Open Stack

On Thu, Jun 5, 2014 at 1:26 PM, Jeff Feng <jhfeng1 at gmail.com> wrote:
> AES block size is 16 bytes. I don't understand why the padding size would >
> 16.
> cipher.block_size should equal to 16.
>
They might be trying to hide the length of the plain text. But its not
readily obvious since there does not appear to be design documentation
or comments in the source code.

While its unusual, I'm not sure its incorrect in the context of
cryptography. See, for example, "On Hiding a Plaintext Length by
Preencryption",
http://cihangir.forgottenlance.com/papers/lengthhiding-corrected.pdf.

Jeff

> On 6/5/2014 12:06 PM, Robert Clark wrote:
>>
>> In the case of 128 bit AES would this not send twice as much padding as
>> needed or is the schema such that padding is always to the nearest 256 ?
>>
>>> -----Original Message-----
>>> From: bounces at canonical.com [mailto:bounces at canonical.com] On
>>> Behalf Of Robert Clark
>>> Sent: 04 June 2014 23:11
>>> To: Clark, Robert Graham
>>> Subject: [Bug 1326474] Re: crypto/utils.py may use too much padding
>>>
>>> Thanks for the clarification guys, this makes sense to me, fwiw I did
>>> ask
>>> about this in the oslo room first and the suggestion was that I log a
>>> bug to
>>> get it double checked.
>>>
>>> Cheers
>>> -Rob
>>>
>>> ** Changed in: oslo
>>>         Status: New => Invalid
>>>
>>> --
>>> You received this bug notification because you are subscribed to the bug
>>> report.
>>> https://bugs.launchpad.net/bugs/1326474
>>>
>>> Title:
>>>    crypto/utils.py may use too much padding
>>>
>>> Status in Oslo - a Library of Common OpenStack Code:
>>>    Invalid
>>>
>>> Bug description:
>>>    I've been reviewing some of the crypto code available in OpenStack and
>>>    noticed something interesting in the padding of
>>>    common/crypto/utils.py.
>>>
>>>
>>>    If the message length is the same as the cipher block size an entire
>>> extra
>>> block of padding is sent along with the message, I'm not sure if this is
>>> desired behaviour (if it is, this bug is invalid) but it certainly
>>> doesn't seem
>>> quite right.
>>>
>>>    If the message length is ever the same as the boundary size ( % 256)
>>>    then an entirely extra block of padding will be applied:
>>>
>>>    ---code---
>>>       r = len(msg) % self.cipher.block_size
>>>            padlen = self.cipher.block_size - r - 1
>>>            msg += b'\x00' * padlen
>>>            msg += bchr(padlen)
>>>    ---/code---
>>>
>>>    So if our msg length is 256,512,,, then 'r' will be 0
>>>    padlen will be 256-0-1 or 255
>>>    msg gets 255 * b'\x00' added and then the number 255 tagged on the
>>> end.
>>>
>>> To manage notifications about this bug go to:
>>> https://bugs.launchpad.net/oslo/+bug/1326474/+subscriptions