RFR (xs) 8059588: deadlock in java/io/PrintStream when verbose javax.net.debug flags are set

Seán Coffey sean.coffey at oracle.com
Fri Mar 27 19:30:28 UTC 2015


Brad,

you're correct. The debug flag for pkcs11 is enabled via 
java.security.debug=pkcs11 - I'll update the bug synopsis. The changeset 
has already been pushed though.

regards,
Sean.

On 27/03/2015 19:27, Bradford Wetmore wrote:
> Sean,
>
> Is this a java.security.debug or a javax.net.debug issue as mentioned 
> in the synopsis?  It imports from sun.security.util.Debug, which is 
> java.security.debug, not javax.net.debug.
>
> In the bug, both variables were set.
>
> ...deleted...
> -Djavax.net.debug=ssl,handshake,record,keygen,session,defaultctx,sslctx,ses 
>
> sioncache,keymanager,trustmanager 
> -Dcom.oracle.security.ucrypto.debug=true
> -Djava.security.debug=all -Djsse.S
>
> Probably the synopsis should be changed.
>
> Brad
>
>
>
> On 3/27/2015 12:05 PM, Seán Coffey wrote:
>> Thanks Sean. Yes - I'll add those braces before pushing.
>>
>> regards,
>> Sean.
>>
>> On 27/03/2015 18:58, Sean Mullan wrote:
>>> This looks fine, just one comment. Can you put braces around the if
>>> debug statement on lines 117-118 (just to be safe).
>>>
>>> --Sean
>>>
>>> On 03/25/2015 09:52 AM, Seán Coffey wrote:
>>>> This issue was reported while debugging a pkcs11 issue. While I can't
>>>> reproduce it, the captured stack trace does show a deadlock issue. Fix
>>>> is to simply stop using the SessionManager class lock to record and
>>>> print the maxActiveSessions in use and replace it with a local lock.
>>>>
>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8059588
>>>> webrev : http://cr.openjdk.java.net/~coffeys/webrev.8059588/webrev/
>>>>
>>>> Regards,
>>>> Sean.
>>



More information about the security-dev mailing list