RFR: 8331008: Implement JEP 478: Key Derivation Function API (Preview) [v24]

[Kevin Driver]

On Fri, 6 Sep 2024 13:31:07 GMT, Viktor Klang <vklang at openjdk.org> wrote:

>> Kevin Driver has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains 31 additional commits since the last revision:
>> 
>>  - Merge remote-tracking branch 'origin/master' into kdf-jep-wip
>>    # Please enter a commit message to explain why this merge is necessary,
>>    # especially if it merges an updated upstream into a topic branch.
>>    #
>>    # Lines starting with '#' will be ignored, and an empty message aborts
>>    # the commit.
>>  - several more review comments
>>  - change impl class to use byte arrays rather than SecretKey objects where possible
>>  - updated delayed provider selection javadoc
>>  - review comments
>>  - use a delegate record to hold the spi and provider
>>  - assorted review comment changes
>>  - another round of review comments
>>  - consistency with wording for addIKM and addSalt
>>  - another round of code review comments
>>  - ... and 21 more: https://git.openjdk.org/jdk/compare/568a513f...a35e98c9
>
> src/java.base/share/classes/com/sun/crypto/provider/HkdfKeyDerivation.java line 259:
> 
>> 257:                     os.writeBytes(CipherCore.getKeyBytes(workItem));
>> 258:                 }
>> 259:                 return os.toByteArray();
> 
> I haven't found any specification indicating that flush is not required before toByteArray(), but it can't hurt right?
> 
> Suggestion:
> 
>                 os.flush();
>                 return os.toByteArray();

@viktorklang-ora added additional comment on deliberately not calling `flush`.

> src/java.base/share/classes/com/sun/crypto/provider/HkdfKeyDerivation.java line 337:
> 
>> 335:             info = new byte[0];
>> 336:         }
>> 337:         int rounds = (outLen + hmacLen - 1) / hmacLen;
> 
> Perhaps make the addition as long, to rule out the risk of overflow?
> Suggestion:
> 
>         int rounds = (0L + outLen + hmacLen - 1) / hmacLen;

@viktorklang-ora: Further addressed in https://github.com/openjdk/jdk/pull/20301/commits/dc0bd1552ea14adaec4cc8e67d4213826c1b175c. Excluding any `hmacLen` values that are not currently supported.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/20301#discussion_r1752653636
PR Review Comment: https://git.openjdk.org/jdk/pull/20301#discussion_r1752657886