From: Marek Vasut <marex@denx.de>
Subject: Re: [PATCH 3/3] crypto: Add Allwinner Security System crypto accelerator
Date: Wed, 23 Jul 2014 21:38:35 +0200
Message-ID: <201407232138.35268.marex@denx.de>
References: <1400771396-9686-1-git-send-email-clabbe.montjoie@gmail.com> <201407231751.02050.marex@denx.de> <53D0045C.6090605@gmail.com>
Mime-Version: 1.0
Content-Type: Text/Plain; charset=iso-8859-1
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: Herbert Xu <herbert@gondor.apana.org.au>, robh+dt@kernel.org,
	pawel.moll@arm.com, mark.rutland@arm.com,
	ijc+devicetree@hellion.org.uk, galak@codeaurora.org,
	rdunlap@infradead.org, maxime.ripard@free-electrons.com,
	linux@arm.linux.org.uk, davem@davemloft.net,
	grant.likely@linaro.org, devicetree@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
	linux-crypto@vger.kernel.org
To: Corentin LABBE <clabbe.montjoie@gmail.com>
In-Reply-To: <53D0045C.6090605@gmail.com>
Sender: linux-crypto-owner@vger.kernel.org

On Wednesday, July 23, 2014 at 08:52:12 PM, Corentin LABBE wrote:
> Le 23/07/2014 17:51, Marek Vasut a =E9crit :
> > On Wednesday, July 23, 2014 at 04:13:09 PM, Herbert Xu wrote:
> >> On Wed, Jul 23, 2014 at 04:07:20PM +0200, Marek Vasut wrote:
> >>> On Wednesday, July 23, 2014 at 03:57:35 PM, Herbert Xu wrote:
> >>>> On Sat, May 24, 2014 at 02:00:03PM +0200, Marek Vasut wrote:
> >>>>>> +	}
> >>>>>> +#endif
> >>>>>> +
> >>>>>> +#ifdef CONFIG_CRYPTO_DEV_SUNXI_SS_MD5
> >>>>>> +	err =3D crypto_register_shash(&sunxi_md5_alg);
> >>>>>=20
> >>>>> Do not use shash for such device. This is clearly and ahash (an=
d
> >>>>> async in general) device. The rule of a thumb here is that you =
use
> >>>>> sync algos only for devices which have dedicated instructions f=
or
> >>>>> computing the transformation. For devices which are attached to=
 some
> >>>>> kind of bus, you use async algos (ahash etc).
> >>>>=20
> >>>> I'm sorry that I didn't catch this earlier but there is no such
> >>>> rule.
> >>>>=20
> >>>> Unless you need the async interface you should stick to the sync
> >>>> interfaces for the sake of simplicity.
> >>>>=20
> >>>> We have a number of existing drivers that are synchronous but
> >>>> using the async interface.  They should either be converted
> >>>> over to the sync interface or made interrupt-driven if possible.
> >>>=20
> >>> Sure, but this device is interrupt driven and uses DMA to feed th=
e
> >>> crypto engine, therefore async, right ?
> >>=20
> >> If it's interrupt-driven, then yes it would certainly make sense t=
o
> >> be async.  But all I see is polling in the latest posting, was the
> >> first version different?
> >=20
> > I stand corrected then, sorry.
> >=20
> > Is it possible to use DMA to feed the crypto accelerator, Corentin?
> >=20
> > Best regards,
> > Marek Vasut
>=20
> Yes, DMA is possible and will be implemented soon.
> So if I have well understood, I keep using async interface.

Yeah, in this case, using DMA and async interface combo is the way to g=
o.

Best regards,
Marek Vasut