* Added PK callback context tests for client/server examples (`SetupPkCallbackContexts`).
* Added new test define for `TEST_PK_PRIVKEY` to allows simulating hardware based private key.
* Added new test.h function for loading PEM key file and converting to DER (`load_key_file`).
* Added way to get private key signature size (`GetPrivateKeySigSize`).
* Added new ECC API `wc_ecc_sig_size_calc` to return max signature size for a key size.
* Added inline comments to help track down handshake message types.
* Cleanup of RSS PSS terminating byte (0xbc) to use enum value.
* Fixed bug with PK callback for `myEccVerify` public key format.
* Fixed bug with PK callback for ED25519 verify key buffer in DoServerKeyExchange.
1. Added some bounds checking on the ciphertext passed into the RSA decrypt function. NIST SP 800-56B specifies that the ciphertext shouldn't be a number larger than the modulus.
2. Added an API test to check that the direct RSA decrypt function returns an error with a "bad" message.
3. Added an ifndef guard to disable the bounds check. Default is to keep the bounds check.
4. RSA Decrypt bounds check only checked the first time into wc_RsaFunction().
--enable-sp-math to include minimal implementation of MP (only with
--enable-sp.)
Add futher functionality for ECC (conditionally compiled):
- check key
- is point on curve
- API to add and double projective points
- API to map from project to affine
- Uncompress point (including sqrt)
Some configuration options will not work with SP math - configure.ac
detects this and errors out.
Change test code to better support SP sizes only.
A recent change to the RSA key generation process is capping the number of attempts of finding a probable prime to a multiple of the prime's size, in FIPS builds. This means it might fail once in a while. (It could also fail for a couple other reasons but this is the most likely.) The API is changed to retry key generation until it succeeds. Non-FIPS builds keep trying until they find a prime.
