RFR: [8u] Workaround VM crash with JNI Weak Refs handling

[Aleksey Shipilev]

On 02/19/2018 01:34 PM, Roman Kennke wrote:
> Sounds like it requires a keep_alive_barrier() in the place where the
> oop is moved from jweak -> local?

Thought so too, and this seems to be what jdk9 and jdk10 are doing for G1. But, it does not fix the
bug. Also, it sometimes fails on asserts with mismatched ranks in SATB and Compile locks, so I guess
we would need to look into that too. The complete fix seems to be much harder than I expected.

> As a workaround, the patch is ok.

Yup.

-Aleksey