Christian Thalinger christian.thalinger at oracle.com
Fri Jan 10 13:04:47 PST 2014

John is on vacation right now but here is what I know.  Years and years ago John wanted to rewrite the inlining policy and so the existing one got moved under UseOldInlining.  It turned out (and here starts the guesswork) that the new inlining was either not as good or wasn’t finished and so was never turned on.  Since then we run with UseOldInlining.

I’ve talked to John about this in the past and if I remember correctly he agreed that we can get rid of it.  Let’s wait for him to come back and decide then what to do.  Can you file a bug so we don’t forget?

On Jan 10, 2014, at 12:23 AM, Aleksey Shipilev <aleksey.shipilev at oracle.com> wrote:

> Hi,
> Does anyone remember the story behind -XX:+UseOldInlining? I have
> googled, wandered through HotSpot code, searched through JIRA, grepped
> through Mercurial to no avail. The history on this predates OpenJDK.
> In seems we have it unconditionally turned on. It makes the inlining
> policy a little less understandable to have additional branch here and
> there. Should we purge this option from the VM codebase? If so, I can
> prepare the proper revision of this scratch change:
>  http://cr.openjdk.java.net/~shade/scratch/purge-old-inlining.patch
> Thanks,
> -Aleksey.

More information about the hotspot-compiler-dev mailing list