[wildfly-dev] app policy race

[Jason Greene]

Memory visibility would be guaranteed. If there is no happens-before ordering then one thread's put is not always visible to another thread's get even if they happen in the proper order. Also corruption of the table could result in a dead table.

E.g.

1. Thread A -> put Kz
2. Other stuff
3. Thread B -> get Kz 
(Returns cached stale value, null)


Granted there could still be some ordering problem. Like we are allowing requests to get handled before the server is started.


Sent from my iPhone

> On Dec 29, 2013, at 4:54 AM, Ales Justin <ales.justin at gmail.com> wrote:
> 
> How would proper map sync help here then?
> 
> Are you saying there is a put and get going on at the same time?
> Imo, this then means WF' dependencies are not done right.
> 
> btw: when do you "corrupt" a HashMap? (wrt my diff manifestation)
> With multiple concurrent puts/removes, or can it be mixture of puts/gets/removes?
> 
>> On Dec 29, 2013, at 2:56, Jason Greene <jgreene at redhat.com> wrote:
>> 
>> It looks exactly how it would manifest. Something isn't in the map that should be, so an error is thrown.
>> 
>>> On Dec 27, 2013, at 1:26 PM, Ales Justin <ales.justin at gmail.com> wrote:
>>> 
>>> That's one thing -- dunno how that did pop-up before. ;-)
>>> 
>>> But I doubt that just non-synchronised collections/maps cause this -- it would manifest differently.
>>> Imo, it must also be a race-condition in WF.
>>> 
>>>> On 27 Dec 2013, at 17:50, Anil Saldhana <Anil.Saldhana at redhat.com> wrote:
>>>> 
>>>> Stefan merged the fixes recently. I guess we need a PicketBox upgrade in WF.
>>>> 
>>>>> On 12/27/2013 06:54 AM, Ales Justin wrote:
>>>>> I often get this error: 
>>>>> * https://gist.github.com/alesj/8146623
>>>>> 
>>>>> I guess it's related to https://issues.jboss.org/browse/SECURITY-777 ?
>>>>> 
>>>>> Do we have a JIRA for WF as well?
>>>>> 
>>>>> -Ales
>>>>  
>>>> _______________________________________________
>>>> wildfly-dev mailing list
>>>> wildfly-dev at lists.jboss.org
>>>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>> 
>>> _______________________________________________
>>> wildfly-dev mailing list
>>> wildfly-dev at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
> _______________________________________________
> wildfly-dev mailing list
> wildfly-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/wildfly-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/wildfly-dev/attachments/20131229/d38e5bec/attachment.html