Search code examples
copensslcryptographyrsalibcrypto

Difference between EVP_PKEY_sign and EVP_DigestSignInit?

So far I've seen two different approaches to RSA signing with OpenSSL:
With EVP_PKEY_sign

 ctx = EVP_PKEY_CTX_new(signing_key, NULL /* no engine */);
 if (!ctx)
     /* Error occurred */
 if (EVP_PKEY_sign_init(ctx) <= 0)
     /* Error */
 if (EVP_PKEY_CTX_set_rsa_padding(ctx, RSA_PKCS1_PADDING) <= 0)
     /* Error */
 if (EVP_PKEY_CTX_set_signature_md(ctx, EVP_sha256()) <= 0)
     /* Error */

 /* Determine buffer length */
 if (EVP_PKEY_sign(ctx, NULL, &siglen, md, mdlen) <= 0)

With EVP_DigestSignInit:

if(1 != EVP_DigestSignInit(mdctx, NULL, EVP_sha256(), NULL, key)) 
          goto err;
if(1 != EVP_DigestSignUpdate(mdctx, msg, strlen(msg))) 
          goto err;
if(1 != EVP_DigestSignFinal(mdctx, NULL, slen)) 
          goto err;
if(!(*sig = OPENSSL_malloc(sizeof(unsigned char) * (*slen)))) 
          goto err;
if(1 != EVP_DigestSignFinal(mdctx, *sig, slen)) 
          goto err;

Are these just two different ways to do the same thing?

Solution

  • Oh. There's a fairly major difference.

    EVP_PKEY_sign() does not hash the data to be signed, and therefore is normally used to sign digests. For signing arbitrary messages, see the EVP_DigestSignInit(3)

    Per https://wiki.openssl.org/index.php/Manual:EVP_PKEY_sign(3)

    So EVP_PKEY_sign is very likely used under the hood in EVP_Digest SignInit, and is intended for applications where the caller will manually format the block to be signed.