Re: [Qemu-devel] [PATCH for-2.9 1/2] crypto: add 3des-ede support when using libgcrypt/nettle

[Longpeng (Mike)]

On 2016/12/7 17:19, Longpeng (Mike) wrote:

> 
> 
> On 2016/12/7 16:52, Gonglei (Arei) wrote:
> 
>>> -----Original Message-----
>>> From: longpeng
>>> Sent: Wednesday, December 07, 2016 4:34 PM
>>> To: address@hidden; address@hidden; address@hidden;
>>> Gonglei (Arei)
>>> Cc: address@hidden; Wubin (H); Zhoujian (jay, Euler); longpeng
>>> Subject: [PATCH for-2.9 1/2] crypto: add 3des-ede support when using
>>> libgcrypt/nettle
>>>
>>> Libgcrypt and nettle support 3des-ede, so this patch add 3des-ede
>>> support when using libgcrypt or nettle.
>>>
>>> Signed-off-by: Longpeng(Mike) <address@hidden>
>>> ---
> ......
>>> --- a/qapi/crypto.json
>>> +++ b/qapi/crypto.json
>>> @@ -63,6 +63,7 @@
>>>  # @aes-192: AES with 192 bit / 24 byte keys
>>>  # @aes-256: AES with 256 bit / 32 byte keys
>>>  # @des-rfb: RFB specific variant of single DES. Do not use except in VNC.
>>> +# @3des-ede: 3DES-EDE with 192 bit / 24 byte keys
>>
>> Missing since 2.9 here.
>>
> 
> Okay, I will add it in V2.
> 
>>>  # @cast5-128: Cast5 with 128 bit / 16 byte keys
>>>  # @serpent-128: Serpent with 128 bit / 16 byte keys
>>>  # @serpent-192: Serpent with 192 bit / 24 byte keys
> ......
>>> +    {
>>> +        /* Borrowed from linux-kernel crypto/testmgr.h */
>>> +        .path = "/crypto/cipher/3des-ede-ecb",
>>> +        .alg = QCRYPTO_CIPHER_ALG_3DES_EDE,
>>> +        .mode = QCRYPTO_CIPHER_MODE_ECB,
>>> +        .key =
>>> +            "0123456789abcdef5555555555555555"
>>> +            "fedcba9876543210",
>>> +        .plaintext =
>>> +            "736f6d6564617461",
>>> +        .ciphertext =
>>> +            "18d748e563620572",
>>> +    },
>>> +#endif
>>
>> Pls adds the CTR mode test case as well.
>>
> 
> Okay!
> 
> But I confuse with the 3DES's iv length now. I saw that cbc(3des-ede)'s iv-len
> can be 8 or 16 bytes, and ctr(3des-ede)'s iv-len is 24 bytes(according to
> linux-kernel crypto/testmgr.h).
> 

s/24/16

> However, I think qcrypto_cipher_get_iv_len() cannot distinguish a cipher alg
> with more than one iv-len.
> 
> Daniel, do you have any idea ?
> 
>> You'd better split the patch set into two patches so that Denial can take 
>> this one.
>> and I can take the patch 2.
>>
> 
> Okay, thanks.
> 
>> Thanks for your work.
>>
>> -Gonglei
>>
>>>      {
>>>          /* RFC 2144, Appendix B.1 */
>>>          .path = "/crypto/cipher/cast5-128",
>>> --
>>> 1.8.3.1
>>>
>>
>>
>> .
>>
> 
> 


-- 
Regards,
Longpeng(Mike)