> -----Original Message----- > From: Antoine Tenart <antoine.ten...@bootlin.com> > Sent: Wednesday, September 11, 2019 5:30 PM > To: Pascal van Leeuwen <pascalv...@gmail.com> > Cc: linux-crypto@vger.kernel.org; antoine.ten...@bootlin.com; > herb...@gondor.apana.org.au; da...@davemloft.net; Pascal Van Leeuwen > <pvanleeu...@verimatrix.com> > Subject: Re: [PATCH 2/2] crypto: inside-secure - Add support for the > Chacha20-Poly1305 > AEAD > > Hello Pascal, > > On Tue, Sep 10, 2019 at 04:38:13PM +0200, Pascal van Leeuwen wrote: > > @@ -43,8 +44,8 @@ struct safexcel_cipher_ctx { > > > > u32 mode; > > enum safexcel_cipher_alg alg; > > - bool aead; > > - int xcm; /* 0=authenc, 1=GCM, 2 reserved for CCM */ > > + char aead; /* !=0=AEAD, 2=IPSec ESP AEAD */ > > + char xcm; /* 0=authenc, 1=GCM, 2 reserved for CCM */ > > You could use an u8 instead. It also seems the aead comment has an > issue, I'll let you check that. > Yes, u8 would be better, I'll fix that. I don't see what's wrong with the comment though? Anything unequal to 0 is AEAD, with value 2 being the ESP variant.
> > - dev_err(priv->dev, "aead: unsupported hash algorithm\n"); > > + dev_err(priv->dev, "aead: unsupported hash algorithmn"); > > You remove the '\' here. > Oops. Must've accidentally happended during editing. Good catch. > > @@ -440,6 +459,17 @@ static int safexcel_context_control(struct > > safexcel_cipher_ctx > *ctx, > > CONTEXT_CONTROL_DIGEST_XCM | > > ctx->hash_alg | > > CONTEXT_CONTROL_SIZE(ctrl_size); > > + } else if (ctx->alg == SAFEXCEL_CHACHA20) { > > + /* Chacha20-Poly1305 */ > > + cdesc->control_data.control0 = > > + CONTEXT_CONTROL_KEY_EN | > > + CONTEXT_CONTROL_CRYPTO_ALG_CHACHA20 | > > + (sreq->direction == SAFEXCEL_ENCRYPT ? > > + CONTEXT_CONTROL_TYPE_ENCRYPT_HASH_OUT : > > + CONTEXT_CONTROL_TYPE_HASH_DECRYPT_IN) | > > + ctx->hash_alg | > > + CONTEXT_CONTROL_SIZE(ctrl_size); > > I think you could use an if + |= for readability here. > Ok, I can do that > > +static int safexcel_aead_chachapoly_crypt(struct aead_request *req, > > + enum safexcel_cipher_direction dir) > > +{ > > + struct safexcel_cipher_req *creq = aead_request_ctx(req); > > + struct crypto_aead *aead = crypto_aead_reqtfm(req); > > + struct crypto_tfm *tfm = crypto_aead_tfm(aead); > > + struct safexcel_cipher_ctx *ctx = crypto_tfm_ctx(tfm); > > + struct aead_request *subreq = aead_request_ctx(req); > > + u32 key[CHACHA_KEY_SIZE / sizeof(u32) + 1]; > > Shouldn't you explicitly memzero the key at the end of the function? > Yes! :-) And good catch, again. > Thanks! > Antoine > > -- > Antoine Ténart, Bootlin > Embedded Linux and Kernel engineering > https://bootlin.com Thanks, Pascal van Leeuwen Silicon IP Architect, Multi-Protocol Engines @ Verimatrix www.insidesecure.com