Re: [PATCH v3 1/2] mac80211: fils_aead: Use crypto api CMAC shash rather than bare cipher

2017-02-07 Thread Johannes Berg 
On Wed, 2017-02-08 at 07:45 +, Ard Biesheuvel wrote:
> On 8 February 2017 at 07:00, Johannes Berg  > wrote:
> > This looks strange to me:
> > 
> > > +static int aes_s2v(struct crypto_shash *tfm,
> > >  size_t num_elem, const u8 *addr[], size_t len[],
> > > u8 *v)
> > >  {
> > > - u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE];
> > > + u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE] = {};
> > > + SHASH_DESC_ON_STACK(desc, tfm);
> > 
> > desc declared
> > 
> > > 
> > > + crypto_shash_digest(desc, tmp, AES_BLOCK_SIZE, d);
> > 
> > used here
> > 
> 
> Each digest() call combines a init()/update()/final() sequence
> 
> > > + crypto_shash_init(desc);
> > 
> > but initialized now?
> > 
> 
> ... for the 6th time, or so. The final vector may require two
> update()s, so we cannot use digest() here. But we can use finup() for
> the last one, which combines update() and final().
> 
> Hence,
> 
> init()/finup()
> 
> or
> 
> init()/update()/finup()
> 
> depending on the length of the last vector.

Great, thanks for the explanation :)

johannes

Re: [PATCH v3 1/2] mac80211: fils_aead: Use crypto api CMAC shash rather than bare cipher

2017-02-07 Thread Ard Biesheuvel 
On 8 February 2017 at 07:00, Johannes Berg  wrote:
> This looks strange to me:
>
>> +static int aes_s2v(struct crypto_shash *tfm,
>>  size_t num_elem, const u8 *addr[], size_t len[],
>> u8 *v)
>>  {
>> - u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE];
>> + u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE] = {};
>> + SHASH_DESC_ON_STACK(desc, tfm);
>
> desc declared
>
>>
>> + crypto_shash_digest(desc, tmp, AES_BLOCK_SIZE, d);
>
> used here
>

Each digest() call combines a init()/update()/final() sequence

>> + crypto_shash_init(desc);
>
> but initialized now?
>

... for the 6th time, or so. The final vector may require two
update()s, so we cannot use digest() here. But we can use finup() for
the last one, which combines update() and final().

Hence,

init()/finup()

or

init()/update()/finup()

depending on the length of the last vector.

Re: [PATCH v3 1/2] mac80211: fils_aead: Use crypto api CMAC shash rather than bare cipher

2017-02-07 Thread Johannes Berg 
This looks strange to me:

> +static int aes_s2v(struct crypto_shash *tfm,
>      size_t num_elem, const u8 *addr[], size_t len[],
> u8 *v)
>  {
> - u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE];
> + u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE] = {};
> + SHASH_DESC_ON_STACK(desc, tfm);

desc declared

> 
> + crypto_shash_digest(desc, tmp, AES_BLOCK_SIZE, d);

used here
 
> + crypto_shash_init(desc);

but initialized now?

johannes