[OpenJDK 2D-Dev] <2D Dev> [7u6] Request for review: 7027300 Unsynchronized HashMap access causes endless loop

[Vladislav Karnaukhov]

Andrew,

I'm not aware this was already fixed and introduced in jdk8. Actually 
I've fixed this in jdk6 as escalated issue, and we agreed with Phil Race 
that I'll forward it into jdk7 and he will do this for jdk8. I don't 
even have his formal approve for jdk6…

Anyway, I agree that we should keep our code as consistent as possible. 
Could you please point me at jdk8 diff?

Regards,
- Vlad

On 19.06.12 11:35, Andrew Brygin wrote:
> Hello Vladislav,
>
> the fix itself seems to be pretty close to the fix for jdk8,
> but introduces different import order, and an additional comment.
>
> From my point of view, we should keep jdk7 and jdk8 sources as close
> as it is possible just to reduce the backporting overhead.
> In this particular case, I do not see a strong need in mentioned 
> modifications
> so could you please consider to do a direct backport of the jdk8 change?
>
> Thanks,
> Andrew
>
> On 18.06.2012 17:31, Vladislav Karnaukhov wrote:
>> Hello,
>>
>> please review a fix for CR 7027300: Unsynchronized HashMap access 
>> causes endless loop.
>>
>> The issue is that SunLayoutEngine.getEngine uses non-synchronized 
>> HashMap, which may cause an endless loop.
>> The fix is to use ConcurrentHashMap instead of HashMap.
>>
>> Bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7027300
>> Webrev: http://cr.openjdk.java.net/~vkarnauk/7027300/webrev.00/
>>
>> Regards,
>> - Vlad
>