the security hooks can start
being invoked as soon as we call register_security(), even if in
practice it won't happen until after the init function completes.

> Same case occurs for Smack, it does some locking initializations and
> setup initial task's security before registration.
> 
> Personally, I feel that it's nicer to let the LSM know if it's
> OK to initialize itself before hitting _the point of no return_ (registration).
> 
> Anyway, I have no problem to implement it using *status if my 
> concerns are wrong.
> 
> > > +static atomic_t security_ops_enabled = ATOMIC_INIT(-1);
> > 
> > I'd suggest getting rid of this atomic and using a spinlock to protect the 
> > global chosen_lsm string, which is always filled when an LSM registers.
> > 
> > >  
> > > +/* Save user chosen LSM */
> > > +static int __init choose_lsm(char *str)
> > > +{
> > > +	strncpy(chosen_lsm, str, SECURITY_NAME_MAX);
> > > +	chosen_lsm[SECURITY_NAME_MAX] = NULL;
> > 
> > You should never need to set the last byte to NULL -- it's initialized to 
> > that and by definition should never be overwritten.
> > 
> > > +int security_module_enable(struct security_operations *ops)
> > > +{
> > > +	if (!ops || !ops->name)
> > > +		return 0;
> > 
> > Lack of ops->name during registration needs to be a BUG_ON.
> > 
> 
> You'll find above three points fixed the next send. Thank you.
> 
> Regards,
> 
-- 
Stephen Smalley
National Security Agency

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/