[M3devel] Win32 threads

[Jay K]

SignalObjectAndWait is widely mentioned as important to implementing condition variables on Win32, however it is documented as not being atomic.

Search the web:

http://msdn.microsoft.com/en-us/library/ms686293(VS.85).aspx

 

"Note that the "signal" and "wait" are not guaranteed to be performed as an atomic operation. Threads executing on other processors can observe the signaled state of the first object before the thread calling SignalObjectAndWait begins its wait on the second object."

 

So it isn't useful, right?

 

Thanks,

 - Jay

 


From: hosking at cs.purdue.edu
To: m3devel at elegosoft.com
Date: Tue, 29 Sep 2009 13:54:48 -0400
Subject: [M3devel] Win32 threads


I believe that the hangs we are seeing are because of the race present in AlertWait, which permit a thread to miss seeing an alert before waiting on the condition.  We want the thread only to wait if there has been no alert.  Unfortunately, the alert is only protected by the global mutex, which must be release before waiting.  Currently, we release the mutex, then wait, with the resulting race.  This requires a more complicated handshake than currently implemented.  Perhaps using SignalObjectAndWait instead as well as a Win32 mutex on each thread.  Similarly to ThreadPThread.m3.







Antony Hosking | Associate Professor | Computer Science | Purdue University
305 N. University Street | West Lafayette | IN 47907 | USA
Office +1 765 494 6001 | Mobile +1 765 427 5484

 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20090929/e023cae6/attachment-0002.html>