On Mon, 5 Apr 2021 17:57:13 GMT, Harold Seigel <hsei...@openjdk.org> wrote:
> Please review this additional cleanup of use of TRAPS in hotspot runtime
> code. The changes were tested with Mach5 tiers 1-2 on Linux, Mac OS, and
> Windows and Mach5 tiers 3-5 on Linux x64.
>
> Thanks, Harold
Hi Harold,
I looked at the changes to synchronization code and looks good except for one
issue below.
Thanks,
Patricio
src/hotspot/share/prims/jni.cpp line 2738:
> 2736:
> 2737: Handle obj(THREAD, JNIHandles::resolve_non_null(jobj));
> 2738: ObjectSynchronizer::jni_exit(THREAD->as_Java_thread(), obj());
Here we would return JNI_ERR if we throw IMSE from jni_exit().
src/hotspot/share/runtime/synchronizer.cpp line 609:
> 607: // intentionally do not use CHECK on check_owner because we must exit
> the
> 608: // monitor even if an exception was already pending.
> 609: if (monitor->check_owner(current)) {
We can actually throw IMSE from check_owner() if this thread is not the real
owner.
-------------
PR: https://git.openjdk.java.net/jdk/pull/3345
