Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 63281 - Hang in getVMInstallType
Summary: Hang in getVMInstallType
Status: RESOLVED DUPLICATE of bug 52474
Alias: None
Product: JDT
Classification: Eclipse Project
Component: Debug (show other bugs)
Version: 3.0   Edit
Hardware: PC Windows 2000
: P3 normal (vote)
Target Milestone: ---   Edit
Assignee: JDT-Debug-Inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2004-05-20 14:27 EDT by David Williams CLA
Modified: 2004-05-25 12:06 EDT (History)
2 users (show)

See Also:


Attachments
ctrl-break results after above hang (21.27 KB, text/plain)
2004-05-20 14:28 EDT, David Williams CLA
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description David Williams CLA 2004-05-20 14:27:54 EDT
I got this hang on 200405140800 build, on windows 2000, using 
java version "1.4.2"
J9 - VM for the Java(TM) platform (build 2.1)
IBM J9SE VM (build 2.1, J2RE 1.4.2 IBM J9 build 20040408 (JIT enabled))

Seemed to be related to 
org/eclipse/jdt/launching/JavaRuntime.getVMInstallType

I got this hang during some automated performance tests, the first step of which 
is to automatically import some test files (java and others) from a CVS 
repository. 

I consider this 'normal' severity, since even is a hang, I'm not sure would 
occur during normal use. But thought I'd report and attach state of the threads 
incase it helps anyone spot a place to improve synchronization.
Comment 1 David Williams CLA 2004-05-20 14:28:55 EDT
Created attachment 10908 [details]
ctrl-break results after above hang
Comment 2 Darin Wright CLA 2004-05-21 11:08:58 EDT
I see the deadlock:

* A build job (Thread Worker-1) is trying to initialize the JRE container. It 
is blocked in getVMInstallType(String id), trying to retrieve all VM install 
types.
* The "All Types Caching" thread has also initialized the JRE container, and 
made it into getVMInstallTypes (owns the monitor Worker-1 is waiting on), and 
has triggered a VM change which triggers a Java core job, which is likely 
blocked on the build job.
Comment 3 Darin Wright CLA 2004-05-21 11:10:59 EDT
Adding John & Philippe as CC's. This is something we need to investigate for 
RC1. Since it does not appear to happen in practive, I do not think we need to 
fix this for M9. (Also note that this was found on an "old" build - May 14).
Comment 4 Darin Wright CLA 2004-05-21 11:23:26 EDT
Marking as RC1, for investigation.
Comment 5 Darin Wright CLA 2004-05-25 12:06:25 EDT

*** This bug has been marked as a duplicate of 52474 ***