RFR: JDK-8008790 - Promotion failed tracing event for all GCs

Erik Helin erik.helin at oracle.com
Wed Mar 6 10:01:18 PST 2013


Hi Jesper,

looks good.

Thanks,
Erik

On 03/06/2013 05:04 PM, Jesper Wilhelmsson wrote:
> Hi,
>
> Additional feedback resulted in a new webrev:
>
> http://cr.openjdk.java.net/~jwilhelm/8008790/webrev.hsx24.3/
>
> Thanks,
> /Jesper
>
>
> On 26/2/13 4:54 PM, Jesper Wilhelmsson wrote:
>> Hi,
>>
>> After some internal feedback a new version of the webrev is available at:
>>
>> http://cr.openjdk.java.net/~jwilhelm/8008790/webrev.hsx24.2/
>>
>> The biggest change is that the promotion failed event now is sent once
>> per
>> thread instead of merging the info into one event.
>>
>> Thanks,
>> /Jesper
>>
>>
>> On 25/2/13 1:53 PM, Jesper Wilhelmsson wrote:
>>> Hi,
>>>
>>> I'm looking for a couple of reviews for the promotion failed tracing
>>> event, now
>>> implemented for ParNew and Serial GC.
>>>
>>> I have replaced the existing counters and booleans with a
>>> PromotionFailedInfo
>>> object and uses that throughout, so now promotion failed reporting
>>> (through the
>>> tracing framework) looks the same and reports the same data for all
>>> young
>>> collectors.
>>>
>>> I have added two new fields to the promotion failed event: firstSize
>>> as reported
>>> by ParNew logging, and smallestSize which is the smallest object size
>>> that
>>> failed to promote.
>>>
>>> Webrew: http://cr.openjdk.java.net/~jwilhelm/8008790
>>>
>>> Testing: Passes JPRT and the existing promotion failed tests with all
>>> collectors.
>>>
>>> Thanks,
>>> /Jesper



More information about the hotspot-gc-dev mailing list