miki-bgd
commented
1 year ago I have found solution, so here is explanation for future readers.
3 . Javax.Crypto AES/CTR/NoPadding algorithm returns results even if you provide array of bytes which is shorter than block size (for AES it is 128b/16B). You provide 4B, it will return E/D(encrypted/decrypted) 4B (this is because E/D is performed on complete block of 16B, but bytes do not have any correlation between them; you can provide bytes 3,4,5, and result will be at those indexes as well). BouncyCastle doesn't return result, until complete internal buffer is full (16B). If DoFinal is invoked, then result is good, but it also performs reset of the counter, which means invoking E/D won't work anymore, because counter is disrupted.
1 . I believe i have understanding about this. IV and counter combination, in AES/CTR must be used only once, in order to preserve security. One instance can do either encryption or decryption. Since counter is initialized from IV (copied, precisely), for both encryption and decryption, it would need 2 IV (remember IV and C combination can be used only once). I believe it was design decision. One thing is little bit unusual (at least to me), this approach means that data is first decrypted by idf, and client will need to encrypt it to get information.
2 . DoFinal will do padding and will return result no matter if buffer is full, but it will also reset counter which means encryptor/decryptor can't be used anymore.
PS. This is just my understanding. Would be nice if someone confirms it, and maybe gives some additional explanation.
Cheers
d-rahul
Hello,
Application works on Android, i tried both in Java and ported to Xamarin C#. However, i am not experienced in cryptography, and would appreciate some explanations about AES algorithm.
Thanks in advance, Miki