db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Diogo Sousa (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (DERBY-6679) Possible Atomicity Violation in BaseMonitor.shutdown()
Date Tue, 29 Jul 2014 16:05:38 GMT

    [ https://issues.apache.org/jira/browse/DERBY-6679?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14077886#comment-14077886
] 

Diogo Sousa commented on DERBY-6679:
------------------------------------

That method only allow one caller, but other methods may remove elements from "services".
 In particular shutdown(Object serviceModule), in line 216, may do so, I think.

> Possible Atomicity Violation in BaseMonitor.shutdown()
> ------------------------------------------------------
>
>                 Key: DERBY-6679
>                 URL: https://issues.apache.org/jira/browse/DERBY-6679
>             Project: Derby
>          Issue Type: Bug
>          Components: Services
>    Affects Versions: 10.10.2.0
>         Environment: all
>            Reporter: Diogo Sousa
>            Priority: Minor
>
> I'm developing a tool for atomicity violation detection and I think it have found an
atomicity violations in derby.
> In org.apache.derby.impl.services.monitor.BaseMonitor there might be an atomicity violation
in method shutdown(), lines 181-206:
> {noformat}
> 		for ( ; ; ) {
> 			synchronized (this) {
>  181:    		position = services.size()  - 1;
> 			if (position == 0)
> 				break;
>                                ...
> 	        	}
>                        ...
> 	        }
> 	       ...
> 206:	      ((TopService) services.get(0)).shutdown();
> {noformat}
> Between the execution of lines 181 and 206 a concurrent thread may remove the only remaining
service, causing the access to "services" in line 206 to fail.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message