RFR (S): 8177764: Humongous Reclaimed log output may be confusing

[Stefan Johansson]

Hi Thomas,

On 2017-05-08 17:52, Thomas Schatzl wrote:
> Hi all,
>
>    can I have reviews for this following small change (plus accompanying
> test case) that fixes log output for the "Humongous Reclaimed" message.
>
> In particular, the value next to it showed the number of reclaimed
> regions, while the previous related messages show the number of
> reclaimed objects.
>
> This resulted in log output where the value of "Humongous Reclaimed" is
> larger than the number of candidates. E.g.:
>
> [267.722s][debug][gc,phases ] GC(486) Humongous Register: 0.1ms
> [267.722s][trace][gc,phases ] GC(486) Humongous Total: 73
> [267.722s][trace][gc,phases ] GC(486) Humongous Candidate: 73
> [...]
> [267.722s][debug][gc,phases ] GC(486) Humongous Reclaim: 1.4ms
> [267.722s][trace][gc,phases ] GC(486) Humongous Reclaimed: 221
>
> This change makes "Humongous Reclaimed" to also report the number of
> reclaimed objects. There is still a gc+heap=info message at every GC
> that shows the number of reclaimed regions (as before/after
> difference).
>
> E.g.:
>
> [90.145s][info ][gc,heap        ] GC(13) Humongous regions: 150->13
>
> I think this information is sufficient to avoid the need for another
> log message.
>
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8177764
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8177764/webrev/
The change looks good,
StefanJ

> Testing:
> jprt, test case failing without patch, succeeding now
>
> Thanks,
>    Thomas
>