Community
Participate
Working Groups
Build Identifier: 20100617-1415 randomly eclipse hangs-up. GUI does not allow to select anything and windows task manager says (not responding). During hanging eclipse uses 25% cpu (my system has 4 cpus) Reproduces only on my windows 7 x64. Reproducible: Sometimes Steps to Reproduce: 1. start eclipse 2. begin work 3. after some time (from 10 min to 5 hours) eclipse hangs-up.
Created attachment 176331 [details] dump of hanging-up eclipse
eclipse starts by user which has only user and debugging user privileges.
This does not look like CDT is involved. There is also no clear indication of a deadlock. The StringPoolJob looks a little suspicious being deeply nested in a recursion. Moving to Platform Resources for now.
The string pool job looks fine. It runs periodically but never does any locking. Vasiliy, it would help if you could attach a couple of dumps at different times so we can see if this is appearing consistently or if it is just noise. Tom, it is hanging here after the application has finished and it is trying to shut down. The main thread is blocked dispatching a synchronous framework event but I can't see what would be blocking this. Any ideas?
(In reply to comment #4) > Tom, it is hanging here after the application has finished and it is trying to > shut down. The main thread is blocked dispatching a synchronous framework event > but I can't see what would be blocking this. Any ideas? I don't either, perhaps there is an enormous # of listeners we are cycling through. This would explain why we pin one CPU for a period of time and exit after we have finished. But I have no idea why there would be so many service listeners.
Vasiliy, you say it hangs randomly, but do you have any ideas about what is going on right before it hangs? This trace looks like it is during shutdown - does it usually occur when you try to shutdown? Does it hang for awhile and then come back to life, or hang forever?
(In reply to comment #6) > Vasiliy, you say it hangs randomly, but do you have any ideas about what is > going on right before it hangs? This trace looks like it is during shutdown - > does it usually occur when you try to shutdown? Does it hang for awhile and > then come back to life, or hang forever? Often it happens when I try to open some modal dialogs (preferences, project properties) or browse between pages in preferences dialog. Sometimes it happens when autocompletion popup going to show. B.t.w On eclipse 3.7 I20100805-1700 the issue doesn't reproduce. 2 dumps are available at: http://narod.ru/disk/24032601000/javaw.zip.html http://narod.ru/disk/24032722000/javaw1.zip.html for download them enter captcha and press "Скачать" (means download) button.
Created attachment 177284 [details] Eclipse configuration
Created attachment 177285 [details] error log
The error log shows you're running out of PermGen space. See: http://wiki.eclipse.org/FAQ_How_do_I_increase_the_permgen_size_available_to_Eclipse%3F for instructions on how to increase this.
(In reply to comment #10) > The error log shows you're running out of PermGen space. See: > http://wiki.eclipse.org/FAQ_How_do_I_increase_the_permgen_size_available_to_Eclipse%3F > for instructions on how to increase this. I temporary decrease space for decrease dump size. Usually I use -Xmx2048m but it eclipse hangs with both (256 and 2048) Mb size values. First dump (http://narod.ru/disk/24032601000/javaw.zip.html ) were created with PermGen == 2048Mb. Second (http://narod.ru/disk/24032722000/javaw1.zip.html ) with PermGen == 256Mb. Oh, I forgot! When eclipse hangs it continue allocate memory and after 15-20 min. it allocate all memory and crashes.
I have _exactly_ the same problem as Vasiliy describes. After reading this report I tested with larger PermSize but no matter what I use, Eclipse hangs after some time. I sometimes get the problem when creating a new class and the editor is about to let me begin edit it, then it hangs. The strange thing is that I have another computer with same SW, ( Eclipse 3.6.1 and Java 1.6_u22 x64, but other HW configuration), that doesn't have this problem at all! This computer _does_ have other Eclipse versions and the 32bit Java installed also, but should that matter?
I noticed when the lock occur, the status bar at lower right says: "Computing additional info". This time the lock happened when I set a class to extend ViewPart.
Created attachment 181334 [details] A screen dump of the error window.
Ok, now I can reproduce this every time. But I wonder if anyone is interested since it's not reproducible in 3.7 There's an exit code when Eclipse is forced to exit. Unfortunately there's no way to copy the text in the window so I attached a image of it. (In previous post)
(In reply to comment #15) > Ok, now I can reproduce this every time. But I wonder if anyone is interested > since it's not reproducible in 3.7 > > There's an exit code when Eclipse is forced to exit. Unfortunately there's no > way to copy the text in the window so I attached a image of it. (In previous > post) I have the same issue. I have Win7 64 bits, AMD Athlon || X2, JDK 1.6 u22 64 bits, and Eclipse 3.6.1 64 bis version. This hang up everytime, eclipse doesn't work. My soultion was to use Eclipse 3.6.0 32 bits, with JDK 1.6 u20 32 bits, adding the folowing arguments in eclipse.ini: -vm C:\Program Files (x86)\Java\jdk1.6.0_21\bin\javaw.exe
Kenneth and Carlos, please provide a stack dump at the time it is hung. See: http://wiki.eclipse.org/How_to_report_a_deadlock Vasiliy, I attempted to view your logs, but the site appeared to be trying to install software on my computer (from bar.yandex.ru) so I closed it. Can you attach the log here, or at least the stack dump portion of the log?
Created attachment 181529 [details] dump1
Created attachment 181542 [details] dump2
Created attachment 181555 [details] Thread dump from Java VisualVM Here's my thread dump.
Created attachment 181556 [details] Stacktrace Here's a stacktrace from "-consolelog -debug" option when starting eclipse. Maybe not so much more info?
Created attachment 185391 [details] yet another thread dump This happens to me, too, on Windows 7 64-bit with both Eclipse 3.6.1 (eclipse-rcp-helios-SR1-win32-x86_64.zip) and 3.7M3 (eclipse-rcp-indigo-M3-win32-x86_64.zip). There is no reliable way to reproduce this issue. Last time it happened to me (see attached thread dump) I had opened the package explorer, selected two classes, and hit delete. Instead of showing the modal "Confirm Delete" dialog, Eclipse hung for ten minutes (using 25% CPU) before I killed it. After restarting, I repeated the exact same operation without any trouble. Other than Subversive, no additional plugins have been installed.
Hi, I am having the same issue, I have a windows 7 Intel Core2Due Dell Laptop, with eclipse-jee-helios-SR1-win32-x86_64. I just upgraded by jre to 6u23-windows-i586s, trying to resolve the issue - with no luck. It happens after around 2-5mins working, mostly when I open a java source code in the editor, or when highlighting something in the editor, put also at other random times too. Its making it unusable.
This is quickly turning into a meta-bug for any kind of deadlock on Windows 7 x64. Not all deadlocks are related. Unless your stack dump is similar to Vasiliy, with a hang due to a blocked synchronous event dispatch from Main thread during shutdown, please enter a separate bug. Eric Jain, yours in particular looks quite different. Please enter a bug against Platform SWT for your case. I'm changing the bug title to avoid this becoming a general dumping ground. Here is the stack trace of the deadlock being tracked here: "main" prio=6 tid=0x000000000052b800 nid=0x2e58 waiting on condition [0x000000000261e000] java.lang.Thread.State: RUNNABLE at org.eclipse.osgi.framework.eventmgr.ListenerQueue.dispatchEventSynchronous(ListenerQueue.java:148) at org.eclipse.osgi.internal.serviceregistry.ServiceRegistry.publishServiceEventPrivileged(ServiceRegistry.java:756) at org.eclipse.osgi.internal.serviceregistry.ServiceRegistry.publishServiceEvent(ServiceRegistry.java:711) at org.eclipse.osgi.internal.serviceregistry.ServiceRegistrationImpl.unregister(ServiceRegistrationImpl.java:225) at org.eclipse.equinox.internal.app.EclipseAppHandle.setAppStatus(EclipseAppHandle.java:163) - locked <0x0000000085b92b30> (a org.eclipse.equinox.internal.app.EclipseAppHandle) at org.eclipse.equinox.internal.app.EclipseAppHandle.setInternalResult(EclipseAppHandle.java:231)
(In reply to comment #21) > Created an attachment (id=181556) [details] > Stacktrace > > Here's a stacktrace from "-consolelog -debug" option when starting eclipse. > Maybe not so much more info? I can at least confirm this is the same as Vasiliy's original bug.
Just to provide another data point, I am running Windows 7, x86-64, Sun Java 6u22. I use a variety of 3.6.x, 3.7.x, and 4.x stream builds and I haven't seen this. Since this happens during shutdown, my theory is there is a root cause we are not seeing (perhaps OutOfMemoryError or similar system error). This is causing the system to shutdown, and is also putting the event queue in a bad state. We try to shutdown in response to the error, but hit an infinite loop or hang during event dispatch that prevents shutdown from completing normally. For anyone seeing this, check your .log file for unrelated errors that occur around the same time as the hang, and attach them here.
(In reply to comment #24) > Eric Jain, yours in particular looks quite different. Please enter a bug > against Platform SWT for your case. Created Bug 332965
Steps to reproduce. Work always. - start eclipse in a new workspace - create a plug-in project. Set name and leave all else as default. - switch to plug-in perspective - create a Class in same package as Activator. (right-click and New) - new class opens in editor. - type after class name 'extends View' and hit CTRL+SPACE to use content assist. For me this will cause Eclipse to hang. No '.log' file is created. At least I can't find it...
Is this bug in pipe for 3.6.2 or is forgotten?
The dumps here make it look like a permgen space issue from https://bugs.eclipse.org/bugs/attachment.cgi?id=181556: Heap PSYoungGen total 349376K, used 800K [0x0000000035490000, 0x000000004a9e0000, 0x000000004a9e0000) eden space 349248K, 0% used [0x0000000035490000,0x00000000355581a8,0x000000004a9a0000) from space 128K, 0% used [0x000000004a9a0000,0x000000004a9a0000,0x000000004a9c0000) to space 128K, 0% used [0x000000004a9c0000,0x000000004a9c0000,0x000000004a9e0000) PSOldGen total 699072K, used 43735K [0x000000000a9e0000, 0x0000000035490000, 0x0000000035490000) object space 699072K, 6% used [0x000000000a9e0000,0x000000000d495da0,0x0000000035490000) PSPermGen total 86016K, used 86016K [0x00000000055e0000, 0x000000000a9e0000, 0x000000000a9e0000) object space 86016K, 100% used [0x00000000055e0000,0x000000000a9e0000,0x000000000a9e0000) Notice that PSPermGen is at 100%. We have seen lots of cases where running out of PermGen space results in hangs (where it should result in OOME). There is nothing the framework can do here. I do wonder if the launcher is setting up the options to set the permgen space appropriately though. It looks like it says 86016K. I think the eclipse.ini should be setting this to a max of256m with the following lines: --launcher.XXMaxPermSize 256m Can you please check your eclipse.ini file for these settings.
Did not happen with 4.3 (Kepler) so I think this issue can be closed.
(In reply to comment #31) > Did not happen with 4.3 (Kepler) so I think this issue can be closed. Not that I am aware of. It seemed to be a permgen issue in the VM causing a deadlock at the VM level.
This bug hasn't had any activity in quite some time. Maybe the problem got resolved, was a duplicate of something else, or became less pressing for some reason - or maybe it's still relevant but just hasn't been looked at yet. If you have further information on the current state of the bug, please add it. The information can be, for example, that the problem still occurs, that you still want the feature, that more information is needed, or that the bug is (for whatever reason) no longer relevant. -- The automated Eclipse Genie.
I think the more recent Java versions have helped out here. Closing as worksforme.
