Debuggability of failures in sun.security.rsa.RSASignature

[Brad Wetmore]

Hi Matthew,

I've just taken a quick look, but yes, this seems to be a usability 
issue that should somehow be addressed, either by adding some 
logging/debugging or throwing a SignatureException.  There's currently 
no logging/debugging in this package.

We'd need to figure out why the original author made the original 
decision to swallow the exception.

I've filed:

     http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8011740

to track this.

Brad


On 4/1/2013 6:49 PM, Matthew Hall wrote:
> Hi,
>
> This code in RSASignature catches javax.crypto.BadPaddingException without
> logging it, and some of the functions in try { ... } have detailed exceptions
> which get lost when this exception is not properly propagated to calling code.
>
> At minimum there should be a security logging debug flag which enables logging
> the exceptions instead of silently suppressing them, otherwise it's impossible
> to troubleshoot or even detect that issued were encountered here without using
> a debugger on it.
>
> Thoughts?
> Matthew.
>
>      // verify the data and return the result. See JCA doc
>      protected boolean engineVerify(byte[] sigBytes) throws SignatureException {
>          byte[] digest = getDigestValue();
>          try {
>              byte[] decrypted = RSACore.rsa(sigBytes, publicKey);
>              byte[] unpadded = padding.unpad(decrypted);
>              byte[] decodedDigest = decodeSignature(digestOID, unpadded);
>              return Arrays.equals(digest, decodedDigest);
>          } catch (javax.crypto.BadPaddingException e) {
>              // occurs if the app has used the wrong RSA public key
>              // or if sigBytes is invalid
>              // return false rather than propagating the exception for
>              // compatibility/ease of use
>              return false; *** PROBLEM LINE ***
>