RFR: 8328119: Support HKDF in SunPKCS11 (Preview) [v8]

[Weijun Wang]

On Thu, 19 Dec 2024 19:41:23 GMT, Martin Balao <mbalao at openjdk.org> wrote:

>> We would like to propose an implementation of the HKDF algorithms for SunPKCS11, aligned with the KDF API proposed for JDK 24 (see [JEP 478: Key Derivation Function API (Preview)](https://bugs.openjdk.org/browse/JDK-8189808)).
>> 
>> This implementation will be under the _Preview_ umbrella until the KDF API becomes stable in a future JDK release. The benefit of this early proposal is to gather more feedback about the KDF API for future improvements.
>> 
>> The `P11KDF` class has the core implementation and Java calls to the PKCS 11 API. Different native mechanism were used to merge key material: CKM_CONCATENATE_BASE_AND_DATA (key and data), CKM_CONCATENATE_BASE_AND_KEY (key and key) and CKM_CONCATENATE_DATA_AND_BASE (data and key). The implementation also supports merging data with data, at the Java level. List of HKDF algorithms supported: HKDF-SHA256, HKDF-SHA384, and, HKDF-SHA512.
>> 
>> Derivation modes supported: extract, expand, and, extract-expand.
>> 
>> We further advanced the consolidation of algorithm and key info in the P11SecretKeyFactory map —this effort started with the PBE support enhancement and has helped to avoid duplication—. The map has now information about HMAC (`HMACKeyInfo` class) and HKDF (`HKDFKeyInfo` class) algorithms. P11Mac is now aligned to take the information from the map.
>> 
>> Generic keys now supported in SecretKeyFactory. Derived keys could be Generic.
>> 
>> Testing
>> 
>>  * [TestHKDF.java](https://github.com/openjdk/jdk/blob/e87ec99b90ff742f531a5031fdeeb9f2e039856d/test/jdk/sun/security/pkcs11/KDF/TestHKDF.java) test added
>>    * All non-SHA1 & non-SHA224 RFC 5869 test vectors checked
>>    * Cross-checking against SunJCE's HKDF implementation for every algorithm possible
>>       * Static assertion data for resilience if SunJCE were not available
>>    * Use of derived key for encryption check
>>    * Concatenation of input key material and salt checked (multiple combinations)
>>    * Multiple derivation types checked (extract only, expand only, and, extract-expand)
>>    * Derive key and derive data checked
>>    * All supported HKDF algorithms tested (HKDF-SHA256, HKDF-SHA384, and, HKDF-SHA512)
>>    * DH and ECDH key derivation for TLS checked
>>    * Informative output for debugging purposes (shown automatically if there is a test failure)
>>      * Note: test failures do not prevent all tests for running
>>    * Test integrated to the SunPKCS11 tests framework
>> 
>>  * No regressions observed in jdk/sun/security/pkcs11 (114 tests pa...
>
> Martin Balao has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Check disabled PKCS #11 mechanisms when concatenating keys and data.
>   
>   Co-authored-by: Martin Balao Alonso <mbalao at redhat.com>
>   Co-authored-by: Francisco Ferrari Bihurriet <fferrari at redhat.com>

OK. I have a minor concern: this factory seems primarily useful for HSMs, and it’s unlikely that software-based providers would support it. Users should be mindful of its intended use. I noticed the CSR already references its connection to the CKK_GENERIC_SECRET type. You could clarify further that this is specifically used by a PKCS#11 provider to "import" an external key into a token. I believe we already agreed to address this in a separate enhancement.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/22215#issuecomment-2556105328