Added support for EC SSH keys encrypted with AES-256-CBC

[peshev]

No description provided.

[magnumripper]

Can we enable some of the commented-out test vectors (nicked from hashcat in #5633) with these changes? If so, I suggest you do so as well. Thanks!

[solardiz]

Thank you for the contribution @peshev!

@Nothing4You This PR reworks your addition from #5641 (why does it, @peshev?), you could want to test whether the "hash" produced by the script as revised here is still also usable with hashcat or not - and let us all know. Thank you!

[peshev]

@solardiz The fact that both I and @Nothing4You submitted similar PRs in a short timespan is probably not a coincidence. I assume both of us were doing the same thing: a CTF-style challenge, part of 38C3 - TELNET-KLARTEXT-REDEN.

His PR allows the ssh2john.py script to produce hash files that work in hashcat - unfortunately they do not work in john.
My version introduces a new cipher type (7), and its implementation in the john code - unfortunately, the hasfile produced by the script doesn't work in hashcat, since hashcat doesn't have cipher type 7.

I'll do some further work on this, in order to allow the ssh2john.py script to produce files that work in both john and hashcat, and fix #5634

[solardiz]

I'll do some further work on this, in order to allow the ssh2john.py script to produce files that work in both john and hashcat, and fix #5634

Thank you! Please also add a doc/NEWS entry summarizing these improvements.

[Nothing4You]

@solardiz The fact that both I and @Nothing4You submitted similar PRs in a short timespan is probably not a coincidence. I assume both of us were doing the same thing: a CTF-style challenge, part of 38C3 - TELNET-KLARTEXT-REDEN.

This was indeed what prompted me to PR this :)

I'm not sure though why this would need to introduce a new hash type rather than making $5$ work with John?
Especially as that is already working with Hashcat, that seems to be the more sensible change, unless there is some fundamental difference in implementations that would prevent this?

[magnumripper]

I'm not sure though why this would need to introduce a new hash type rather than making $5$ work with John?
Especially as that is already working with Hashcat, that seems to be the more sensible change, unless there is some fundamental difference in implementations that would prevent this?

Right, we definitely want hashcat and john use the same input "hash" format if at all possible. For brand new formats to both, we normally seek consensus on the format beforehand nowadays.