,
11 tweets,
3 min read
Read on Twitter
So, I took a look at swiss online voting system code that someone leaked, and having written, deployed and audited large enterprise java code...that thing triggers every flag.
The core reencryption mixnet code is spread across dozens of different files, not included the auxiliary/utility/deployment packages.
Also this work in progress is reassuring
![]()
* a "packed" ciphertext can be mixed, and the data can be recovered, but the proofs dont verify."" src="/images/1px.png" data-src="https://pbs.twimg.com/media/DzofYx9UcAAMYGY.png">
Also this work in progress is reassuring
* a "packed" ciphertext can be mixed, and the data can be recovered, but the proofs dont verify."" src="/images/1px.png" data-src="https://pbs.twimg.com/media/DzofYx9UcAAMYGY.png">The code looks like it is doing the right things, but this is code that was not written to be easily audited which is concerning for such a security sensitive system.
(tbf it's hard to write enterprise java in an easily audible way, but I've seen it done)
(tbf it's hard to write enterprise java in an easily audible way, but I've seen it done)
Part of my job used to be "here is a large enterprise system and some documentation, certify that the team are following good security practice and highlight any vulnerabilities and risks"
If someone had given me this system, I'd be feeling pretty confident I'd find stuff.
If someone had given me this system, I'd be feeling pretty confident I'd find stuff.
oh ffs, how about some basic defense in depth people. This is not how you start a function designed to verify a proof as being correct.
*headdesk*
*headdesk*
If anyone can point me towards the tests for the proof code that would be good because....I can't find any.
I've just spent an hour of my precious life trying to determine the encryption parameters used in the above proofs and I still have no clue, every example config/parameter that relates to it seems to be no where close to sensible or secure - which is a big flag by itself.
See also this subthread:
I feel like this old tweet of mine on reviewing crypto code is relevant:
Anyway, if this was an actual review, instead of me being having a perverted sense of a relaxing Sunday, then I would probably be prioritizing packages and writing a diplomatically worded email to the team about time constraints.
Thankfully that is not my job this time.
Thankfully that is not my job this time.
the scoping doc for the public pen test is about as expected.
