So, the problem with USB tokens that we basically have two choices: - Unauditable black boxes built on *supposedly* more secure ICs that require NDAs to develop for - Open and auditable, but definitely pwnable off the shelf microcontrollers. Which poison do you prefer?
-
Show this thread
-
Replying to @marcan42
Well, I'd imagine that most people will keep them on them and within their sight at all times. If they can be trusted to revoke the keys if they leave their sight, and if the owner of the key doesn't mind the end user having access - that'd probably be most of the threat model.
1 reply 0 retweets 1 like -
Replying to @justinrwlynn @marcan42
Using epoxy resin embedding w/ non-conductive glitter and photo identitifcation cards, it should be possible to create a tamper resistant case for microcontrollers which are otherwise physically insecure - given the above.
1 reply 0 retweets 0 likes -
Replying to @justinrwlynn
You can still voltage/clock glitch the things pretty easily through epoxy; if you need to disable decoupling capacitors that's pretty easy to do with a well-placed drill hole. TBH decapsulation isn't even that hard.
1 reply 0 retweets 2 likes -
Replying to @marcan42
The point is they'd have to cause physical damage and the glitter would be impossible to replicate without extreme cost. It isn't to make the device tamper proof, just tamper resistant for a few minutes or so.
2 replies 0 retweets 0 likes -
Replying to @justinrwlynn
Yeah, but revoking PGP/SSH keys is less trivial, so a higher level of physical tamper resistance to key extraction is desirable depending on your use case.
1 reply 0 retweets 2 likes -
Replying to @marcan42
Suggest certificate authentication with ssh if possible, with a key as fallback. But yes, the epoxy would help give you time to revoke (say a call to a twilio app or something) - and the primary protection is keeping the key on your person at all times.
1 reply 0 retweets 0 likes -
Replying to @justinrwlynn @marcan42
you could produce a layered multi-PCB construction with a randomised protection grid within each multi-layer PCB the interruption of which would zeroise the microcontroller ... etc. etc. there are lots of possiblities here.
1 reply 0 retweets 0 likes -
Replying to @justinrwlynn
Then you need it to use battery-backed SRAM. Which is how high-end HSMs work, but kind of significantly increases the cost/complexity (and chance of losing the key material accidentally).
1 reply 0 retweets 1 like -
Replying to @marcan42
possibly, or a supercap w/ ultra-low power sram for seed material - or possibly flash memory encrypted with some PBKDF function over the PIN code + a random data in the OTP-rom in the micro.
2 replies 0 retweets 0 likes
Yeah, you'd definitely want cryptographic binding between the PIN and the key material (which at least older YubiKeys absolutely did *not* do, not sure about newer ones). That helps no matter the µC you use.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.