Anyway, if you rely on BitLocker in TPM mode (boot without PIN), you should know that anyone can steal your computer, sniff 32 bytes off of the LPC bus, stick them into libbde, and decrypt your disk. Yes, it's that easy. Solder 7 wires to $favorite_fpga_board, decrypt drive.
-
Show this thread
-
We all knew it was going to be this dumb, but today I'm here to tell you it *is* this dumb and I just implemented it.
4 replies 29 retweets 139 likesShow this thread -
Replying to @marcan42 @mjos_crypto
Great work! Do you think a similar attack could be made against Android "DE" storage on a Pixel phone?
1 reply 0 retweets 0 likes -
Replying to @ciphergoth @mjos_crypto
ARM devices are usually a few orders of magnitude closer to being potentially secure than x86 systems. The "TPM" on those is integrated into the SoC. I'm sure you could still find a way in though, like a glitching attack. This is why I refuse to use Android FBE.
1 reply 0 retweets 0 likes -
Replying to @marcan42 @mjos_crypto
You prefer FDE to FBE? Because all storage is CE, none is DE? Would be interested to know more about the kind is attack you have in mind...
1 reply 0 retweets 0 likes -
Replying to @ciphergoth @mjos_crypto
Yes, except I use a different (much longer) FDE passphrase that is not my normal unlock code (this is possible on rooted devices). Also FDE encrypts *everything*, including all metadata, so I'm immune from "someone used the wrong data class" bugs.
2 replies 0 retweets 0 likes -
The problem with FBE and all the DE stuff (and what iPhone does too) is that it's quite secure if nobody made any mistakes all the way through the stack; everything has to work together perfectly. A security flaw could make the whole thing moot, as we've seen many times.
1 reply 0 retweets 0 likes -
With FDE I know that the storage is cryptographically bound to the long and not crackable passphrase, period. As long as the much simpler code from that to the storage crypto is secure, if my phone is off, you aren't getting any data out of it.
1 reply 0 retweets 1 like -
Replying to @marcan42 @mjos_crypto
I see what you mean. At least the relevant code is open source.
1 reply 0 retweets 0 likes -
Replying to @ciphergoth @mjos_crypto
It isn't. The Qualcomm bootloaders aren't open source and are in the critical path; also the *other* thing they put into TrustZone is DRM blob nonsense (Widevine) and *that* is also exploitable (and then you can escalate). See e.g. https://googleprojectzero.blogspot.com/2017/07/trust-issues-exploiting-trustzone-tees.html …
1 reply 0 retweets 1 like
If you mean for FDE, that still goes through TrustZone for key derivation, but the inputs and outputs are open source, so as long as the TrustZone side isn't doing anything monumentally stupid (like stashing away a copy of your credentials) you're fine.
-
-
Tn that case, the TrustZone side provides rate limiting, so compromising TrustZone lets you attack the passphrase offline and much faster, but if it's long enough to be secure that isn't a problem.
0 replies 0 retweets 0 likesThanks. 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.