[13] RFR JDK-7107615 "scalability bloker in javax.crypto.JceSecurity"

[Alan Bateman]

On 13/05/2019 10:04, Daniel Fuchs wrote:
> Hi Valery,
>
> On 11/05/2019 00:36, Valerie Peng wrote:
>> http://cr.openjdk.java.net/~valeriep/7107615/webrev.02/
>>
>> Please let me know if you have more comments.
>
> If I'm not mistaken, the only thing that references
> the IdentityWrapper<Provider> is the key in the WeakHashMap.
> Therefore, it is only weakly referenced and can be immediatly
> garbage collected. I don't think this is what you want?
>
> I believe what you are trying to achieve there is rather to use
> a plain ConcurrentHashMap, and have IdentityWrapper extend
> WeakReference<Provider> instead. You may need to store
> the hashCode in IdentityWrapper so that it doesn't change
> when the underlying Provider is garbage collected.
>
> Then you can use a ReferenceQueue to purge the map regularly.
Right, plus getVerificationResult is accessing the WeakHashMap without 
synchronization.

I think it would be useful to get a summary on whether this issue is 
trying to address one or two points. For the scalability point then I 
assume a CHM + computeIfAbsent would help. If the issue is also that you 
want the key (Provider) to be weak then it will require additional work, 
as Daniel points out.

-Alan