Encryption output buffer description in algif_aead.c file

Encryption output buffer description in algif_aead.c file

[Harsh Jain]

Hi Stephan,

It seems now AEAD encryption operation also returns AAD to user space
in output buffer. Following comment in aead_recvmsg() needs updation:

/*
373          * AEAD memory structure: For encryption, the tag is appended to the
374          * ciphertext which implies that the memory allocated for
the ciphertext
375          * must be increased by the tag length. For decryption, the tag
376          * is expected to be concatenated to the ciphertext. The plaintext
377          * therefore has a memory size of the ciphertext minus the
tag length.
378          *
379          * The memory structure for cipher operation has the following
380          * structure:
381          *      AEAD encryption input:  assoc data || plaintext
382          *      AEAD encryption output: cipherntext || auth tag
383          *      AEAD decryption input:  assoc data || ciphertext || auth tag
384          *      AEAD decryption output: plaintext
385          */


Regards
Harsh Jain

Re: Encryption output buffer description in algif_aead.c file

[Stephan Mueller]

Am Freitag, 24. Juni 2016, 17:24:02 schrieb Harsh Jain:

Hi Harsh,


> 379          * The memory structure for cipher operation has the following
> 380          * structure:
> 381          *      AEAD encryption input:  assoc data || plaintext
> 382          *      AEAD encryption output: cipherntext || auth tag
> 383          *      AEAD decryption input:  assoc data || ciphertext || auth
> tag 384          *      AEAD decryption output: plaintext

Right, it returns AAD prepended to the stated output. Do you want to provide a 
patch?

Ciao
Stephan

Re: Encryption output buffer description in algif_aead.c file

[Harsh Jain]

Yes, I will share the patch.

regards
Harsh Jain



On Fri, Jun 24, 2016 at 5:31 PM, Stephan Mueller <smueller@chronox.de> wrote:
> Am Freitag, 24. Juni 2016, 17:24:02 schrieb Harsh Jain:
>
> Hi Harsh,
>
>
>> 379          * The memory structure for cipher operation has the following
>> 380          * structure:
>> 381          *      AEAD encryption input:  assoc data || plaintext
>> 382          *      AEAD encryption output: cipherntext || auth tag
>> 383          *      AEAD decryption input:  assoc data || ciphertext || auth
>> tag 384          *      AEAD decryption output: plaintext
>
> Right, it returns AAD prepended to the stated output. Do you want to provide a
> patch?
>
> Ciao
> Stephan

Re: Encryption output buffer description in algif_aead.c file

[Gary R Hook]

On 06/24/2016 07:01 AM, Stephan Mueller wrote:
> Am Freitag, 24. Juni 2016, 17:24:02 schrieb Harsh Jain:
>
> Hi Harsh,
>
>
>> 379          * The memory structure for cipher operation has the following
>> 380          * structure:
>> 381          *      AEAD encryption input:  assoc data || plaintext
>> 382          *      AEAD encryption output: cipherntext || auth tag
>> 383          *      AEAD decryption input:  assoc data || ciphertext || auth
>> tag 384          *      AEAD decryption output: plaintext
>
> Right, it returns AAD prepended to the stated output. Do you want to provide a
> patch?

If testmgr.c is any model, the caller is expected to populate the 
destination
buffer with the AAD. Is my understanding correct? And should this 
comment clarify
that point: i.e. the length of the destination is the sum of the lengths 
of the
aad + ciphertext + tag?

Gary

Re: Encryption output buffer description in algif_aead.c file

[Stephan Mueller]

Am Freitag, 24. Juni 2016, 07:44:49 schrieb Gary R Hook:

Hi Gary,

> On 06/24/2016 07:01 AM, Stephan Mueller wrote:
> > Am Freitag, 24. Juni 2016, 17:24:02 schrieb Harsh Jain:
> > 
> > Hi Harsh,
> > 
> >> 379          * The memory structure for cipher operation has the
> >> following
> >> 380          * structure:
> >> 381          *      AEAD encryption input:  assoc data || plaintext
> >> 382          *      AEAD encryption output: cipherntext || auth tag
> >> 383          *      AEAD decryption input:  assoc data || ciphertext ||
> >> auth tag 384          *      AEAD decryption output: plaintext
> > 
> > Right, it returns AAD prepended to the stated output. Do you want to
> > provide a patch?
> 
> If testmgr.c is any model, the caller is expected to populate the
> destination
> buffer with the AAD. Is my understanding correct? And should this
> comment clarify
> that point: i.e. the length of the destination is the sum of the lengths
> of the
> aad + ciphertext + tag?

It may make sense if you would look at libkcapi which handles the input/output 
appropriately. Especially, the kcapi_aead_getdata() function sets up the right 
pointers.

It is correct that the AAD data is taken from the input.

Ciao
Stephan