From: Herbert Xu Subject: Re: [PATCH v6 1/4] crypto: AF_ALG: add AEAD support Date: Mon, 29 Dec 2014 21:33:19 +1100 Message-ID: <20141229103319.GB13334@gondor.apana.org.au> References: <5682082.ffPqvQlSqN@tachyon.chronox.de> <5002301.TQO37u96dE@tachyon.chronox.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Daniel Borkmann , 'Quentin Gouchet' , 'LKML' , linux-crypto-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Stephan Mueller Return-path: Content-Disposition: inline In-Reply-To: <5002301.TQO37u96dE-PJstQz4BMNNP20K/wil9xYQuADTiUCJX@public.gmane.org> Sender: linux-api-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-crypto.vger.kernel.org On Thu, Dec 25, 2014 at 11:01:47PM +0100, Stephan Mueller wrote: > > + err = -ENOMEM; This should be EINVAL. > + if (!aead_sufficient_data(ctx)) > + goto unlock; So we're checking two things here, one that we have enough data for AD and two we have the authentication tag. The latter is redundant as the underlying implementation should be able to cope with short input so we should only check the assoclen here. Also this check should be moved to the sendmsg side as that'll make it more obvious as to what went wrong. PS we should add a length check for missing/partial auth tags to crypto_aead_decrypt. We can then remove such checks from individual implementations. Thanks, -- Email: Herbert Xu Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt