ECDH secrets generated by BouncyCastle Java API and by OpenSSL are different

Question

I'm trying to make use of elliptic curve crypto. I need two implementations of the same thing, one in Java and one in C. I'm testing them using two key pairs which were generated using the curve secp256k1. When I generate the derived secret in Java I always get a different number from what I get from OpenSSL.

Java code:

/* privateKey and peerPublicKey are generated with the following parameters */
ECParameterSpec paramSpec = ECNamedCurveTable.getParameterSpec("secp256k1");
/* ... */
Provider BC = new BouncyCastleProvider();
KeyAgreement agr = KeyAgreement.getInstance("ECDH", BC);
agr.init(privateKey);
agr.doPhase(peerPublicKey, true);
byte[] secret = agr.generateSecret();

C code

/* pkey and peerkey are generated using EC_KEY_new_by_curve_name(NID_secp256k1) */
/* and than wrapped in an EVP_PKEY */
EVP_PKEY_CTX *ctx = EVP_PKEY_CTX_new(pkey, NULL);
uint8_t *secret = NULL;
size_t secret_len;
EVP_PKEY_derive_init(ctx);
EVP_PKEY_derive_set_peer(ctx, peerkey);
EVP_PKEY_derive(ctx, NULL, &secret_len);
secret = malloc(secret_len);
EVP_PKEY_derive(ctx, secret, &secret_len);

I'm sure that the keys are valid and that they are the same both in C and in Java code, but I don't understand why the derived secret is different. Am I missing something?

Thanks

Answer 1

EVP_PKEY_CTX *ctx = EVP_PKEY_CTX_new(pkey, NULL);
uint8_t *secret = NULL;
size_t secret_len;
EVP_PKEY_derive_init(ctx);
EVP_PKEY_derive_set_peer(ctx, peerkey);
EVP_PKEY_derive(ctx, NULL, &secret_len);
secret = malloc(secret_len);
EVP_PKEY_derive(ctx, secret, &secret_len);

This code looks like its missing a few steps. For example, EVP_PKEY_paramgen_init is not present.

The OpenSSL wiki has an example at Elliptic Curve Diffie-Hellman. I'm going to copy/paste it below to avoid the link-only answer, but I believe its the work of Matt Caswell.

EVP_PKEY_CTX *pctx, *kctx;
EVP_PKEY_CTX *ctx;
unsigned char *secret;
EVP_PKEY *pkey = NULL, *peerkey, *params = NULL;

/* Create the context for parameter generation */
if(NULL == (pctx = EVP_PKEY_CTX_new_id(EVP_PKEY_EC, NULL))) handleErrors();

/* Initialise the parameter generation */
if(1 != EVP_PKEY_paramgen_init(pctx)) handleErrors();

/* We're going to use the ANSI X9.62 Prime 256v1 curve */
if(1 != EVP_PKEY_CTX_set_ec_paramgen_curve_nid(pctx, NID_X9_62_prime256v1)) handleErrors();

/* Create the parameter object params */
if (!EVP_PKEY_paramgen(pctx, &params)) handleErrors();

/* Create the context for the key generation */
if(NULL == (kctx = EVP_PKEY_CTX_new(params, NULL))) handleErrors();

/* Generate the key */
if(1 != EVP_PKEY_keygen_init(kctx)) handleErrors();
if (1 != EVP_PKEY_keygen(kctx, &pkey)) handleErrors();

/* Get the peer's public key, and provide the peer with our public key -
 * how this is done will be specific to your circumstances */
peerkey = get_peerkey(pkey);

/* Create the context for the shared secret derivation */
if(NULL == (ctx = EVP_PKEY_CTX_new(pkey, NULL))) handleErrors();

/* Initialise */
if(1 != EVP_PKEY_derive_init(ctx)) handleErrors();

/* Provide the peer public key */
if(1 != EVP_PKEY_derive_set_peer(ctx, peerkey)) handleErrors();

/* Determine buffer length for shared secret */
if(1 != EVP_PKEY_derive(ctx, NULL, secret_len)) handleErrors();

/* Create the buffer */
if(NULL == (secret = OPENSSL_malloc(*secret_len))) handleErrors();

/* Derive the shared secret */
if(1 != (EVP_PKEY_derive(ctx, secret, secret_len))) handleErrors();

EVP_PKEY_CTX_free(ctx);
EVP_PKEY_free(peerkey);
EVP_PKEY_free(pkey);
EVP_PKEY_CTX_free(kctx);
EVP_PKEY_free(params);
EVP_PKEY_CTX_free(pctx);

/* Never use a derived secret directly. Typically it is passed
 * through some hash function to produce a key */
return secret;

---

When I generate the derived secret in Java I always get a different number from what I get from OpenSSL.

Each run of the protocol will produce different results. That's because each party picks a random value for each run of the protocol. That is, the a in g^a is random and different for each run, so the public key A = g^a is different for each run.

If everything is working correctly, you'll never see the parties use the same values, or one party to reuse a past value. Independent executions will never produce the same result. It does not matter if its OpenSSL ↔ OpenSSL, OpenSSL ↔ Java, or Java ↔ Java. They will always produce different results.