RFR: 8008243: Zero: Implement fast-bytecodes

Severin Gehwolf sgehwolf at redhat.com
Wed Nov 25 10:04:02 UTC 2015


Hi,

Roman asked me to get his patch integrated. This is basically a rebase
of his patch to latest JDK 9 hs-rt + some incorporated review feedback.

Bug: https://bugs.openjdk.java.net/browse/JDK-8008243
webrev: http://cr.openjdk.java.net/~sgehwolf/webrevs/JDK-8008243/00/

The old review thread was here:
http://mail.openjdk.java.net/pipermail/hotspot-dev/2013-February/008449.html

With this patch I see a speedup of the Zero interpreter of about 18%.

I don't think we need an extra cache->is_resolved() condition for
putfield since this is already done earlier in the putfield case. As to
the breakpoint conditional I'm not sure how that should look like. Does
anybody have some pointers?

Since the patch relies on ordering of fast bytecodes I've added an
assert which attempts to catch this.

Thoughts?

Note: Once approved I'd need someone to sponsor this patch for me.

Cheers,
Severin


More information about the hotspot-dev mailing list