RFR: Bulk backports sh/jdk11 -> sh/jdk8u

[Roman Kennke]

This backports outstanding changes from mid-Dec to ~now from 
shenandoah/jdk11 to shenandoah/jdk8u.

Find the list of changes here:
http://cr.openjdk.java.net/~rkennke/backports-sh-jdk8-2019-03-15/changes.txt

The full webrev:
http://cr.openjdk.java.net/~rkennke/backports-sh-jdk8-2019-03-15/webrev.00/

Disregard the C2 parts (under src/share/vm/opto) here. This backport 
contains a rather huge refactoring and reshaping of the barrier 
expansion code. It should now match (for the most part) what we have in 
jdk11+. Differences are: ripped-out loop-strip-mining, a couple of jdk8u 
specific idiosyncrasies (e.g. new (C) XYZNode), and for now split-if 
optimization path is disabled (I assume Roland is working on it and we 
might turn it on in a follow-up patch):

http://cr.openjdk.java.net/~rkennke/backports-sh-jdk8-2019-03-15/webrev-c2barriershape/

This one needs careful reviewing and also testing. It is a somewhat 
risky change, but sortof necessary. Overall I believe it reduces our 
upstream exposure, and makes our backporting story much better. It also 
prepares the way for LRB which will arrive ... at some point ;-)


Roman