[PATCH 4/5] LSM: Define SELinux function to measure security state

[John Johansen]

On 6/15/20 10:44 AM, Mimi Zohar wrote:
> (Cc'ing John)
> 
> On Mon, 2020-06-15 at 10:33 -0700, Casey Schaufler wrote:
>> On 6/15/2020 9:45 AM, Lakshmi Ramasubramanian wrote:
>>> On 6/15/20 4:57 AM, Stephen Smalley wrote:
>>>
>>> Hi Stephen,
>>>
>>> Thanks for reviewing the patches.
>>>
>>>>> +void security_state_change(char *lsm_name, void *state, int state_len)
>>>>> +{
>>>>> +       ima_lsm_state(lsm_name, state, state_len);
>>>>> +}
>>>>> +
>>>>
>>>> What's the benefit of this trivial function instead of just calling
>>>> ima_lsm_state() directly?
>>>
>>> One of the feedback Casey Schaufler had given earlier was that calling an IMA function directly from SELinux (or, any of the Security Modules) would be a layering violation.
>>
>> Hiding the ima_lsm_state() call doesn't address the layering.
>> The point is that SELinux code being called from IMA (or the
>> other way around) breaks the subsystem isolation. Unfortunately,
>> it isn't obvious to me how you would go about what you're doing
>> without integrating the subsystems.
> 
> Casey, I'm not sure why you think there is a layering issue here.
>  There were multiple iterations of IMA before it was upstreamed.  One
> iteration had separate integrity hooks(LIM).  Only when the IMA calls
> and the security hooks are co-located, are they combined, as requested
> by Linus.
> 
I don't see the layering violation here either, Casey has already
responded and I don't have anything to add

> There was some AppArmour discussion about calling IMA directly, but I
> haven't heard about it in a while or seen the patch.
> 
its lower priority than other work. I think calling IMA directly has use
cases but must be done very carefully, and well reviewed. I have would
have more concerns with IMA calling directly into the various LSMs.