[PATCH] crypto: ccree: fix xts-aes-ccree blocksize

[PATCH] crypto: ccree: fix xts-aes-ccree blocksize

[Corentin Labbe]

loading ccree on next-20220207 with CRYPTO_MANAGER_EXTRA_TESTS show a warning:
alg: skcipher: blocksize for xts-aes-ccree (1) doesn't match generic impl (16)
alg: self-tests for xts-aes-ccree (xts(aes)) failed (rc=-22)

After setting the correct blocksize, selftests pass.

Fixes: 67caef08a71f ("crypto: ccree - enable CTS support in AES-XTS")
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
---
 drivers/crypto/ccree/cc_cipher.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/crypto/ccree/cc_cipher.c b/drivers/crypto/ccree/cc_cipher.c
index 78833491f534..d955fe15cf40 100644
--- a/drivers/crypto/ccree/cc_cipher.c
+++ b/drivers/crypto/ccree/cc_cipher.c
@@ -1010,7 +1010,7 @@ static const struct cc_alg_template skcipher_algs[] = {
 	{
 		.name = "xts(paes)",
 		.driver_name = "xts-paes-ccree",
-		.blocksize = 1,
+		.blocksize = XTS_BLOCK_SIZE,
 		.template_skcipher = {
 			.setkey = cc_cipher_sethkey,
 			.encrypt = cc_cipher_encrypt,
@@ -1140,7 +1140,7 @@ static const struct cc_alg_template skcipher_algs[] = {
 		 */
 		.name = "xts(aes)",
 		.driver_name = "xts-aes-ccree",
-		.blocksize = 1,
+		.blocksize = XTS_BLOCK_SIZE,
 		.template_skcipher = {
 			.setkey = cc_cipher_setkey,
 			.encrypt = cc_cipher_encrypt,
-- 
2.34.1

Re: [PATCH] crypto: ccree: fix xts-aes-ccree blocksize

[Gilad Ben-Yossef]

Hi Again,

Thank you for taking the time to look into this!

However, this is an old topic that has been discussed before and the
answer really is that the selftests are wrong. They are looking at the
wrong thing. Yes, I know...

See the discussion here:
https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg40576.html

I also also point out this is actually documented in the code:

+               /* See
https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg40576.html
+                * for the reason why this differs from the generic
+                * implementation.
+                */

Thanks again!
Gilad


On Wed, Feb 9, 2022 at 9:06 AM Corentin Labbe <clabbe@baylibre.com> wrote:
>
> loading ccree on next-20220207 with CRYPTO_MANAGER_EXTRA_TESTS show a warning:
> alg: skcipher: blocksize for xts-aes-ccree (1) doesn't match generic impl (16)
> alg: self-tests for xts-aes-ccree (xts(aes)) failed (rc=-22)
>
> After setting the correct blocksize, selftests pass.
>
> Fixes: 67caef08a71f ("crypto: ccree - enable CTS support in AES-XTS")
> Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
> ---
>  drivers/crypto/ccree/cc_cipher.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/crypto/ccree/cc_cipher.c b/drivers/crypto/ccree/cc_cipher.c
> index 78833491f534..d955fe15cf40 100644
> --- a/drivers/crypto/ccree/cc_cipher.c
> +++ b/drivers/crypto/ccree/cc_cipher.c
> @@ -1010,7 +1010,7 @@ static const struct cc_alg_template skcipher_algs[] = {
>         {
>                 .name = "xts(paes)",
>                 .driver_name = "xts-paes-ccree",
> -               .blocksize = 1,
> +               .blocksize = XTS_BLOCK_SIZE,
>                 .template_skcipher = {
>                         .setkey = cc_cipher_sethkey,
>                         .encrypt = cc_cipher_encrypt,
> @@ -1140,7 +1140,7 @@ static const struct cc_alg_template skcipher_algs[] = {
>                  */
>                 .name = "xts(aes)",
>                 .driver_name = "xts-aes-ccree",
> -               .blocksize = 1,
> +               .blocksize = XTS_BLOCK_SIZE,
>                 .template_skcipher = {
>                         .setkey = cc_cipher_setkey,
>                         .encrypt = cc_cipher_encrypt,
> --
> 2.34.1
>


-- 
Gilad Ben-Yossef
Chief Coffee Drinker

values of β will give rise to dom!

Re: [PATCH] crypto: ccree: fix xts-aes-ccree blocksize

[Herbert Xu]

On Thu, Feb 17, 2022 at 09:36:41PM +0200, Gilad Ben-Yossef wrote:
> Hi Again,
> 
> Thank you for taking the time to look into this!
> 
> However, this is an old topic that has been discussed before and the
> answer really is that the selftests are wrong. They are looking at the
> wrong thing. Yes, I know...
> 
> See the discussion here:
> https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg40576.html
> 
> I also also point out this is actually documented in the code:
> 
> +               /* See
> https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg40576.html
> +                * for the reason why this differs from the generic
> +                * implementation.
> +                */

Indeed.  We should instead change the generic algorithm as well
as the other drivers that implement XTS.

Thanks,
-- 
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt