Meltdown and Spectre Vulnerability

[Justin.Abendroth]

Are PA devices effected by the meltdown and spectre vulnerability and if so, are these signatures known yet by PA?

[djr]

Default Palo Action on any new threat = Alert 

...as far as I can tell.   I have always presumed it is deliberate, but I don't know how long they leave it before they feel confident enough to block.  We have spotted some very mature Critical or High threats which are still only "alert" as default action. 

[OtakarKlier]

Hello,

The defaults are set to alert as a safety measure. WHat we do is copy the default Vulnerability policies and manuall set the action based on severity. I.e. anything medium and higher is blocked or reset, etc. This way when new signatures come out, they may be classified as high or critical, but our policies are set so that they will be blocked in some fashion.

Hope that makes sense.

Regards,

[chyates]

So you override the default actions for those severities?  I have that configuration in place in some places, but the way the default actions as described to my be an SE from Palo Alto which got me thinking about that approach.  He said something to the effect that default action is evaluated very differently than severity.  They are basically set based on confidence that there will be no false positives.

Another example he gave was the Poodle signature which simply detected any SSLv3 - which you certainly wouldn't want to block, particularly at that time.  Granted that was an informational signature, not medium or higher, but you can look at Vulnerability Protection signatures '( severity contains 'critical' ) and ( action contains 'alert' )' and see 1704 signatures (over 1/10th) where the default action is alert (and one that is allow.)  

Do you even alert below medium?  All I have investigated in those severities have ended up on our whitelist so far.

[djr]

We do drop/reset Critical and High because we have seen serious vulnerabilities being ignored because they are alert only, but we had worked out that the PAN approach was more based on confidence.  That makes lots of sense, but only if the action changes after a few days or so when they have worked out how reliable the detection is.  We have found that they stay on alert for months which just gives too large a window for attack.

We do alert everything but our logs are huge!  It's a good point maybe we don't need to, we just tend to log as much as possible so we can see what is going on.

[OtakarKlier]

Hello @chyates,

Yes, I am actually thinking of just performing a server side reset on my external facing zones so my servers connection gets reset but the attackers does not so it keeps their session open :).

Cheers!