Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752828AbeAJTO7 (ORCPT + 1 other); Wed, 10 Jan 2018 14:14:59 -0500 Received: from mail-wr0-f179.google.com ([209.85.128.179]:39432 "EHLO mail-wr0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752493AbeAJTOz (ORCPT ); Wed, 10 Jan 2018 14:14:55 -0500 X-Google-Smtp-Source: ACJfBotO/NZultfbK9iDq8klJP/kqex+YYhljFhBnjlmKymqCKdKmBtbi+o0vOIlwQ3Iqspz2hkiQQ== Date: Wed, 10 Jan 2018 20:14:50 +0100 From: Corentin Labbe To: Fabien DESSENNE Cc: Alexandre TORGUE , "arei.gonglei@huawei.com" , "corbet@lwn.net" , "davem@davemloft.net" , "herbert@gondor.apana.org.au" , "jasowang@redhat.com" , "mcoquelin.stm32@gmail.com" , "mst@redhat.com" , "linux-arm-kernel@lists.infradead.org" , "linux-crypto@vger.kernel.org" , "linux-doc@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "virtualization@lists.linux-foundation.org" , Benjamin GAIGNARD Subject: Re: [PATCH 1/6] Documentation: crypto: document crypto engine API Message-ID: <20180110191450.GB12214@Red> References: <20180103201109.16077-1-clabbe.montjoie@gmail.com> <20180103201109.16077-2-clabbe.montjoie@gmail.com> <849600f6-3082-c090-205d-75e3d14b38f9@st.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <849600f6-3082-c090-205d-75e3d14b38f9@st.com> User-Agent: Mutt/1.7.2 (2016-11-26) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: On Wed, Jan 10, 2018 at 02:13:13PM +0000, Fabien DESSENNE wrote: > Hi Corentin, > > > Thank you for this new version which I have testes successfully with the > stm32 hash & cryp drivers. > > As a general comment on this patchset, I would say that it does not > cover all async requests: typically I need (for the pending stm32 cryp > driver uprade) to use CryptoEngine to process AEAD requests which is not > covered here. > > Could you please consider adding the 'transfer' and 'finalize' EXPORTed > functions for aead requests? (the implementation is quite trivial) > > Have also a look at struct acomp_req (acompress.h) and struct > kpp_request (kpp.h) which also use "struct crypto_async_request base" > > > BR > > Fabien > Hello Thanks for your review and test (could I add your tested-by ?). I didn't add aead (and kpp/acompress), since I do not have any way to test it. Since you have a way to test aead, I will add it to the next release. Regards > > On 03/01/18 21:11, Corentin Labbe wrote: > > Signed-off-by: Corentin Labbe > > --- > > Documentation/crypto/crypto_engine.rst | 46 ++++++++++++++++++++++++++++++++++ > > 1 file changed, 46 insertions(+) > > create mode 100644 Documentation/crypto/crypto_engine.rst > > > > diff --git a/Documentation/crypto/crypto_engine.rst b/Documentation/crypto/crypto_engine.rst > > new file mode 100644 > > index 000000000000..b0ed37f9fb0c > > --- /dev/null > > +++ b/Documentation/crypto/crypto_engine.rst > > @@ -0,0 +1,46 @@ > > +============= > > +CRYPTO ENGINE > > +============= > > + > > +Overview > > +-------- > > +The crypto engine API (CE), is a crypto queue manager. > > + > > +Requirement > > +----------- > > +You have to put at start of your tfm_ctx the struct crypto_engine_reqctx > > +struct your_tfm_ctx { > > + struct crypto_engine_reqctx enginectx; > > + ... > > +}; > > +Why: Since CE manage only crypto_async_request, it cannot know the underlying > > +request_type and so have access only on the TFM. > > +So using container_of for accessing __ctx is impossible. > > +Furthermore, the crypto engine cannot know the "struct your_tfm_ctx", > > +so it must assume that crypto_engine_reqctx is at start of it. > > + > > +Order of operations > > +------------------- > > +You have to obtain a struct crypto_engine via crypto_engine_alloc_init(). > > +And start it via crypto_engine_start(). > > + > > +Before transferring any request, you have to fill the enginectx. > > +- prepare_request: (taking a function pointer) If you need to do some processing before doing the request > > +- unprepare_request: (taking a function pointer) Undoing what's done in prepare_request > > +- do_one_request: (taking a function pointer) Do encryption for current request > > + > > +Note: that those three functions get the crypto_async_request associated with the received request. > > +So your need to get the original request via container_of(areq, struct yourrequesttype_request, base); > > + > > +When your driver receive a crypto_request, you have to transfer it to > > +the cryptoengine via one of: > > +- crypto_transfer_cipher_request_to_engine() > > +- crypto_transfer_skcipher_request_to_engine() > > +- crypto_transfer_akcipher_request_to_engine() > > +- crypto_transfer_hash_request_to_engine() > > + > > +At the end of the request process, a call to one of the following function is needed: > > +- crypto_finalize_cipher_request > > +- crypto_finalize_skcipher_request > > +- crypto_finalize_akcipher_request > > +- crypto_finalize_hash_request