[aarch64-port-dev ] RFR: Fix code differences against shenandoah/jdk8u

[Roman Kennke]

Am 20.09.2018 um 14:50 schrieb Aleksey Shipilev:
> http://cr.openjdk.java.net/~shade/shenandoah/aarch64-sync-1/webrev.01/
> 
> After recent push to aarch64-port/shenandoah-8u, I diff'ed the hotspot repositories, and these are
> the differences between out development shenandoah/jdk8u and aarch64-port/shenandoah-8u. After
> applying this patch, there are no differences anymore.
> 
> History:
> 
>  *) src/share/vm/opto/library_call.cpp: this is whitespace difference that was left over after
> Shenandoah barriers addition. Upstream does not have these newlines.
> 
>  *) src/share/vm/opto/phaseX.cpp: webrev would show "if (m_op == Op_CmpI)" block was moved, but int
> fact it is Shenandoah-specific "if (m->is_ShenandoahBarrier())" block swapped places -- the webrev
> tool minimized the change showing the smaller block as diff. It is probably an innocuous difference,
> but difference nevertheless. The block was misplaced by earlier merge here:
> 
> http://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/hotspot/diff/21088d271abb/src/share/vm/opto/phaseX.cpp
> 
>  *) src/share/vm/prims/jni.cpp: this seems to be the confusion between multiple merges. The barrier
> story around JNI accesses is complicated, and I am not sure sh/jdk gets it correctly. Still, it is
> better to sync up the two, so fixes from sh/jdk8u proliferate in due course.
> 
> Testing: tier3_gc_shenandoah {release|fastdebug}
> 
> Thanks,
> -Aleksey
> 

Looks good.
Roman