Re: [PATCH v3 1/3] crypto: qat - use pre-allocated buffers in datapath

[Giovanni Cabiddu <giovanni.cabiddu@intel.com>]

Belated response on this - I was out last week.

On Mon, Apr 11, 2022 at 05:37:24PM +0000, Eric Biggers wrote:
> On Sun, Apr 10, 2022 at 08:47:05PM +0100, Giovanni Cabiddu wrote:
> > If requests exceed 4 entries buffers, memory is allocated dynamically.
> > 
> > In addition, remove the CRYPTO_ALG_ALLOCATES_MEMORY flag from both aead
> > and skcipher alg structures.
> > 
> 
> There is nothing that says that algorithms can ignore
> !CRYPTO_ALG_ALLOCATES_MEMORY if there are too many scatterlist entries.  See the
> comment above the definition of CRYPTO_ALG_ALLOCATES_MEMORY.
From the conversation in [1], I assumed that a cap on the number of
pre-allocated entries in the scatterlists was already agreed.

> If you need to introduce this constraint, then you will need to audit the users
> of !CRYPTO_ALG_ALLOCATES_MEMORY to verify that none of them are issuing requests
> that violate this constraint, then add this to the documentation comment for
> CRYPTO_ALG_ALLOCATES_MEMORY.
Makes sense. I see that the only users of !CRYPTO_ALG_ALLOCATES_MEMORY
are dm-crypt and dm-integrity but I haven't done an audit on those yet
to understand if they use more than 4 entries.

Regards,

[1] https://lore.kernel.org/linux-crypto/20200722072932.GA27544@gondor.apana.org.au/

-- 
Giovanni