On Thu, 4 Jun 2026 05:02:51 GMT, David Holmes <[email protected]> wrote:

> I filed this enhancement request out of frustration, but I'm not sure that 
> accepting any frame from java/jdk/sun.* is the right way to deal with this as 
> it could hide completely unexpected situations.
> 
> That said, I am struggling to see exactly what the stack inspections are for 
> ...

Yeah fair point, I had the same thought. from what I understand, the stack 
inspections verify the thread went through the expected code path (bringState 
-> recursion ->run) and that recursion depth is correct, though checkThreadInfo 
already validates the sleeping behavior independently via thread state and wait 
count

Based on my understanding the core issue is SleepingThread had 12 hardcoded JDK 
method names that broke on every Thread.sleep change. a narrower filter would 
just bring back the same problem. tests still reject unexpected non-JDK frames 
and verify stack depth and only JDK implementation details get accepted, which 
is what the bug asks for. same pattern as strace001.java from your JDK-8330302 
fix.

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

PR Comment: https://git.openjdk.org/jdk/pull/31357#issuecomment-4620446053

Reply via email to