RFR 8242184: CRL generation error with RSASSA-PSS

Valerie Peng valerie.peng at oracle.com
Wed Apr 8 03:46:40 UTC 2020


Hi Max,

Not all of the comments are related to the changes in the webrev, just 
notice some PSS related inconsistency and thought I will ask:

<AlgorithmId.java>

- For RSASSA-PSS keys, existing code in 
getDefaultAlgorithmParameterSpec(...) (line 1121) decides the default 
based on key size. I think we should consider checking if the key 
contains PSS parameters. If present, use it as default.

- In the checkKeyAndSigAlgMatch(...) method (line 1027), don't we need 
to add checking for RSASSA-PSS signature? RSASSA-PSS sig can take both 
RSA and RSASSA-PSS keys.

- The getEncAlgFromSigAlg(...) method just returns the key algorithm as 
result when the specified signature algorithm does not contain "with". 
As RSASSA-PSS signature can use both RSASSA-PSS and RSA keys, should it 
still return key algorithm?

- The makeSigAlg(...) method does not work for RSASSA-PSS.

<X509CRLImpl.java>

- The sign() methods of X509CertImpl class do not generate default 
parameters automatically. Have you considered adding a sign() method to 
X509CRLImpl class which has extra signature parameter spec argument and 
move the default parameter call to the caller instead of inside 
X509CRLImpl? I think it's more suitable for the caller to generate the 
default unless there are many callers all need the same functionality.

Thanks,

Valerie

On 4/6/2020 8:11 PM, Weijun Wang wrote:
> Please review the fix at
>
>     http://cr.openjdk.java.net/~weijun/8242184/webrev.00/
>
> The major change is inside X509CRLImpl.java to allow params setting and reading.
>
> I also take this chance to:
>
> 1. Provide a default -sigalg for "keytool -genkeypair -keyalg rsassa-pss".
>
> 2. Revert a former change in X509CertImpl.java, which might be a safer call.
>
> Thanks,
> Max
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/security-dev/attachments/20200407/2a3feef6/attachment.htm>


More information about the security-dev mailing list