RFR: 8308764: Reporting errors from create_vm may crash

[David Holmes]

On Thu, 1 Jun 2023 15:18:02 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

>> src/hotspot/share/runtime/threads.cpp line 552:
>> 
>>> 550:   if (status != JNI_OK) {
>>> 551:     // Don't delete main_thread here as we may need to report exceptions.
>>> 552:     *canTryAgain = false; // don't let caller call JNI_CreateJavaVM again
>> 
>> So could the jni code try to create a vm, fail and then go on to do other stuff?  Would this then be a memory leak?
>
> maybe this should be a if (!HAS_PENDING_EXCEPTION) main_thread->smr_delete(); and then in the caller will exit with vm_exit_during_initialization() which does exit() so not necessary to delete the main thread.

> Would this then be a memory leak?

Yes but at this point we are "leaking everything that VM initialization created but does not clean up. So adding a JavaThread instance to that seems insignificant.

> maybe this should be a if (!HAS_PENDING_EXCEPTION) main_thread->smr_delete(); 

That would work too. I can make that switch.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/14238#discussion_r1213820157