On 04.02.2013 18:25, Alan Bateman wrote:
On 04/02/2013 12:28, Alexey Utkin wrote:
Hi,
Please review the fix.
Bug description:
http://bugs.sun.com/view_bug.do?bug_id=8007454
https://jbs.oracle.com/bugs/browse/JDK-8007454
The suggested fix:
http://cr.openjdk.java.net/~uta/openjdk-webrevs/JDK-8007454/webrev.01/
Summary:
Old version of code for the [Java_java_lang_ProcessImpl_create]
function could not work right in some cases. The
[SetHandleInformation] call worked partially and just because
HANDLE_FLAG_INHERIT == TRUE == 1 and FALSE == 0.
As the parameter is a DWORD then it shouldn't be using a BOOLEAN so I
agree with that.
Can you explain why HANDLE_FLAG_PROTECT_FROM_CLOSE is needed?
Hmm... That was my mistake.
HANDLE_FLAG_PROTECT_FROM_CLOSE bit has to be removed from mask.
I mixed the variable names and did not understand clearly the handles
life circle.
My tests with redirected, direct and terminated input/output modes prove
that the number of opened
handles has GC-related saturation value.
New version:
http://cr.openjdk.java.net/~uta/openjdk-webrevs/JDK-8007454/webrev.02/
Thanks,
-uta
