RFR(s) 8170873: PPC64: Poor StrictMath performance due to non-optimized compilation

[Gustavo Romero]

Hi,

Could the following webrev be reviewed please?

webrev 1/2: http://cr.openjdk.java.net/~gromero/8170873/v2/
webrev 2/2: http://cr.openjdk.java.net/~gromero/8170873/v2/jdk/
bug       : https://bugs.openjdk.java.net/browse/JDK-8170873

It's a backport of 8170153:

8170153: PPC64/s390x/aarch64: Poor StrictMath performance due to non-optimized
compilation

Bug       : https://bugs.openjdk.java.net/browse/JDK-8170153
Webrev 1/2: http://cr.openjdk.java.net/~gromero/8170153/v2/
Webreb 2/2: http://cr.openjdk.java.net/~gromero/8170153/v2/jdk/
Review    : http://mail.openjdk.java.net/pipermail/hotspot-dev/2016-November/025348.html
URL 1/2   : http://hg.openjdk.java.net/jdk9/jdk9/rev/3c7b02f5fa7c
URL 2/2   : http://hg.openjdk.java.net/jdk9/jdk9/jdk/rev/b6bad6302dc8

The original change affects s390x and aarch64 as well but since both
architectures are not present on 8u both were excluded from this new webrev
turning it, in effect, a ppc-only fix on 8u.

I understand that that change is particularly important to PPC64 since [1] is
not planned to be backported.

I've tested the same way I did on 9 (jtreg) and no regression was observed [2].

I did not run against the JCK tests as performend by Martin [3]
since I'm still in the process of solving an issue related to the OCTLA.

Thank you.


Regards,
Gustavo

[1] https://bugs.openjdk.java.net/browse/JDK-8134780
[2] http://mail.openjdk.java.net/pipermail/hotspot-dev/2016-November/025283.html
[3] http://mail.openjdk.java.net/pipermail/hotspot-dev/2016-November/025317.html