RFR: 8255564: InterpreterMacroAssembler::remove_activation() needs to restore thread right after VM call on x86_32

[Zhengyu Gu]

On Thu, 29 Oct 2020 06:25:34 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> Currently, it restores thread register a bit too late for reset_last_Java_frame().
>> 
>> It is probably not a problem right now, cause there is no 32-bit GC that supports concurrent stack processing. It crashes Shenandoah GC with concurrent stack processing on x86_32, which I am working on.
>
> Thanks! This is a regression since JDK-8255233, I linked the bugs.

Thanks, David and Aleksey!

-------------

PR: https://git.openjdk.java.net/jdk/pull/919