RFR 8235515: TestBasicLogOutput.java crashes due to EXCEPTION_ACCESS_VIOLATION in HandshakeThreadsOperation::do_handshake

David Holmes david.holmes at oracle.com
Wed Dec 11 06:28:27 UTC 2019


On 11/12/2019 8:13 am, Patricio Chilano wrote:
> Hi David,
> 
> On 12/10/19 4:19 PM, David Holmes wrote:
>> Hi Patricio,
>>
>> On 11/12/2019 1:06 am, Patricio Chilano wrote:
>>> Hi,
>>>
>>> Could you review this small fix? Signaling the _done semaphore has to 
>>> be done after logging, otherwise we could be accessing an already 
>>> destroyed HandshakeThreadsOperation object when executing name().
>>
>> Great catch!
>>
>> If not too late can you add a comment after:
>>
>> +   // Use the semaphore to inform the VM thread that we have 
>> completed the operation
>> +   _done.signal();
>>
>> // It is no longer safe to refer to 'this' as the VMThread may have 
>> destroyed this operation
> I already pushed the fix, sorry about that, but if you are okay I can 
> add that comment in 8232733(Remove need to grab Threads_lock while 
> processing handshakes) since I'm touching same code.

Sounds good to me :)

Thanks,
David

> Thanks for looking at this David!
> 
> Patricio
>> Thanks,
>> David
>>
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8235515
>>> Webrev: http://cr.openjdk.java.net/~pchilanomate/8235515/v01/webrev/
>>>
>>> Thanks!
>>> Patricio
> 


More information about the hotspot-runtime-dev mailing list