Re: [EXT] Re: [PATCH] crypto: caam/hash - fix asynchronous hash

[Eric Biggers]

On Wed, Jan 17, 2024 at 10:48:39AM +0000, Gaurav Jain wrote:
> 
> 
> > -----Original Message-----
> > From: Eric Biggers <ebiggers@xxxxxxxxxx>
> > Sent: Wednesday, January 17, 2024 10:03 AM
> > To: Gaurav Jain <gaurav.jain@xxxxxxx>
> > Cc: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>; David S . Miller
> > <davem@xxxxxxxxxxxxx>; Horia Geanta <horia.geanta@xxxxxxx>; Pankaj
> > Gupta <pankaj.gupta@xxxxxxx>; Varun Sethi <V.Sethi@xxxxxxx>; Meenakshi
> > Aggarwal <meenakshi.aggarwal@xxxxxxx>; Aisheng Dong
> > <aisheng.dong@xxxxxxx>; Silvano Di Ninno <silvano.dininno@xxxxxxx>; linux-
> > crypto@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; dl-linux-imx <linux-
> > imx@xxxxxxx>
> > Subject: [EXT] Re: [PATCH] crypto: caam/hash - fix asynchronous hash
> > 
> > Caution: This is an external email. Please take care when clicking links or
> > opening attachments. When in doubt, report the message using the 'Report this
> > email' button
> > 
> > 
> > On Tue, Jan 16, 2024 at 03:14:05PM +0530, Gaurav Jain wrote:
> > > ahash_alg->setkey is updated to ahash_nosetkey in ahash.c so updating
> > > the handling of setkey in caam driver.
> > >
> > > Fixes: 2f1f34c1bf7b ("crypto: ahash - optimize performance when
> > > wrapping shash")
> > > Signed-off-by: Gaurav Jain <gaurav.jain@xxxxxxx>
> > > ---
> > >  drivers/crypto/caam/caamalg_qi2.c | 4 ++--
> > >  drivers/crypto/caam/caamhash.c    | 4 ++--
> > >  2 files changed, 4 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/drivers/crypto/caam/caamalg_qi2.c
> > > b/drivers/crypto/caam/caamalg_qi2.c
> > > index a148ff1f0872..93a400e286b4 100644
> > > --- a/drivers/crypto/caam/caamalg_qi2.c
> > > +++ b/drivers/crypto/caam/caamalg_qi2.c
> > > @@ -4571,7 +4571,7 @@ static int caam_hash_cra_init(struct crypto_tfm
> > > *tfm)
> > >
> > >       ctx->dev = caam_hash->dev;
> > >
> > > -     if (alg->setkey) {
> > > +     if (crypto_hash_alg_has_setkey(halg)) {
> > >               ctx->adata.key_dma = dma_map_single_attrs(ctx->dev, ctx->key,
> > >                                                         ARRAY_SIZE(ctx->key),
> > >                                                         DMA_TO_DEVICE,
> > > @@ -4611,7 +4611,7 @@ static int caam_hash_cra_init(struct crypto_tfm
> > *tfm)
> > >        * For keyed hash algorithms shared descriptors
> > >        * will be created later in setkey() callback
> > >        */
> > > -     return alg->setkey ? 0 : ahash_set_sh_desc(ahash);
> > > +     return crypto_hash_alg_has_setkey(halg) ? 0 :
> > > + ahash_set_sh_desc(ahash);
> > >  }
> > >
> > >  static void caam_hash_cra_exit(struct crypto_tfm *tfm) diff --git
> > > a/drivers/crypto/caam/caamhash.c b/drivers/crypto/caam/caamhash.c
> > > index 290c8500c247..4d50356b593c 100644
> > > --- a/drivers/crypto/caam/caamhash.c
> > > +++ b/drivers/crypto/caam/caamhash.c
> > > @@ -1804,7 +1804,7 @@ static int caam_hash_cra_init(struct crypto_tfm
> > *tfm)
> > >       } else {
> > >               if (priv->era >= 6) {
> > >                       ctx->dir = DMA_BIDIRECTIONAL;
> > > -                     ctx->key_dir = alg->setkey ? DMA_TO_DEVICE : DMA_NONE;
> > > +                     ctx->key_dir = crypto_hash_alg_has_setkey(halg)
> > > + ? DMA_TO_DEVICE : DMA_NONE;
> > >               } else {
> > >                       ctx->dir = DMA_TO_DEVICE;
> > >                       ctx->key_dir = DMA_NONE; @@ -1862,7 +1862,7 @@
> > > static int caam_hash_cra_init(struct crypto_tfm *tfm)
> > >        * For keyed hash algorithms shared descriptors
> > >        * will be created later in setkey() callback
> > >        */
> > > -     return alg->setkey ? 0 : ahash_set_sh_desc(ahash);
> > > +     return crypto_hash_alg_has_setkey(halg) ? 0 :
> > > + ahash_set_sh_desc(ahash);
> > >  }
> > >
> > 
> > Thanks.  Did you also consider putting something in struct caam_hash_alg (the
> > struct in which this driver embeds its ahash_alg structure) that indicates whether
> > the algorithm is an HMAC or not?  Other drivers use that solution.
> 
> Crypto/ahash.c has this API to check the setkey so I used to differentiate between HMAC & only hash.
> Let me know if this change is not sufficient, will add the flag in caam_hash_alg.
> 

Currently crypto_hash_alg_has_setkey() isn't used outside of crypto/ahash.c.
Since there's an alternative that matches what the other drivers do, I think I'd
prefer that you did that.  We can then make crypto_hash_alg_has_setkey() a
static function, private to crypto/ahash.c.

- Eric