[8u] RFR: 8242480: Negative value may be returned by getFreeSwapSpaceSize() in the docker

[Severin Gehwolf]

Hi Andrew,

On Wed, 2020-11-25 at 06:35 +0000, Andrew Hughes wrote:
> On 15:37 Mon 02 Nov     , Severin Gehwolf wrote:
> > Hi,
> > 
> > Please review this backport for Oracle JDK parity. The JDK 11 patch did
> > apply cleanly after JDK-8221710 which is in the approval queue, but the
> > test was not working and needed some fixes:
> > 
> >  * Common.java in OpenJDK 8u uses unified logging VM args. Used
> >    -XX:+UnlockDiagnosticVMOptions -XX:+PrintContainerInfo instead.
> >  * The test library location needed updates
> > 
> > Bug: https://bugs.openjdk.java.net/browse/JDK-8242480
> > webrev: https://cr.openjdk.java.net/~sgehwolf/webrevs/JDK-8242480/jdk8/01/webrev/
> > 
> > Testing: Regression test on Linux x86_64. Failed before, passes after.
> > 
> > Thoughts?
> > 
> > Thanks,
> > Severin
> > 
> 
> I think the Common.java change should go in under a separate bug. It's not
> an issue with this fix, but a bug introduced by 8223147: "JFR Backport".

I'm not disagreeing, but practical matters need to be taken into
account. It's not worth the extra overhead. Separate bug, separate
approval/review. So my preference would be to keep it as part of this
fix as it seems to be the first test which makes use of Common. Without
it, the test fails. We shouldn't delay this fix any further.

> It's also worth noting that your changes are backed up by the duplicate copy
> of Common.java found in the HotSpot repository [0].

Yes.

> If these tests are HotSpot-specific, should they not be in the HotSpot
> repository?

These tests test the container capabilities in Java code
(Metrics.java). Container support has been added to hotspot and core
libraries. These tests aren't hotspot specific. They are for the Java
side.

Thanks,
Severin

> [0] https://hg.openjdk.java.net/jdk8u/jdk8u/hotspot/file/83661fdee9f0/test/testlibrary/com/oracle/java/testlibrary/Common.java#l67
> 
> Thanks,