RFR: Upstream merge 2018-06-30

[Roman Kennke]

Now I'm thinking I should probably redo the merge, but only up to:

changeset:   51326:e8d55141afd2
user:        jwilhelm
date:        Thu Jun 28 22:28:41 2018 +0200
summary:     8206006: Build failed on Windows

... which is the last changeset that both jdk/jdk11 and jdk/jdk have in
common, to have a basis for a clean fork.

Let me see if that's easily possible. Hold on.

Roman

> This merges upstream changes from jdk/jdk up to:
> 
> changeset:   51330:12133a6e2613
> user:        jlahoda
> date:        Fri Jun 29 10:41:10 2018 +0200
> summary:     8205418: Assorted improvements to source code model
> 
> 
> It includes some interesting (GC related) highlights:
> 
> - 8205336: Modularize allocations in assembler
> 
> - 8204540: Automatic oop closure devirtualization
>  this one caused a mess and some code re-arrangements in Shenandoah, but
> overall I think it's good
> 
> - 8205683: Refactor heap allocation to separate concerns
>  this one caused mess and I need to work with upstream/eosterlund to
> clean it up. See what I had to do in shenandoahHeap.cpp
> 
> All merged changesets:
> http://cr.openjdk.java.net/~rkennke/upstream-merge-2018-06-30/outgoing.txt
> 
> Shenandoah-related diff:
> http://cr.openjdk.java.net/~rkennke/upstream-merge-2018-06-30/shenandoah-changes.patch
> 
> Testing (ongoing): tier3_gc_shenandoah (fastdebug/non-PCH release/non-PCH)
> 
> As always, aarch64 *may* be broken (I gave my best). Will fix if needed
> afterwards.
> 
> Roman
>