OK, looks good. Good luck debugging! On Tue, Jul 5, 2016 at 1:19 PM, Roger Riggs <Roger.Riggs@oracle.com> wrote:

Hi Martin,

On 7/5/2016 3:38 PM, Martin Buchholz wrote:

Instead of unconditionally printing, why not print only when a zombie is discovered? All the information should still be available, I think...

I'm not sure what to trust since this has not been an issue until recently. In some local tests, the parent came back empty; that's probably normal in the case where the process is/has already terminated.

One recent change on Solaris just before the first appearance was: 8155808 Process.getInputStream().skip() method is faulty <https://bugs.openjdk.java.net/browse/JDK-8155808>

Which should only get involved when there is standard output. The Zombies test does not consume standard out. But I'll need to look into that to see how the exit value from the perl script could be affected by the change.

here's some misleading indentation...

+ p2.waitFor(); + throw new Error(zombies + " zombies!");

fixed, I did some of the editing on solaris with an emacs using hard tabs.

Thanks, Roger

On Tue, Jul 5, 2016 at 12:29 PM, Roger Riggs <Roger.Riggs@oracle.com> wrote:

...

Please review adding diagnostics to an intermittently failing test; java/lang/ProcessBuilder/Zombies.java.

Recently, the Zombies test has started intermittently reporting 1 Zombies errors on Solaris.

Webrev: http://cr.openjdk.java.net/~rriggs/webrev-zombies-8160848/

The original issue is open pending more information: https://bugs.openjdk.java.net/browse/JDK-8160151

Thanks, Roger