[OpenJDK 2D-Dev] Request for review/OpenJDK6: 6892485: Deadlock in SunGraphicsEnvironment / FontManager

[joe.darcy at oracle.com]

On 5/3/2010 6:35 PM, Martin Buchholz wrote:
> Hi Roman,
>
> I can confirm that your fix addresses the hang
> in my hacky test case FontDeadlock.java,
> so we at Google will use your fix for openjdk6,
> and would like to see it in openjdk6 proper.
>
> Joe, could you give this fix a thumbs up?
>   

Hmm.  Is this is noreg-hard and your hacky test is too hacky to check in 
as a regression test?

I'm inclined to let this go in, but want to make sure there is 
appropriate test coverage.

-Joe

> Roman, please commit this to openjdk6
> (or I can do it for you, if you prefer).
>
> Martin
>
> On Wed, Mar 3, 2010 at 12:33, Igor Nekrestyanov
> <Igor.Nekrestyanov at sun.com> wrote:
>   
>> Fix is ok with me.
>>
>> Here is the process description for openjdk6 -
>> http://j2se.sfbay/web/bin/view/CoreTech/OpenJDK6
>>
>> -igor
>>
>> On 3/3/10 7:01 AM, Roman Kennke wrote:
>>     
>>> This is the fix for the deadlock problem in
>>> SunGraphicsEnvironment/FontManager backported to OpenJDK6. It is
>>> slightly different from the OpenJDK7 version due to the restructured
>>> font manager, but the basic idea is the same: instead of sync'ing on 2
>>> different lock objects (with the potential of entering the monitors in
>>> the wrong order) we only sync on one (the FontManager class in this
>>> case, was the FontManager instance in JDK7).
>>>
>>> http://cr.openjdk.java.net/~rkennke/6892485/webrev.02/
>>>
>>> Ok to push?
>>>
>>> Also, is there any additional stuff that needs to be done for pushing to
>>> OpenJDK6, I never did that and I see that there are not team
>>> repositories. When I have 2 ok's, should I push directly to the
>>> jdk6/jdk6-gate/jdk repository?
>>>
>>> Thanks, Roman
>>>
>>>
>>>       
>>