Re: [PATCH 01/15] crypto: skcipher - Add tailsize attribute

[Eric Biggers <ebiggers@kernel.org>]

On Sat, Dec 02, 2023 at 12:55:02PM +0800, Herbert Xu wrote:
> This patch adds a new tailsize attribute to skcipher and lskcipher
> algorithms.  This will be used by algorithms such as CTS which may
> need to withhold a number of blocks until the end has been reached.
> 
> When issuing a NOTFINAL request, the user must ensure that at least
> tailsize bytes will be supplied later on a final request.
> 
> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
> ---
>  crypto/lskcipher.c                 |  1 +
>  crypto/skcipher.c                  | 16 ++++++++++++++-
>  include/crypto/internal/skcipher.h |  1 +
>  include/crypto/skcipher.h          | 33 ++++++++++++++++++++++++++++++
>  4 files changed, 50 insertions(+), 1 deletion(-)
> 
> diff --git a/crypto/lskcipher.c b/crypto/lskcipher.c
> index 0b6dd8aa21f2..2a602911f4fc 100644
> --- a/crypto/lskcipher.c
> +++ b/crypto/lskcipher.c
> @@ -300,6 +300,7 @@ static void __maybe_unused crypto_lskcipher_show(
>  	seq_printf(m, "ivsize       : %u\n", skcipher->co.ivsize);
>  	seq_printf(m, "chunksize    : %u\n", skcipher->co.chunksize);
>  	seq_printf(m, "statesize    : %u\n", skcipher->co.statesize);
> +	seq_printf(m, "tailsize     : %u\n", skcipher->co.tailsize);

Do we really want to add new attributes like this to /proc/crypto?

I worry about userspace starting to depend on these algorithm attributes in a
weird way.

What is the use case for exposing them to userspace?

- Eric