cfb mode for AES

[cadorn]

Are there plans to implement cfb mode for AES?

I am trying to move to forge exclusively and am missing this.

Currently using: https://github.com/openpeer/cifre/blob/master/aes.js (which has a bug with data > 15 chars)

[dlongley]

@cadorn -- I added CFB mode support and examples for using it. Let me know if there are any issues.

[dlongley]

@cadorn, while I was at it, I added CFB, OFB, and CTR cipher modes.

[cadorn]

@dlongley Wow. Thanks for the quick addition.

Is the AES encryption/decryption in CFB mode compatible with the following?:

AES (Rijndael- 128) in CFB mode with a 32 byte key size, 16 byte block size, 16 byte feedback size

[cadorn]

Given:

key: 861009ec4d599fab1f40abc76e6f89880cff5833c79c548c99f9045f191cd90b
iv: d927ad81199aa7dcadfdb4e47b6dc694
data: MY-DATA-AND-HERE-IS-MORE-DATA

After encryption I get:

80eb666a9fc9e263faf71e87ffc94451d7d8df7cfcf2606470351dd5ac3f70bd

While my existing JS implementation (cifre) and PHP implementation give me:

80eb666a9fc9e263faf71e87ffc94451d7d8df7cfcf2606470351dd5ac

Notice the extra 3f70bd at the end of the forge result.

If I decrypt the forge result with cifre I get:

MY-DATA-AND-HERE-IS-MORE-DATA

When I decrypt result from cifre with forge I get:

MY-DATA-AND-HERE

Any idea what the extra characters at the end are and how they can be removed from the forge result?

[dlongley]

It sounds like you're having trouble with the padding. There may or may not be a bug in forge related to this, but the padding shouldn't be affected by the cipher block mode (correction, the default padding should not be affected but stream mode shouldn't be using it). Here's some code below that seems to be working just fine for me:

var key = '861009ec4d599fab1f40abc76e6f89880cff5833c79c548c99f9045f191cd90b';
var iv = 'd927ad81199aa7dcadfdb4e47b6dc694';
var data = 'MY-DATA-AND-HERE-IS-MORE-DATA';

key = forge.util.hexToBytes(key);
iv = forge.util.hexToBytes(iv);

var cipher = forge.aes.createEncryptionCipher(key, 'CFB');
cipher.start(iv);
cipher.update(forge.util.createBuffer(data));
cipher.finish();
var encrypted = cipher.output;
console.log('encrypted: ' + encrypted.toHex());

cipher = forge.aes.createDecryptionCipher(key, 'CFB');
cipher.start(iv);
cipher.update(encrypted);
cipher.finish();
var decrypted = cipher.output.getBytes();
console.log('decrypted: ' + decrypted);

assert.equal(data, decrypted);

The above code prints this for me:

encrypted: 80eb666a9fc9e263faf71e87ffc94451d7d8df7cfcf2606470351dd5ac3f70bd
decrypted: MY-DATA-AND-HERE-IS-MORE-DATA

What are the differences with your code?

[dlongley]

Wow. Thanks for the quick addition.

Sure! It wasn't too difficult to add ... and I had been meaning to for quite a while so I figured that since someone else needed it I'd see if it could be done quickly.

Is the AES encryption/decryption in CFB mode compatible with the following?:

AES (Rijndael- 128) in CFB mode with a 32 byte key size, 16 byte block size, 16 byte feedback size

Yes.

[dlongley]

@cadorn, it looks like the cifre and PHP code you're using is not padding the output; you should always get multiples of the block size and that output is only 29 bytes long (58 in hex) when it should be 32.

[dlongley]

@cadorn, it looks like I need to change the default padding for CFB (none) -- let me look into that.

[dlongley]

@cadorn, padding should have been turned off for any cipher stream modes (the new ones just added) with automatic truncation rather than default PKCS#7 padding. This has been corrected in the latest commit.

[cadorn]

Great. Thanks!