Re: [RFC 6/8] cert: add l_cert_pkcs5_pbkdf2_from_key_id

[Denis Kenzior <denkenz@gmail.com>]

Hi James,

On 11/18/22 15:16, James Prestwood wrote:
> The same pbkdf2 algorithm but uses a key ID as the password.
> ---
>   ell/cert-crypto.c | 27 +++++++++++++++++++++++++++
>   ell/cert.h        |  6 +++++-
>   ell/ell.sym       |  1 +
>   3 files changed, 33 insertions(+), 1 deletion(-)
> 

<snip>

> diff --git a/ell/cert.h b/ell/cert.h
> index f637588..ce430fa 100644
> --- a/ell/cert.h
> +++ b/ell/cert.h
> @@ -76,7 +76,11 @@ bool l_cert_pkcs5_pbkdf2(enum l_checksum_type type, const char *password,
>   				const uint8_t *salt, size_t salt_len,
>   				unsigned int iter_count,
>   				uint8_t *out_dk, size_t dk_len);
> -
> +bool l_cert_pkcs5_pbkdf2_from_key_id(enum l_checksum_type type,

missing LIB_EXPORT?

> +					int32_t key_id, const uint8_t *salt,
> +					size_t salt_len,
> +					unsigned int iter_count,
> +					uint8_t *out_dk, size_t dk_len);

So personally I'd rather have l_cert_pkcs5_pbkdf2 take a struct l_checksum * as 
the first parameter instead of creating two almost identical constructors. 
Especially since we already would have a special l_checksum_hmac_* constructor 
that takes a key id.

Regards,
-Denis