Add XMSS Serialize/Deserialize

[ducnguyen-sb]

LMS

This PR slightly modify #1533

• Return OQS_STATUS
• Flatten control-flow

I tried to keep the same logic as possible. Please let me know if there is something wrong @ashman-p

XMSS

Add serialize and deserialize function to XMSS.

• Does this PR change the input/output behaviour of a cryptographic algorithm (i.e., does it change known answer test values)? (If so, a version bump will be required from x.y.z to x.(y+1).0.)
• Does this PR change the list of algorithms available -- either adding, removing, or renaming? Does this PR otherwise change an API? (If so, PRs in oqs-provider, OQS-OpenSSL, OQS-BoringSSL, and OQS-OpenSSH will also need to be ready for review and merge by the time this is merged.)

[ducnguyen-sb]

So I have this piece of code that we discuss in the meeting:
So we check if the address of array secret_key_data is either NULL or not. If it's not NULL, we assume it's already initialized, thus we return error.
To reach line 39, sk->secret_key_data must be NULL. Which make the memcpy is SEGFAULT.

I'm thinking of creating static once = 1 as static variable to make sure the initialization happen only once.
But this lead to:

• We will have 2 ways to allocate memory for sk->secret_key_data, either in deserialize, and in key generation.
• If I use static variable, I don't know in the case we multithread/fork process, would it become complicate to make sure the sk->secret_key_data happen only once with static variable.

[ducnguyen-sb]

After a while I change the logic of the code to:

/* Deserialize XMSS byte string into an XMSS secret key data */
OQS_STATUS OQS_SECRET_KEY_XMSS_deserialize_key(OQS_SIG_STFL_SECRET_KEY *sk, const size_t sk_len, const uint8_t *sk_buf) {
	if (sk == NULL || sk_buf == NULL || (sk_len != sk->length_secret_key)) {
		return OQS_ERROR;
	}

	if (sk->secret_key_data != NULL) {
		// Key data already present
		// We dont want to trample over data
		return OQS_ERROR;
	}

	// Assume key data is not present
	sk->secret_key_data = malloc(sk_len);
	if (sk->secret_key_data == NULL) {
		return OQS_ERROR;
	}

	memcpy(sk->secret_key_data, sk_buf, sk_len);

	return OQS_SUCCESS;
}

What do you think @ashman-p @dstebila ?

[ducnguyen-sb]

Committed in 7e4f2c9

[ashman-p]

This looks fine to me.
It might be good to walk through the use-case where it is necessary to call "deserialize' on a working secret key object. Those use-cases would help determine the desired behaviors for each expected outcome.

[ducnguyen-sb]

This looks good to me. I'll merge this PR.