Microsoft's Bitlocker and TPM encryption combo defeated

https://www.pcgamer.com/microsofts-bitlocker-and-tpm-encryption-combo-defeated-with-a-dollar10-raspberry-pi-and-a-bit-of-braininess/

by cannibalXxx • 2 years ago

6 5 2 years ago

pcgamer.com

nonrandomstring2• 2 years ago

Microsoft says Bitlocker works best when used in combination with a TPM chip

Microsoft is wrong and has issued bad advice.

Any disk encryption works best when used only with an ephemeral pass-phrase known solely to the operator. TPM is an okay solution for a server in a physically secure rack that must be booted remotely without human intervention.

omgCPhuture3• 2 years ago

1. TPM is Common Criteria certified to about a chip bank card. I think 5 or so on CC eval.

2. Many, if not most, consumer motherboards wrongly implements TPM and easily reveals secrets to a logic analyzer.

3. I am yet to see th TPM that is not on that situation AND/OR has several side channel problems.

4. SoC method really just makes physics a problem as die nanometer shrinks; but even before that, not think implementation well through from a mindset like the problem solving mind of an experienced, skilled hacker and end up implementing bugs in hardware format.

TEE and the likes are no exception. DPA, DFA, Accoustic analytics, pick your SCA poison when you are done with the weak math or usual code suspects like buffers, data type overflows, logic bugs, etc. Fault injections and bounds overwriting are classic bugs with endless ways to abuse the given situation.

MS has been known to shit the bed coding, over & over. The fact is you never know who sees opurtunity where closed source corp culture only sees $ in the short term: Nobody will ever know & the bottom line better.

5. Knowledge is power, ignorance is dangerous. That's how you end up implementing your idea of defense & end up with someone else's idea of a doormat.

6. Never underestimate hubris, ignorance or man's capacity to persuade himself to be lazy or any combination there of.

If you cannot and have not inspected it you should not trust it. </blog>.

nonrandomstring2• 2 years ago

And was I dreaming that Cavium/Marvell had probably got funky dual elliptic curve RNGs? Didn't that come out in Jacob Appelbaum's PhD and then... wind/crickets/tumbleweeds....

omgCPhuture3• 2 years ago

The fact is most encryption is backdoored, but more and more by way of implementation not kleptographic backdoors. That can be summed up in some math, more over ISO+world+dog have seen weakened math from the NSA who cleverly publish field manuals admitting to doing it over and over and never working. Implementation specific buggery hidden in plain sight takes skill or flouride in the tap water. Fact is most commerical cell phones, commercial routers, computers all have that sort of thing. Just ask Cisco, who will deny IPSEC was weakened for the NSA and FBI. RSA BSAFE, CtyptoPhone from Crypto AG. NagraVision.

Apache_scalp.c was a clever orchestration that likely shook Theo De Raadt to the Xanax Core when GObbles GOBBLED about it on openbsd.org using historical (un)hidden references. Fact is if you can audit it and you do not have the prerequisits needed to have an informed audit you should assume you know nothing other than some stranger wrote some code that claims this or that.

Microsoft's Bitlocker and TPM encryption combo defeated

Related Stories