RFR (S): 8010294: Refactor HeapInspection to make it more reusable

Bengt Rutisson bengt.rutisson at oracle.com
Tue Mar 26 03:27:20 PDT 2013


Looks good.

Bengt

On 3/25/13 2:16 PM, Erik Helin wrote:
> All,
>
> I have updated the change based on feedback from Bengt and Jesper.
>
> Differences between webrev.01 and webrev.02:
> - Added #else clause in HeapInspection::start_of_perm_gen to make sure
>   that no compilers will isse any warnings.
> - Removed the variable need_epilogue in
>   HeapInspection::instance_inspection, just kept the comment.
> - Updated the @bug annotation of the test to the correct bug number,
>   8010294.
>
> New webrev located at:
> http://cr.openjdk.java.net/~ehelin/8010294/webrev.02/
>
> Thanks,
> Erik
>
> On 03/25/2013 12:29 PM, Erik Helin wrote:
>> Bengt,
>>
>> On 03/25/2013 11:58 AM, Bengt Rutisson wrote:
>>>
>>> Hi Erik,
>>>
>>> Changes look good!
>>>
>>> One minor nit. I think the test contains the wrong bug number:
>>>
>>> * @bug 8004924
>>
>> Agree, I will fix.
>>
>> Thanks,
>> Erik
>>
>>> Bengt
>>>
>>> On 3/25/13 9:32 AM, Erik Helin wrote:
>>>> All,
>>>>
>>>> I have updated the change based on internal feedback.
>>>>
>>>> The changes between webrev.00 and webrev.01 are mostly in the test 
>>>> code:
>>>> - The test uses correct @summary description
>>>> - The test now uses the correct command line flags
>>>> - The test is more documented to describe how the test code is run
>>>>
>>>> There is also a very minor change in the hotspot code. I am now using
>>>> the operator != instead of operator > when comparing the missed count.
>>>> This was done just to make the diff one line smaller :)
>>>>
>>>> Please see new webrev at:
>>>> http://cr.openjdk.java.net/~ehelin/8010294/webrev.01/
>>>>
>>>> Thanks,
>>>> Erik
>>>>
>>>> On 03/22/2013 10:06 AM, Erik Helin wrote:
>>>>> Hi all,
>>>>>
>>>>> I have refactored the HeapInspection class to prepare for the
>>>>> vm/gc/detailed/object_count_after_gc event.
>>>>>
>>>>> The refactoring mainly consists of breaking up the rather long
>>>>> heap_inspect function to multiple small functions that can be reused.
>>>>>
>>>>> I also moved the public enums used to initialize KlassInfoTable and
>>>>> KlassInfoHisto to private constants. Now they no longer have to be
>>>>> passed as an argument to the constructor.
>>>>>
>>>>> I also added a test to ensure that my refactoring did not break
>>>>> anything
>>>>> (the test is not that extensive but at least it ensures that the 
>>>>> basic
>>>>> stuff is working).
>>>>>
>>>>> Webrev:
>>>>> http://cr.openjdk.java.net/~ehelin/8010294/webrev.00/
>>>>>
>>>>> Bug:
>>>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8010294
>>>>>
>>>>> Testing:
>>>>> - JPRT
>>>>> - A new jtreg test (see webrev)
>>>>>
>>>>> Thanks,
>>>>> Erik
>>>>
>>>
>>>
>>
>



More information about the hotspot-gc-dev mailing list