keytool and key password on PKCS#12 stores

[Weijun Wang]

Yes, you are right.

This is a behavior change (also it's not a correct behavior now), we 
will think about if it has any negative impact.

Thanks
Max

On 3/14/13 10:44 PM, Bruno Harbulot wrote:
> Hello,
>
> I've always thought that the keystore password and the key password
> itself had to be the same in a PKCS12 keystore, due to constraints in
> the PKCS#12 format. It appears this is not quite the case with the
> Java implementation, especially with a PKCS#12 store generated with
> keytool.
>
> When importing a PKCS#12 store into another PKCS#12 store, keytool
> changes the store password, but doesn't change the actual key
> password. However, keytool -keypasswd doesn't allow to change the key
> password itself on a PKCS#12 store.
>
> To test this, generate a PKCS#12 keystore (keystore1.p12):
>
>      keytool -genkeypair -dname "CN=test" -storetype PKCS12 -keystore
> keystore1.p12 -storepass testabcd
>
> Import/export this store into another PKCS#12 keystore (keystore2.p12):
>
>     keytool -importkeystore -srckeystore keystore1.p12 -srcstoretype
> PKCS12 -srcstorepass testabcd -destkeystore keystore2.p12
> -deststoretype PKCS12 -deststorepass test1234
>
> This will change the store password from testabcd to test1234, but the
> key password will still be testabcd. This can be tested with the
> following code:
>
>      KeyStore ks = KeyStore.getInstance("PKCS12");
>      InputStream ksis = new FileInputStream("keystore2.p12");
>      try {
>          ks.load(ksis, "test1234".toCharArray()); // New password
>      } finally {
>          if (ksis != null) {
>              ksis.close();
>          }
>      }
>      KeyManagerFactory kmf = KeyManagerFactory.getInstance(KeyManagerFactory
>              .getDefaultAlgorithm());
>      // New password, fails:
>      kmf.init(ks, "test1234".toCharArray());
>      // Old password, works:
>      // kmf.init(ks, "testabcd".toCharArray());
>
>
> I must admit I'm not familiar with the details of the PKCS#12
> specifications, but I think all the other tools I've seen that use the
> PKCS#12 format only have a password for the store, not for the
> individual keys. In fact, it seems that Firefox is capable of
> importing and using keystore2.p12 from the example above using only
> the store password, despite producing an error message.
> Shouldn't keytool -importkeystore also change the key passwords (or at
> least there seems to be a padding issue)? Here is the error message:
>
>      Caused by: javax.crypto.BadPaddingException: Given final block not
> properly padded
> 	at com.sun.crypto.provider.CipherCore.doFinal(CipherCore.java:811)
> 	at com.sun.crypto.provider.CipherCore.doFinal(CipherCore.java:676)
> 	at com.sun.crypto.provider.PKCS12PBECipherCore.implDoFinal(PKCS12PBECipherCore.java:355)
> 	at com.sun.crypto.provider.PKCS12PBECipherCore$PBEWithSHA1AndDESede.engineDoFinal(PKCS12PBECipherCore.java:387)
> 	at javax.crypto.Cipher.doFinal(Cipher.java:2087)
> 	... 6 more
>
>
> Best wishes,
>
> Bruno.
>