Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 353829 - BundleLoaderProxy.getBasicBundleLoader needs to be synchronized.
Summary: BundleLoaderProxy.getBasicBundleLoader needs to be synchronized.
Status: RESOLVED WONTFIX
Alias: None
Product: Equinox
Classification: Eclipse Project
Component: Framework (show other bugs)
Version: 3.7   Edit
Hardware: PC Windows XP
: P3 normal (vote)
Target Milestone: ---   Edit
Assignee: equinox.framework-inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-08-03 17:38 EDT by John Ross CLA
Modified: 2012-05-14 10:43 EDT (History)
1 user (show)

See Also:

Attachments
Adds synchronization to getBasicBundleLoader method. (1.02 KB, patch)
2011-08-03 17:46 EDT, John Ross CLA 		no flags Details | Diff
Alternate: Make loader variable volatile. (1.26 KB, patch)
2011-08-03 19:21 EDT, John Ross CLA 		no flags Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description John Ross CLA 2011-08-03 17:38:04 EDT 
getBasicBundleLoader accesses the loader variable which is set within the synchronized getBundleLoader0 method.
Comment 1 John Ross CLA 2011-08-03 17:46:29 EDT 
Created attachment 200862 [details]
Adds synchronization to getBasicBundleLoader method.
Comment 2 Thomas Watson CLA 2011-08-03 17:52:53 EDT 
We need to evaluate if adding such synchronization could introduce deadlock.  If getBasicBundleLoader is called as a result of calling getBundleLoader0 then deadlock may occur if there are dependency cycles between bundles.
Comment 3 John Ross CLA 2011-08-03 19:21:05 EDT 
Created attachment 200867 [details]
Alternate: Make loader variable volatile.
Comment 4 Thomas Watson CLA 2011-08-04 08:27:41 EDT 
(In reply to comment #3)
> Created attachment 200867 [details]
> Alternate: Make loader variable volatile.

I prefer the first approach if we can determine that it is safe.  I had not done the analysis to determine if there is deadlock potential with the first patch.
Comment 5 John Ross CLA 2011-08-04 09:58:23 EDT 
(In reply to comment #4)
> (In reply to comment #3)
> > Created attachment 200867 [details] [details]
> > Alternate: Make loader variable volatile.
> I prefer the first approach if we can determine that it is safe.  I had not
> done the analysis to determine if there is deadlock potential with the first
> patch.

I took a look for a bit yesterday and didn't see anything obvious but a second pair of eyes wouldn't hurt.
Comment 6 Thomas Watson CLA 2012-05-14 10:43:18 EDT 
No plans to fix this.  A general restructuring of the code is going to happen when we prototype and implement OSGi inter-op with Java 8 Module system.