RFR(xs): 8186465: Each j.l.Reference elapsed time log is incorrect
sangheon.kim at oracle.com
Thu Aug 24 05:09:56 UTC 2017
Thank you for the review!
On 08/23/2017 02:08 PM, Stefan Karlsson wrote:
> Hi Sangheon,
> On 2017-08-23 23:04, sangheon.kim wrote:
>> Hi all,
>> Could I have some reviews for this tiny fix?
>> The problem is each j.l.References has wrong time log as related
>> function is using wrong variable.
>> Actually it should be using '_ref_proc_time_ms' but accidentally
>> using '_par_phase_time_ms'. Setter is using correct one but getter is
>> referring wrong one.
>> CR: https://bugs.openjdk.java.net/browse/JDK-8186465
>> webrev: http://cr.openjdk.java.net/~sangheki/8186465/webrev.0/
>> Testing: manually checked log values
> Looks good, but I think you should add a regression test for this.
> Maybe the test could at least verify that the sub-phases are not
> larger than the enclosing phase?
Added a new sub-test in TestPrintReferences.java to see:
1. Reference Processing time >= Soft + Weak + Final + Phantom Reference
2. Each References >= phase 1(for Soft) + phase 2 + phase 3
Testing: JPRT, manual test to see sub-phases are not larger than the
FYI, from my local run:
Reference Processing: 115.6ms > 115.5 + 0 + 0 + 0
SoftReference: 115.5ms > 115.4 (36.5 + 37.2 + 41.7)
More information about the hotspot-gc-dev