On Sun, 9 Aug 2026 00:18:15 GMT, Shiv Shah <[email protected]> wrote:

> Converts suspend001 to ThreadWrapper and fixes the hang. The test assumed 
> MyThread always grabs the shared lock first once main releases it, which 
> isn't guaranteed. With virtual threads the single wakeup goes to the first 
> thread that blocked, that's Suspended, and a suspended virtual thread can't 
> mount, so MyThread was never woken and the debuggee stalled.
> 
> Now each thread gates on its own lock so nothing depends on who gets the 
> wakeup. Also removed the platform only comment and switched thread discovery 
> to the same rendezvous the other converted jdb tests use
> 
> ---------
> - [x] I confirm that I make this contribution in accordance with the [OpenJDK 
> Interim AI Policy](https://openjdk.org/legal/ai).

One possibility is to get this change pushed now, but without the deadlock fix. 
That would require problem listing it, which would also require a new CR to 
address the deadlock fix. Or maybe the existing CR should be used for the 
deadlock issue, and you can create a new CR for pushing the virtual thread 
support sans the deadlock fix, and also have that CR/PR handle problem listing. 
No hurry on any of this. We can wait and see how the JVM fix discussion unfolds.

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

PR Comment: https://git.openjdk.org/jdk/pull/32267#issuecomment-5257018243

Reply via email to