floodyberry / poly1305-donna Goto Github PK

The code in poly1305-donna-16.h failed to pass the selftest and most of the test vectors from
http://tools.ietf.org/html/rfc7539. See my post and the answer at http://stackoverflow.com/questions/31806863/verification-of-poly1305-donna-16-h-code.

Hi, we played with poly1305-donna and we like that it's quite fast. We'd like and would like to ask how difficult it is, with this specific implementation, to cause a collision (e.g. generate a different message with the same MAC) for an attacker that would know the secret key passed to poly1305_auth()? Of course, we'd use a key only once per message.

Thank you and sorry for asking here, don't know where else.