Re: AEAD Question

["Juan Pablo Nariño Mendoza" <jnarino@gmail.com>]

Hello Stephan
Thank you for your fast answer

My idea is to do zero copy encryption and the cipher I am using is
authenc(hmac(sha1),cbc(aes).

The layout of my buffer is

espHeader(AD) || IV || plaintext || Integrity (TAG)

As I see it, the SGs do point to the same buffer, but they do not overlap

Below, some parts of my code


// Prepare data
sg_init_table(sg,ARRAY_SIZE(sg));


sg_set_buf(&sg[1],cipherText_p,espHdrLen);
sg_set_buf(&sg[0],cipherText_p+espHdrLen+ivLen,plen+intLen);

// Put plaintext in cipherText + espHdr + ivLen
// put esp header info in ciphertext

req_p = aead_givcrypt_alloc(aead_p,GFP_KERNEL);
aead_givcrypt_set_callback(req_p,0,hwencpoc_testCipher_cb,cipherText_p);
aead_givcrypt_set_crypt(req_p,&sg[0],&sg[0],plen,iv_p);
aead_givcrypt_set_assoc(req_p,&sg[1],espHdrLen);
aead_givcrypt_set_giv(req_p,cipherText_p+espHdrLen,0);

ret=crypto_aead_givencrypt(req_p);


However, when I get the callback, the associated data has been
overwritten (first 8 bytes of the cipherText buffer), and the
integrity is wrong. The IV and the encryption are correct. What I am
doing wrong? Thank you again

Juan

2016-10-26 18:32 GMT+02:00 Stephan Mueller <smueller@chronox.de>:
> Am Mittwoch, 26. Oktober 2016, 18:17:14 CEST schrieb Juan Pablo Nariño
> Mendoza:
>
> Hi Juan,
>
>> Sorry in advance for making what shall be a basic question for this
>> list, but I have really ran out of ideas.
>>
>> Can someone explain me please, how does the memory layout and sg lists
>> work for the IPSec case, with generated IV case?
>>
>> As I understand, it is like this (ascii art)
>>
>> AD | IV Space | Plain Text | Auth Tag
>>  ^                   ^
>>
>> asg              sg
>
> The memory layout is correct when using the RFC4106 cipher.
>
> How you partition your physical memory with the SGL is your choice as long as
> your data in the SGL is lined up as you mentioned above.
>>
>> But I am having the AD overwritten, and the AUTH tag and encryption is
>> not correct? Could someone please point me in the right direction?
>> Thank you again
>
> Maybe the src and dst SGL somehow partially overlap?
>>
>> Regards
>>
>> Juan
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-crypto" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
> Ciao
> Stephan