On Fri, 17 Jan 2025 19:45:08 GMT, Martin Balao <mba...@openjdk.org> wrote:
>> src/jdk.crypto.cryptoki/share/classes/sun/security/pkcs11/P11HKDF.java line >> 117: >> >>> 115: boolean isExtract = false, isExpand = false; >>> 116: boolean isData = retType == byte[].class; >>> 117: assert isData || retType == SecretKey.class : "Invalid return >>> type."; >> >> Style comment: >> >> I don't see why passing the return-type is preferable here, making this >> `assert` check necessary. It's only internal code that is passing a >> `retType` argument anyway. Why not do something similar to what the software >> implementation does and just return a byte[] from this method? > > In some cases we need to return a `SecretKey` (a `P11SecretKey` instance, > internally) that represents a key inside the token. In some cases, we can > extract its bytes and create a key again with key translation, but it's > costly. In some others (e.g. the key is not extractable or is sensitive), we > cannot do that. I see, so you are attempting to cover three cases then: 1) raw bytes 2) present `SecretKey` 3) token `SecretKey` In case three, the data would never have been available to the provider, so you do not have bytes to return -- and it would not make sense to represent the token as a byte[] I suppose. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/22215#discussion_r1920690305
