[URGENT] [8u] Re-cast MetadataOnStack patch to be Shenandoah-specific

[Zhengyu Gu]

Hi Aleksey,

The nontrivial divergence was introduced by [1], intended to fix a hard 
to reproduce crash. Apparently, it does not work, I am more leaning 
toward backing it out if it is still an option.

Thanks,

-Zhengyu


[1] http://hg.openjdk.java.net/shenandoah/jdk8/hotspot/rev/b261d2db2ea2

On 10/7/21 04:36, Aleksey Shipilev wrote:
> There is a hard requirement for 8u-aarch64 integration that any 
> non-mainline patch in shared code was either trivial, or be obviously 
> Shenandoah-specific. Unfortunately, at least one of the changes affects 
> the shared path. With this patch, I demote it to Shenandoah-specific 
> paths only.
> 
> This blocks 8u integration.
> 
> Note: a similar problem exists in jni.cpp, see: 
> https://builds.shipilev.net/patch-openjdk-shenandoah-jdk8/hotspot/src/share/vm/prims/jni.cpp.sdiff.html 
> -- I'll take a stab at it, but Zhengyu, please take a look as well. That 
> one is urgent as well.
> 
> 8u webrev:
>    
> https://cr.openjdk.java.net/~shade/shenandoah/8u-metadata-on-stack/webrev.01/ 
> 
> 
> Testing: hotspot_gc_shenandoah
>