RFR: 6991327 : using -Xprof trigger native memory leak

[Zhengyu Gu]

Thanks for quick review.

-Zhengyu

On 8/29/2013 11:33 PM, David Holmes wrote:
> Looks okay to me.
>
> Thanks,
> David
>
> On 30/08/2013 3:52 AM, Zhengyu Gu wrote:
>> The is a simple fix for memory leak.
>>
>> FlatProfiler::record_thread_tick allocates array for JavaThread list,
>> but never free it.
>>
>>
>> External bug: http://bugs.sun.com/view_bug.do?bug_id=6991327
>> Internal bug: https://bugs.openjdk.java.net/browse/JDK-6991327
>> Webrev: http://cr.openjdk.java.net/~zgu/6991327/webrev.00/
>> <http://cr.openjdk.java.net/%7Ezgu/6991327/webrev.00/>
>>
>> Test:
>>    Ran test case (2) on bug report with NMT detail tracking on Linux 32,
>> diff report clearly shows memory leak at
>> FlatProfiler::record_thread_tick().  pmap also shows total memory grows
>> continuously.
>>
>>    With the fix, the call site no longer shows on diff report, and there
>> are not obviously memory leaks on the report. pmap shows total memory
>> stops growing after a while.
>>
>> Sanity check:
>>    Passed JPRT with -Xprof flag
>>
>> Thanks,
>>
>> -Zhengyu
>>