NewLC - Same encryption using c++ symbian & C# .net - Comments

Re: Same encryption using c++ symbian & C# .net

alh — Thu, 07 Dec 2006 13:29:45 +0100

Its not a "limitation", its just how block cryptos work. They work on fixed size blocks.

If the output from your encryption with the managed class, isn't exactly 32 bytes, or a multiple of it, it means it has some more information in it.

To encrypt more then one block you use some kind of "chaining" wich is to use some information from block 1 to encrypt block 2, some from block 2 to encrypt block 3 etc, this to counter "known plain text" attacks. There are different techniques to do this chaining, where I think CBC is the most common. (havn't fiddled with crypto in a long time so I might be wrong though)

You should read up some on encryption theory, or you are bound to make some small mistakes that breaks the security.

Re: Same encryption using c++ symbian & C# .net

Brekko74 — Thu, 07 Dec 2006 11:40:36 +0100

Thank you very much, now i'm trying to use in .net the Rijndael class instead of RijndaelManaged.

rbrunner:
Encrypted data is encoded (Base64) just to avoid binary data format. Then i send it with a socket to a server over the internet.
While in debug i can see the encrypted buffer and i can see that is differend from the same source text encrypted from the pc server program using the RijndaelManaged class of .net
So, i don't think the transport could be the reason....maybe....

alh:
I found on internet that 32 bytes is the real limit os AES encryption ... so it seems you are on the right way...
RijdaelManaged class allow me to encrypt a longer string as surely manage the chaining of encrypted blocks...
As i don't know how this class separates encrypted blocks i think i will have to manage the chaining both in pc and symbian.
Now i will try to use the simple Rijndael class on .net ...
Later i will post the codes (pc and symbian) if i find a working solution.

Thank you very much by now!

Re: Same encryption using c++ symbian & C# .net

alh — Wed, 06 Dec 2006 12:40:48 +0100

Rijndael is a block crypto, with 32 byte blocks, that is your "limitation".

Seems your example implementation only handles one block, and you have to do the chaining yourself.

If you encrypt just one block with either pc or symbian version and same key and init vector, your result should be decryptable on both sides.

My guess is that the "managed" version not outputs just the block but some extra header information or such.

Re: Same encryption using c++ symbian & C# .net

rbrunner — Tue, 05 Dec 2006 19:45:57 +0100

How do you transport the data? Is the data binary and the method of transport that you use safe for binary data? It might be that it is not, that one way (from PC to phone) works purely by chance, and the other way (from phone to PC) does not.

Anyway, what's your application? The application usually decides which kind of encryption (weak, strong, very strong) is needed.

Re: Same encryption using c++ symbian & C# .net

MOGi — Tue, 05 Dec 2006 17:57:38 +0100

I have not studied the issue. But one thing came to my mind. Can it not be due to different endianness on PC and Symbian HW?

Same encryption using c++ symbian & C# .net

Brekko74 — Tue, 05 Dec 2006 15:06:51 +0100