RFR: Remaining Shenandoah backports jdk/jdk -> sh/jdk11

[Aleksey Shipilev]

On 5/16/19 7:06 PM, Roman Kennke wrote:
>>> This backports the following changes to sh/jdk11:
>>>
>>> JDK-8221751: Shenandoah: Improve SATB enqueueing
>>> JDK-8221848: Shenandoah: ArrayCopy post-barrier improvements
>>> JDK-8222227: Shenandoah: Fix Traversal GC weak roots handling in
>>> final-traversal pause
>>> JDK-8222259: Shenandoah: Pre-evacuate string-dedup roots in Traversal GC
>>> JDK-8222188: Shenandoah: Adjust Shenandoah work gang types
>>> JDK-8218468: Load barrier slow path node should be MachTypeNode
>>> JDK-8223651: Build fails with --with-jvm-features=-jfr and
>>> --disable-precompiled-headers
>>>
>>> Changes transplanted with minimal and simple changes.
>>> Testing: hotspot_gc_shenandoah fine
>>>
>>> Ok?
>>
>> Okay. Is there a webrev that shows the changes, for the record?
> 
> Oops, forgot to include it:
> 
> http://cr.openjdk.java.net/~rkennke/backport-jdk11-2019-05-16/webrev.00/

*) The indenting is weird here in shenandoahBarrierSetAssembler_aarch64.cpp -- is it the same way in
upstream?

  90   if (is_oop) {
  91       Label done;
  92
  93       // Avoid calling runtime if count == 0
  94       __ cbz(count, done);
  95
  96       // Is updating references?
  97       Address gc_state(rthread, in_bytes(ShenandoahThreadLocalData::gc_state_offset()));
  98       __ ldrb(rscratch1, gc_state);
  99       __ tbz(rscratch1, ShenandoahHeap::UPDATEREFS_BITPOS, done);
 100
 101     __ push(saved_regs, sp);

Otherwise looks fine.

-Aleksey