Mike Miller created DAEMON-286:
----------------------------------
Summary: Race condition during stopping service
Key: DAEMON-286
URL: https://issues.apache.org/jira/browse/DAEMON-286
Project: Commons Daemon
Issue Type: Bug
Components: Procrun
Affects Versions: 1.0.13
Environment: Windows 7 64 bit
Reporter: Mike Miller
Priority: Minor
While debugging an issue where ProcRun NT Service hangs in the stopping state,
I noticed a race condition that causes 2 possible code paths during stopping.
serviceStop() calls afxJavaStart(...). I have observed apxJavaStart() returns
both TRUE & FALSE for a successful call to my stop method. I think this is
caused by a potential race condition in apxJavaStart() & __apxJavaWorkerThread.
The afxJavaStart() waits for lpJava->hWorkerSync event to be set and then
checks the lpJava->dwWorkerStatus. The race condition is that
__apxJavaWorkerThread sets the dwWorkerStatus to 1 and sets the event but there
is no guarantee that afxJavaStart() will see the 1. If the
__apxJavaWorkerThread completes the stop call before afxJavaStart has a chance
to check the dwWorkerStatus, the __apxJavaWorkerThread may have changed the
dwWorkerStatus back to 0. So depending on timing, it appears apxJavaStart()
could return either TRUE or FALSE even if the stop method behaved the same.
This race condition results in the stopService() method not always calling
afxjavaWait(hWorker, INFINITE,...).
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
