Code review request: three native memory tracking related bugs (original) (raw)

Vladimir Kozlov vladimir.kozlov at oracle.com
Fri Jul 13 17:36:43 PDT 2012

Zhengyu Gu wrote:

7181989: NMT ON: Assertion failure when NMT checks thread's native CR: http://bugs.sun.com/bugdatabase/viewbug.do?bugid=7181989 Webrev: http://cr.openjdk.java.net/~zgu/7181989/webrev.00/

Good.

We try to assert Thread's stack base to ensure that Thread::recordstackbaseandsize() to record native stack to NMT, but there is scenario that the thread fails to start, which no native stack is created and should not be asserted. 7181986: NMT ON: Assertion failure when running jdi ExpiredRequestDeletionTest CR: http://bugs.sun.com/bugdatabase/viewbug.do?bugid=7181986 Webrev: http://cr.openjdk.java.net/~zgu/7181986/webrev.00/

Good.

This is a racing condition when C/C++ runtime exit handler is ran before NMT worker thread exits. The exit handler calls querylock's destructor while NMT worker thread is still holding it. The fix is to make querylock a heap object, instead of static object, but the drawback is that, it does not seem that querylock can be safely deleted. Also, I reassigned MemSnaphot lock and query lock, so they participate in the deadlock detection logic. 7182543: NMT ON: Aggregate a few NMT related bugs CRhttp://bugs.sun.com/bugdatabase/viewbug.do?bugid=7182543 Webrev: http://cr.openjdk.java.net/~zgu/7182543/webrev.00/

I think the next should be guarantee() instead of check since all MemRecorders should be deleted in preceding code:

if (MemRecorder::_instance_count == 0) {

Vladimir

- Fixed generationsinused calculation - Wait MemRecorder instance count to drop to zero before completely shutdown NMT - Added assertion for JavaThread in threadblocked state.

Thanks, -Zhengyu

More information about the hotspot-dev mailing list