Regin: Nation-state ownage of GSM networks

Curious; Symantec's report said the RC5 encryption was performed in CFB mode, but this is not the case. Instead, we have a weird combination of CBC and CFB, with CFB being used solely for the last block (instead of padding or ciphertext-stealing). This has the look of seriously legacy code.

This seems to be a very old mode: I have trouble finding references to it with less than 20 years. This mode is in [1, pg. 151] and [2, pg. 77], as far as I can find, both of which were published in 1982. [2] also introduced ciphertext stealing, albeit a bugged version [3, §6].

[1] http://faculty.nps.edu/dedennin/publications/Denning-Cryptog...

[2] http://www.amazon.com/Cryptography-Dimension-Computer-Securi... [sorry, no PDF link]

[3] http://web.cs.ucdavis.edu/~rogaway/papers/steal.pdf

How did they get to these GSM stations from the infected windows systems ?

Do BSC/BTS typically have IP-accessible configuration ?

Regardless of how they connected (I thought SS7 ...) wouldn't there be a login to access the switch BSC controls ... how would the malware have that ?

I don't understand how the malware running on the windows system executed the commands on the BSC.

The module names "legspin" and "willischeck" refer to the sport of Cricket.

> "Finally, the word 'shit' appears in many places throughout the code and modules."

Someone's boss is gonna be peeved! (Also, this would be a good indicator of a native English speaker developing the software; I doubt most developers put curse words into their code in a non-native language)

I don't find the GSM attacks unusual or interesting. Sure, it might be new for an Internet worm, but government-sponsored orgs have been pushing attacks on infrastructure like GSM networks since the L0pht days (though back then I think satellites were more in vogue), and they're constantly looking for new ways to infiltrate foreign networks of any type. This just shows someone's pet research project from a decade ago made it out of the lab.

It's also interesting to note that while this is useful for nation-states, they're not necessarily developed directly by a nation-state, and there could be multiple organizations using the same malware package. One way to see multiple independent actors using the same package would be to look for different cryptographic keys, C&C servers, and discrete distributed victim-networks.

I am surprised that the data files that are meant to be sent to the C&C aren't encrypted with a public key so only the intended recipients can decrypt them. This would thwart discovering what information had already been extracted after an infection was discovered.

The remarkable delay in reporting these findings is deliberate?

Normally reading something like this wouldn't interest me as it would be over my head but I found this fascinating. Thanks for posting it. It sounds like it's been known about for several years - why is it only being talked about now?

So according to the timestamps it's pretty clear it's only the US, not UK nor Israel, who developed this software.