[Cryptography] Fwd: OPENSSL FREAK

[ianG]

I'm going to assume that every device has a remote management interface.

Then, channeling all the posts, and not crediting names:



1. the RMI must be capable of defending itself because likely it itself 
might not be able to be upgraded.  Therefore it has a small and strong 
kernel.

2. The RMI can be used to upgrade the device.

3. A death notice could just be limited to killing the networking?

4. The RMI could be used to kill the device.

5. If the owner just wants the device to keep running because it works 
for her, even when we know it to be secure, then she can disconnect it 
from the net / run a private net.  Which she'll probably want to do anyway.

5. If she doesn't firewall it into a private net, then likely over time, 
the device might get hit by a death notice.

6. Which would be a bit shocking.  So maybe there needs to be a warning? 
  A grace period?  Some way in which the device alerts the user that she 
needs to disconnect the fridge.

7. Complexity of this arrangement seems to indicate that all such 
devices will then be coupled with private networks and private 
management consoles which manage the devices.

8. Then, the message might be that if you place your target on the open 
net, it's fair game, in whatever circumstance.  It's up to the installer 
and/or owner to put in the appropriate protection.  If not, they are 
being negligent.  With appropriate doco and PR, we could get this 
established sufficient to survive in court.

9. *We are looking at the wrong part*.  Instead of looking at the 
device, we should be thinking about the protocol for upgrade, 
management, tunnelling, virtual death imitation, etc, and the app&box 
that does that work.

10. Someone will make a lot of money here because that box is likely a 
lot higher margin than the average target ;-)



Close?

iang