RFR (XS): 8165860: workgroup classes are missing volatile qualifiers for lock-free code

Erik Österlund erik.osterlund at oracle.com
Tue Sep 20 08:50:31 UTC 2016


Hi Mikael,

On 2016-09-20 10:44, Mikael Gerdin wrote:
>
>
> On 2016-09-20 10:43, Thomas Schatzl wrote:
>> Hi Erik,
>>
>> On Tue, 2016-09-20 at 10:28 +0200, Erik Österlund wrote:
>>> Hi Thomas,
>>>
>>> On 2016-09-19 17:00, Thomas Schatzl wrote:
>>>> On Mon, 2016-09-19 at 15:50 +0200, Mikael Gerdin wrote:
>>>>> On 2016-09-19 15:48, Erik Österlund wrote:
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> Mikael Gerdin noticed a compiler workaround could be done in a
>>>>>> nicer-looking way.
>>>>    so just assigning the volatile to the temp is not sufficient to
>>>> make sure the compiler generates code to only read it once?
>>> The problem was more that solaris studio (only) complained that I was
>>> losing the volatile qualifier when I did it the straight forward way
>>> (just assigning from the volatile).
>>> I tried again, and it doesn't seem to complain any longer. I must
>>> have done something slightly different the first time I tried this -
>>> sorry about that.
>>>
>>> If you prefer just assign from the volatile, then here is a new
>>> webrev with that variant. I agree that probably looks better.
>>>
>>> Full webrev: http://cr.openjdk.java.net/~eosterlund/8165860/webrev.02
>>> /
>>> Incremental webrev:
>>> http://cr.openjdk.java.net/~eosterlund/8165860/webrev.01_02/
>>>
>>
>> Looks best =D.
> +1

Thanks for the review!

/Erik

> /Mikael
>>
>> I can sponsor it if you want.
>>
>> Thomas
>>



More information about the hotspot-gc-dev mailing list