Re: [PATCH] LSM: ModPin LSM for module loading restrictions

[Casey Schaufler]

On 10/16/2013 3:43 PM, Kees Cook wrote:
> On Wed, Oct 16, 2013 at 2:42 PM, Casey Schaufler <casey@xxxxxxxxxxxxxxxx> wrote:
>> On 10/16/2013 1:47 PM, Tetsuo Handa wrote:
>>> Kees Cook wrote:
>>>> Any update on this? It'd be nice to have it in linux-next.
>>> What was the conclusion at LSS about multiple concurrent LSM support?
>>> If we agreed to merge multiple concurrent LSM support, there will be nothing to
>>> prevent this module from merging.
>>>
>> Yeah.
> The discussion at LSS basically centered around the catch-22 of not
> being able to stack, and not having anything to stack (since Yama got
> an hard-coded exception). So I sent this LSM as one I'd been waiting
> for stacking on. Essentially, I'm breaking the catch-22 by sending
> this. I'd like it to get into the tree so we don't have a catch-22
> about stacking any more. :)
>
>> The conclusion was that it needs to be staged because it's
>> too much to swallow all at once. I can see that. It's going
>> to be a lot of work to rearrange and rebase. That's a chunk
>> of time I don't expect to have for a while. It looks good
>> to happen, but don't hold supper for me.
> Do you want me to take a stab at it? It sounds like it was desirable
> to cut the current series into two halves? The core changes first, and
> the userspace interface changes next?

My read on it was a three phased approach:

First, move the cap "module" checks out of the other modules
and directly into security.c. There would be no "default" module.
If another module is loaded, call the hook it defines if the cap
check passes. Add /sys/kernel/security/lsm to make it easy to
find out what module (if any) is active.

Second, allow more than one LSM to get called if so requested.
Call them all, and return the error code of the last failure.
Refuse to load more than one module that uses an exclusive feature;
netlabel, secmark or XFRM.

Finally, put in all the gimmicks to decide who gets which of the
networking facilities.

Yes, If you've got the cycles to work with it I'd be happy for the help.


>
> -Kees
>

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