Java crash in libglib-2.0 after upgrade from 13.04 to 13.10

[Claude Durocher]

Running smartgit 4.6.4 on 13.10 64 bits. After registering the product, smartgit crash when trying to open a new repository. Java error log :

# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007fa59061f9c0, pid=12494, tid=140349308167936
#
# JRE version: 7.0_25-b30
# Java VM: OpenJDK 64-Bit Server VM (23.7-b01 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libglib-2.0.so.0+0x389c0] g_str_hash+0x0

I tried different version of Java (Oracle v7 and v6 jre) with same result. Also, Eclipse display blank menus so there's a general java problem with displays.

[Eclipse-p (eclipse-p)]

Created attachment 236604
crash log

A user on Linux reported native crashs for SmartSVN (SWT-based application):

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j org.eclipse.swt.internal.gtk.OS._gtk_widget_show(J)V+0
j org.eclipse.swt.internal.gtk.OS.gtk_widget_show(J)V+8
j org.eclipse.swt.widgets.MenuItem.createHandle(I)V+249
j org.eclipse.swt.widgets.Widget.createWidget(I)V+2
j org.eclipse.swt.widgets.MenuItem.<init>(Lorg/eclipse/swt/widgets/Menu;I)V+19

[Claude Durocher (claude-d)]

I found a workaround : if I start smartgit like this, it won't crash :

UBUNTU_MENUPROXY= ./smartgit.sh

So the issue seem related to appmenu.

[Quinn Balazs (qbalazs)]

There should be a file containing the full context of the OpenJDK issue. All you added here is what the problematic frame, which is helpful in context, but not particularly on its own. See if you can't upload the error log, and once you have, set this back to new and someone will take a look.

Quinn Balazs

[Claude Durocher (claude-d)]

Here's the full log:

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007ff34dd759c0, pid=13319, tid=140683267766016
#
# JRE version: 7.0_25-b30
# Java VM: OpenJDK 64-Bit Server VM (23.7-b01 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libglib-2.0.so.0+0x389c0] g_str_hash+0x0
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# If you would like to submit a bug report, please include
# instructions on how to reproduce the bug and visit:
# https://bugs.launchpad.net/ubuntu/+source/openjdk-7/
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

--------------- T H R E A D ---------------

Current thread (0x00007ff358009000): JavaThread "main" [_thread_in_native, id=13320, stack(0x00007ff360237000,0x00007ff360338000)]

siginfo:si_signo=SIGSEGV: si_errno=0, si_code=1 (SEGV_MAPERR), si_addr=0x0000000000000000

Registers:
RAX=0x00007ff358a9d230, RBX=0x00007ff358a9d230, RCX=0x0000000000000000, RDX=0x00007ff358a9d230
RSP=0x00007ff3603344a8, RBP=0x00007ff3590872a0, RSI=0x0000000000000000, RDI=0x0000000000000000
R8 =0x00007ff3590cc1a0, R9 =0x00007ff35825ed52, R10=0xaaaaaaaaaaaaaaab, R11=0x00007ff3585e1100
R12=0x00007ff358ad72a0, R13=0x00007ff358ad72a0, R14=0x0000000000000000, R15=0x0000000000000000
RIP=0x00007ff34dd759c0, EFLAGS=0x0000000000010206, CSGSFS=0x0000000000000033, ERR=0x0000000000000004
  TRAPNO=0x000000000000000e

Top of Stack: (sp=0x00007ff3603344a8)
0x00007ff3603344a8: 00007ff34dd74b30 0000000000000000
0x00007ff3603344b8: 0000000000000000 00007ff358a9d230
0x00007ff3603344c8: 0000000058fc9890 0000000000000000
0x00007ff3603344d8: 00007ff358a9d230 00007ff358a6ff70
0x00007ff3603344e8: 00007ff358ad72a0 00007ff358ad72a0
0x00007ff3603344f8: 0000000000000000 0000000000000000
0x00007ff360334508: 00007ff3173bac62 00007ff358ef5870
0x00007ff360334518: 0000000000000012 00007ff358b31d10
0x00007ff360334528: 00007ff34dd9e8aa 00007ff3173b4c70
0x00007ff360334538: 00007ff35881c4c0 00007ff358a9d230
0x00007ff360334548: 00007ff358ef5870 0000000000000012
0x00007ff360334558: 00007ff358b31d10 00007ff358b100f0
0x00007ff360334568: 00007ff3173b53a2 0000000000000002
0x00007ff360334578: 0000000000000012 00007ff360334790
0x00007ff360334588: 00007ff34dd9ebb3 00007ff358a9d230
0x00007ff360334598: 00007ff34c8e8b00 00007ff358a9d230
0x00007ff3603345a8: 00007ff35881c4c0 0000000000000002
0x00007ff3603345b8: 00007ff3603347f0 00007ff360334790
0x00007ff3603345c8: 00007ff3173b6e46 00007ff358a9d230
0x00007ff3603345d8: 00007ff3173b8921 00007ff358a61a90
0x00007ff3603345e8: 00007ff358a61a90 00007ff3603347f0
0x00007ff3603345f8: 00007ff358570ac0 00007ff3603347f0
0x00007ff360334608: 00007ff359113b90 00007ff358a9d230
0x00007ff360334618: 00007ff3585a0890 00007ff3590f1d70
0x00007ff360334628: 00007ff34e050d50 0000000000000000
0x00007ff360334638: 00007ff359113b90 0000000000000002
0x00007ff360334648: 00007ff3603347f0 00007ff360334790
0x00007ff360334658: 00007ff34e04e188 0000000000000000
0x00007ff360334...

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Eclipse-p (eclipse-p)]

The version.txt file inside the swt.jar shows "version 4.331".

[Eclipse-p (eclipse-p)]

Ubuntu 13.10

[Akurtakov (akurtakov)]

Is there any chance that you can give a snippet to reproduce the problem?

[Eclipse-p (eclipse-p)]

Created attachment 236645
crash log

[Eclipse-p (eclipse-p)]

Created attachment 236646
crash log

[Eclipse-p (eclipse-p)]

Created attachment 236647
crash log

[Eclipse-p (eclipse-p)]

The crashes happen after completely different steps, the last crash occurred directly after starting while the splash screen was showing. Following errors are locked multiple times to the console:

(SmartGit/Hg:18473): GLib-GObject-CRITICAL **: g_closure_add_invalidate_notifier: assertion 'closure->n_inotifiers < CLOSURE_MAX_N_INOTIFIERS' failed

[Akurtakov (akurtakov)]

Thomas, I'm asking for some small pure Swt code snippet runnable separately (aka class with main method) that shows the problem. The crash and the invalidate notifier are not related (me is 99% sure) and it's discussed in bug 407077.

[Eclipse-p (eclipse-p)]

Please try following code:

  final Display display = new Display();

  final Shell shell = new Shell(display);
  shell.setLayout(new FillLayout());

  final Menu menuBar = new Menu(shell, SWT.BAR);
  final MenuItem fileMenuBarItem = createMenuItem(SWT.CASCADE, "File", menuBar);
  final Menu fileMenu = new Menu(shell, SWT.DROP_DOWN);
  fileMenuBarItem.setMenu(fileMenu);
  shell.setMenuBar(menuBar);

  shell.addListener(SWT.Activate, new Listener() {
   @Override
   public void handleEvent(Event event) {
    System.err.println("rebuilding menu");

    final MenuItem[] items = fileMenu.getItems();
    for (MenuItem item : items) {
     item.dispose();
    }

    createMenuItem(SWT.PUSH, "Push", fileMenu);
    createMenuItem(SWT.CHECK, "Check", fileMenu);
    createMenuItem(SWT.RADIO, "Radio", fileMenu);
   }
  });

  final Menu menu = new Menu(shell, SWT.POP_UP);
  createMenuItem(SWT.PUSH, "Push", menu);
  createMenuItem(SWT.CHECK, "Check", menu);
  createMenuItem(SWT.RADIO, "Radio", menu);

  final Button button = new Button(shell, SWT.PUSH);
  button.setText("Click Me");
  button.addListener(SWT.Selection, new Listener() {
   @Override
   public void handleEvent(Event event) {
    final Point location = display.map(button.getParent(), null, button.getLocation());
    final Point size = button.getSize();
    menu.setLocation(location.x, location.y + size.y);
    menu.setVisible(true);
   }
  });

  shell.setSize(shell.computeSize(SWT.DEFAULT, SWT.DEFAULT));
  shell.open();

  while (!shell.isDisposed()) {
   if (!display.readAndDispatch()) {
    display.sleep();
   }
  }

  menu.dispose();
  display.dispose();

And open some other windows, especially dialogs, click the button.

[Akurtakov (akurtakov)]

Missed the createMenuItem method?

[Eclipse-p (eclipse-p)]

Sorry:

 private static MenuItem createMenuItem(int style, String text, Menu menu) {
  final MenuItem item = new MenuItem(menu, style);
  item.setText(text);
  return item;
 }

[Akurtakov (akurtakov)]

I played with it a lot but wasn't able to reproduce on fedora 19 system (gtk 2.24.22). Does it crash for your everytime you do certain action? Can you simulate the crash in the snippet?

[Eclipse-p (eclipse-p)]

I only can reproduce the problem on Ubuntu 13.10, not on Ubuntu 12.04. It only has us been reported against Ubuntu 13.10, not against previous versions, though the code has not changed there for a long time. I guess, there is something specific in Ubuntu 13.10 which may not be reproducible on other systems.

[Eclipse-p (eclipse-p)]

According to https://twitter.com/md__5/status/391352267313405952 the bug seems to be related to UBUNTU_MENUPROXY.

[Akurtakov (akurtakov)]

Further investigations would need to be done by someone having the time to install and test the Ubuntu version in question. I don't have such time but I would be happy to review a patch if provided.

[Dac Chartrand (conner-bw)]

I have the same bug using SmartSVN (same vendor, historically at least), and their support narrowed it down to a problem with "Native crash in org.eclipse.swt.internal.gtk.OS._gtk_widget_show," see:

https://bugs.eclipse.org/bugs/show_bug.cgi?id=419729

Dear: Java devs, please read thread above, Comment #15 in particular.

Cheers.

[Quinn Balazs (qbalazs)]

This bug has been filed in Launchpad (Ubuntu Bug Tracker) as lp#1241101, filed against the java-common package in Ubuntu.

User using Ubuntu 13.10 has reported crashes using SmartGit v4.6.4

# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007fa59061f9c0, pid=12494, tid=140349308167936
#
# JRE version: 7.0_25-b30
# Java VM: OpenJDK 64-Bit Server VM (23.7-b01 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libglib-2.0.so.0+0x389c0] g_str_hash+0x0

The other issue addressed in the LP bug is the SmartSVN issue discussed here.

This issue in LP can be found here https://bugs.launchpad.net/ubuntu/+source/java-common/+bug/1241101 .

[Quinn Balazs (qbalazs)]

Tracking this issue in eclipse-bugs.

Quinn Balazs

[Quinn Balazs (qbalazs)]

In the previous comment I attached a different portion of the log file than I had intended to. Here is the relevant portion.

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j org.eclipse.swt.internal.gtk.OS._gtk_widget_show(J)V+0
j org.eclipse.swt.internal.gtk.OS.gtk_widget_show(J)V+8
j org.eclipse.swt.widgets.MenuItem.createHandle(I)V+249
j org.eclipse.swt.widgets.Widget.createWidget(I)V+2
j org.eclipse.swt.widgets.MenuItem.<init>(Lorg/eclipse/swt/widgets/Menu;I)V+19

[Dac Chartrand (conner-bw)]

Also see bug #1208019

Not a crash per say, but some coordinated effort to fix global menus that were working fine in 13.04. but must now be worked around setting UBUNTU_MENUPROXY would be prudent.

Best regards,

[Quinn Balazs (qbalazs)]

Posted here on behalf of Julien Aubin.

[Cristiano Cavallari (cristiano-cavallari)]

I have the same problem with Smartsvn.

[Quinn Balazs (qbalazs)]

Strangely, something that has worked for people in a few other situations has been to change GTK2 appearance to anything other than oxygen-gtk. If someone wouldn't mind checking that out, I'd be interested to see if that workaround works here, or if there are multiple issues at hand here.

[Quinn Balazs (qbalazs)]

I've taken a look at this, and seemingly randomly*, java frames flipflop between what we've seen here, and the

j org.eclipse.swt.internal.gtk.OS._g_object_get_qdata(JI)J+0
j org.eclipse.swt.internal.gtk.OS.g_object_get_qdata(JI)J+9
j org.eclipse.swt.widgets.Display.removeWidget(J)Lorg/eclipse/swt/widgets/Widget;+19
j org.eclipse.swt.widgets.Combo.deregister()V+21
j org.eclipse.swt.widgets.Widget.releaseWidget()V+1

that we've seen in bug #1205452. Changing GTK2 appearance will on occasion clear up the issue, but works nowhere near 100% of the time. Once I've done a bit more digging and am sure that these issues are at their roots the same, bug #1205452 can be marked as a duplicate of this issue, and maybe a bug pattern can be written for this issue (23 affected in #1205452, 16 affected here, and most people using Eclipse or a derived product will run into this issue)

*Nothing is random, I'm working on associating these differences to user actions.

[Netziro (netziro)]

Quinn, i've got the same scenario and changing GTK2 appearance did solve the issue. I had 2 machines affected and both solved changing the GTK2 opt

[Cristiano Cavallari (cristiano-cavallari)]

I change GTK apperance using gtk-theme-switch2 command, I can see the change on the othere applications (es. firefox) but smart svn crash with the same error.

Stack: [0x00007f275beb7000,0x00007f275bfb8000], sp=0x00007f275bfb3be8, free space=1010k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libglib-2.0.so.0+0x389c0] __float128+0x0

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j org.eclipse.swt.internal.gtk.OS._gtk_widget_show(J)V+0
j org.eclipse.swt.internal.gtk.OS.gtk_widget_show(J)V+8
j org.eclipse.swt.widgets.MenuItem.createHandle(I)V+249
j org.eclipse.swt.widgets.Widget.createWidget(I)V+2

[Netziro (netziro)]

in my case that's the stack

Stack: [0x00007fa5c172b000,0x00007fa5c182c000], sp=0x00007fa5c18281f0, free space=1012k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libgobject-2.0.so.0+0x19528] g_object_get_qdata+0x18

[error occurred during error reporting (printing native stack), id 0xb]

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
J org.eclipse.swt.internal.gtk.OS._g_object_get_qdata(JI)J
J org.eclipse.swt.internal.gtk.OS.g_object_get_qdata(JI)J
j org.eclipse.swt.widgets.Display.removeWidget(J)Lorg/eclipse/swt/widgets/Widget;+19
j org.eclipse.swt.widgets.Combo.deregister()V+21
j org.eclipse.swt.widgets.Widget.releaseWidget()V+1
j org.eclipse.swt.widgets.Control.releaseWidget()V+1
j org.eclipse.swt.widgets.Composite.releaseWidget()V+1
j org.eclipse.swt.widgets.Combo.releaseWidget()V+1
j org.eclipse.swt.widgets.Widget.release(Z)V+82
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.TabFolder.releaseChildren(Z)V+54
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Composite.releaseChildren(Z)V+31
j org.eclipse.swt.widgets.Canvas.releaseChildren(Z)V+42
j org.eclipse.swt.widgets.Decorations.releaseChildren(Z)V+22
j org.eclipse.swt.widgets.Shell.releaseChildren(Z)V+100
j org.eclipse.swt.widgets.Widget.release(Z)V+38
j org.eclipse.swt.widgets.Control.release(Z)V+94
j org.eclipse.swt.widgets.Widget.dispose()V+23
j org.eclipse.swt.widgets.Shell.dispose()V+20
j org.eclipse.jface.window.Window.close()Z+65
j org.eclipse.jface.dialogs.Dialog.close()Z+26
j org.eclipse.search.internal.ui.SearchDialog.close()Z+40
j org.eclipse.search.internal.ui.util.ExtendedDialogWindow.buttonPressed(I)V+46
j org.eclipse.jface.dialogs.Dialog$2.widgetSelected(Lorg/eclipse/swt/events/SelectionEvent;)V+17
j org.eclipse.swt.widgets.TypedListener.handleEvent(Lorg/eclipse/swt/widgets/Event;)V+1133
J org.eclipse.swt.widgets.EventTable.sendEvent(Lorg/eclipse/swt/widgets/Event;)V
j org.eclipse.swt.widgets.Widget.sendEvent(Lorg/eclipse/swt/widgets/Event;)V+25
j org.eclipse.swt.widgets.Display.runDeferredEvents()Z+92
j org.eclipse.swt.widgets.Display.readAndDispatch()Z+46
j org.eclipse.jface.window.Window.runEventLoop(Lorg/eclipse/swt/widgets/Shell;)V+23

[spectator (lists-a)]

FWIW, eclipse started crashing after upgrade to 13.10, always # Problematic frame: # C [libgobject-2.0.so.0+0x16bfe]
Upon reading #29 I changed GTK2 appearance a couple of times from oxygen-gtk to ambiance etc. For me this seems to touch the problem. Oxygen crashes, others do not. Reproducible.

[Scot Bernard (pablo-bueti)]

True, as the twitter says changing the .desktop file exec line to:

Exec=env UBUNTU_MENUPROXY=0 "[yoursmartgitbinpath]/smartgithg.sh" %u

Works, and runs with menues not integrated into unity panel.

[Cristiano Cavallari (cristiano-cavallari)]

Removing the two environment variables Smart Svn don't crash anymore.

GTK_IM_MODULE="ibus"
GTK_MODULES="overlay-scrollbar:unity-gtk-module"

For me that solved the problem.

[Cristiano Cavallari (cristiano-cavallari)]

It's sufficient remove only the variable
GTK_MODULES="overlay-scrollbar:unity-gtk-module"

[aanno (thomas-pasch)]

*** Bug 420499 has been marked as a duplicate of this bug. ***

[aanno (thomas-pasch)]

As I already mentioned on #1245468 I could not verify that changing GTK_MODULES, UBUNTU_MENUPROXY, or GTK_IM_MODULE helps in any way.

However, I could verify that the problem GOES AWAYS IN KUBUNTU/KDE when doing:

System Settings -> Application Themes -> GTK -> Choose GTK2 Theme

Choose 'Radiance' instead of 'oxygen-gtk'

[José Carlos Valero (sirjoseca)]

When changing UBUNTU_MENUPROXY this bug disappear to me. At least in SmartSVN.

[aanno (thomas-pasch)]

Dear sirjoseca,

well I guess it is solved for SmartSVN. But I want to use eclipse...

[José Carlos Valero (sirjoseca)]

Well, indeed i have to use UBUNTU_MENUPROXY= XXX in both, eclipse and SmartSVN to get them working properly (SmartSVN closes unexpectedly, and eclipse doesn't show menus any other way), and I have checked that in both cases disabling ubuntu global menu solve this issue, at least in my computer.

Maybe my comment is pointless, just to try.

Thanks for your reply aano.

[aanno (thomas-pasch)]

Dear sirjoseca,

I rechecked this carefully. For me (as a Kubuntu/KDE user) using

export UBUNTU_MENUPROXY=''

and afterwards starting eclipse makes things A LOT better but the problem does NOT go away. I've still got spurious Java VM crashes, most commonly with SWT 'cancel' buttons. But I have to admit that this is relatively rare.

However, with

System Settings -> Application Themes -> GTK -> Choose GTK2 Theme

and choosing 'Radiance' instead of 'oxygen-gtk' the problem is gone COMPLETELY.

[Quinn Balazs (qbalazs)]

Can you confirm/nonconfirm that this issue is resolved if a desktop environment other than Unity is used?

[aanno (thomas-pasch)]

At present, I could only speak for kubuntu/kde. The problem is gone when using 'Radiance' as 'GTK2 Theme'. However, if I understand the KDE guys right, this simply means that libglib is NOT used. Per consequence the error cannot occur...

[Quinn Balazs (qbalazs)]

I believe that to be correct. If I get a bit of spare time today, i'll mess around with different desktop environments.

[Mrbublor (mrbublor)]

I am using Open Suse 13.1 and I can reproduce this every time. I did not see this until I installed 13.1 from a fresh install, then install the ADT bundle from Android developer site.

Per instructions found on the Ubuntu site, I can change the GTK2 theme from oxygen to Raleigh and the crash goes away.

[Sebastien Senechal (altagir)]

you can setup that directly for adt/eclipse, without changing the style globally on KDE:
launch :
$ GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc /opt/adt-bundle-linux-x86_64-20130917/eclipse/eclipse

[gerard ziemski (gerard-ziemski)]

https://bugs.launchpad.net/bugs/1234253 is a similar issue, but none of the workarounds provided here work. The only workaround that does work in that case is to lower the CPU affinity to 1.

[Volkan Gezer (volkangezer)]

For KDE users:

Follow this: System Settings -> Application View -> GTK -> GTK 2 Theme Raleigh instead of oxygen.

[Christophe Chisogne (cchisogne)]

On Kubuntu 13.10 (64bit), the problem is easy to reproduce when using Eclipse from the latest Android bundle and a JDK 1.7 (tested with jdk1.7.0_25-x64).

Switching GTK2 theme from oxygen-gtk to Raleigh is working in my configuration (glogal change or GTK2_RC_FILES variable).

On my box, the crash is very often in the same frame (libgobject2 g_object_get_qdata). Running gdb on the core dump does not add much information.

  # SIGSEGV (0xb) at pc=0x00007f605618d528, pid=18070, tid=140052081948416
  #
  # JRE version: 7.0_25-b15
  # Java VM: Java HotSpot(TM) 64-Bit Server VM (23.25-b01 mixed mode linux-amd64 compressed oops)
  # Problematic frame:
  # C [libgobject-2.0.so.0+0x19528] g_object_get_qdata+0x18

[Quinn Balazs (qbalazs)]

Gtk+ 2.0 is definitely a component of this issue.

[Akurtakov (akurtakov)]

(In reply to Joe Bublor from comment #19)
> I am using Open Suse 13.1 and I can reproduce this every time. I did not
> see this until I installed 13.1 from a fresh install, then install the ADT
> bundle from Android developer site.
>
> Per instructions found on the Ubuntu site, I can change the GTK2 theme from
> oxygen to Raleigh and the crash goes away.

Joe, I would dare to say it's a bug in oxygen theme. We had few such on Fedora too especially with oxygeng-gtk2 theme.

[Peter Kelder (peter-kelder-quicknet)]

Can reproduce it when using the Archi application. The Archi application is an Enterprise Architecture modeling tool according the Archimate language.
Choosing the oxygen-gtk2 theme it crashes as soon as you do anything in the UI, like:
- closing a tab
- dragging a object on the convas
- ...

[Hendy Irawan (ceefour)]

Bug report in KDE: https://bugs.kde.org/show_bug.cgi?id=324438

[Tom B (0-tom-4)]

That's a different bug. I can confirm that does fix several issues with oxygen-gtk but not the g_object_get_qdata crash.

See https://bugs.kde.org/show_bug.cgi?id=329814 for the g_object_get_qdata crash.

[Tom B (0-tom-4)]

Has anyone managed to track the package which is the source of this bug?

Here's my understanding so far:

It's not an Eclipse bug as the same version of Eclipse running on Ubuntu 13.04 does not suffer the problem

It doesn't appear to stem directly from oxygen-gtk as the QtCurve theme suffers from the same problem which hints that it's something in a library that QtCurve and oxygen-gtk both rely on that other themes do not.

It's unlikely to be a Java bug as it affects both OpenJDK and Oracle's JDK.

A similar problem from 2010 is mentioned here: https://bugs.launchpad.net/ubuntu/+source/eclipse/+bug/598371

It's very similar, but this was certainly fixed somehow since then because 13.04 was unaffected.

[Quinn Balazs (qbalazs)]

We've been concentrating on gtk, which is a component. I'd like to suggest taking a fresh look at Unity's global menu component. There we're changes made to the global menu in 13.10, which may account for this not showing up in 13.04. I'm going to mess around with that side of things and see if I can get anything worthwhile from it.

[Quinn Balazs (qbalazs)]

Rolling back to the version of Unity from Raring updates (7.0.0daily13.06.19~13.04-0ubuntu1) I cannot replicate this issue, whereas with 7.1.2+13.10.20131014.1-0ubuntu1 I am able to produce a crash. If someone could verify/disprove a connection between Unity and this issue that would be very helpful. Hesitant to throw Unity up as a potential package at fault until we are able to narrow down what is actually going on.

I do agree with Tom that Java-Common be removed as a potential culprit, as it seems to span several Java versions and affects both OpenJDK and Oracle's JDK, but for the same reasons, I'd rather hang on there for a bit. Gtk seems to be part of the issue, but may (if Unity doesn't end as a dead end) tie into Unity at some point. Regarding shared libs between QtCurve and gtk2-engines-oxygen, there isn't a lot that stands out. There is of course libc6, qt4-x11, and kde4libs. It's entirely possible that there was a change in one of those libraries that is causing this behavior.

For the moment if you plan to test Unity 7.0.0 use the .deb file for your architecture.

[Claude Durocher (claude-d)]

When I opened the ticket, I also mentioned a workaround (set env. var. UBUNTU_MENUPROXY to blank) before starting a java application. This solves the problem for me for smartgit and eclipse.

This is clearly an indication of a problem with Unity menus.

[Andreas Ntaflos (daff)]

Just so that this doesn't get overlooked: I have this problem with Apache Directory Studio (an Eclipse RCP application) running on Kubuntu 13.10 using KDE 4.12 and the oxygen-gtk engine. It is definitely not related to Unity directly, though I don't know how many libraries or how much code oxygen-gtk and Unity share.

In any case, the workaround of running (in my case) GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc ApacheDirectoryStudio still applies.

[Netziro (netziro)]

Totally agree with Andreas.
I've got the same persistent issue using KDE and oxygen theme.
Gotta be some shared library

Alessio Nobile
mail sent from Android
On 28 Jan 2014 16:31, "Andreas Ntaflos" <email address hidden> wrote:

> Just so that this doesn't get overlooked: I have this problem with
> Apache Directory Studio (an Eclipse RCP application) running on Kubuntu
> 13.10 using KDE 4.12 and the oxygen-gtk engine. It is definitely not
> related to Unity directly, though I don't know how many libraries or how
> much code oxygen-gtk and Unity share.
>
> In any case, the workaround of running (in my case)
> GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc
> ApacheDirectoryStudio still applies.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1241101
>
> Title:
> Java crash in libglib-2.0 after upgrade from 13.04 to 13.10
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/eclipse/+bug/1241101/+subscriptions
>

[auspex (auspex)]

The GTK2_RC_FILES theme workaround doesn't help me, nor does UBUNTU_MENUPROXY.

[Tom B (0-tom-4)]

I'm not sure this is helpful or not but:

I don't have Unity installed. I'm running Linux Mint which is built from Ubuntu and I'm running KDE. Whether that rules out a unity bug or not I don't know.

Setting UBUNTU_MENUPROXY has no effect at all.

Using GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc works, but it's the same workaround as previously because it just disables the oxygen-gtk theme.

[Andreas Ntaflos (daff)]

@auspex: the GTK2_RC_FILES workaround should work for this bug, how are you calling it? Should look like this:

$ GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc /path/to/eclipse/or/whatever

Alternatively you can set the GTK2 theme to Raleigh globally and try starting your crashing application then.

If neither of these helps then you are probably experiencing a different bug, or at least a very different manifestation that apparently hasn't been encountered yet, going by the information in this thread.

[Cristian Adam (cristianadam)]

Using GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc worked also for me using Kubuntu 13.10 64bit with oxygen-gtk theme.

It's an annoying bug because evertying seems to work and at certain operations the application (based on Eclipse) crashes.

[MiXiM (mixim)]

As Mr. Adam says above, i use the same environment, and the same error occurs at certain operations when i use the oxygen-gtk theme.

The workaround with the GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc works for me also, but looks not very good :(

[Claude Durocher (claude-d)]

Just a quick note : I ran a test on Trusty (30-jan build) and the bug is still present.

[David V (charon00)]

Just wanted to comment that this bug is also present in the KDE environment, but not Gnome, on a 64-bit Fedora 20 system. I encountered it when using an SDK leveraged from Eclipse. The SDK crashes frequently when I click the "OK" button on a window. The problematic frame is:

C [libgobject-2.0.so.0+0x19b48] g_object_get_qdata+0x18

Changing the GTK+ Appearance from Oxygen to Raleigh as others have mentioned is working for me (so far).

[auspex (auspex)]

@daff Why would you assume a different bug? I have tried it with both forms
    UBUNTU_MENUPROXY='' GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc ./adt-bundle-linux-x86_64-20131030/eclipse/eclipse -data .
and setting the theme in System Settings, and I _can_ use Eclipse for a while, but sooner or later I'm going to crash in exactly the same location `# C [libgobject-2.0.so.0+0x19528] g_object_get_qdata+0x18`

[dapaldo (dpalomar)]

Same issue in Ubuntu 14.04 (Unity).
Tested with eclipse 3.8.2 and 4.3.1 with oracle-java7 and openjdk-7.
After install and first time start, eclipse freeze and crash, second start works fine, but when download a project from git, or download a maven project or starting tomcat, eclipse crash.

[Antonio Ferraro (anfe67)]

I have my own RCP application, built with Eclipse Helios, for multiple platforms. The problem happens for me when I test the application build for Linux x86-gtk. Platform: Kubuntu 13.10 (virtualbox machine), JRE: 1.6u45. I get the crash while I use the preference pages and hit the OK button. No .log is found in workspace/.metadata, just the hs_err_pidxxxx.log in the app install directory.

The Solaris sparc/intel and win32/64 builds of the app work fine in Solaris, and so does the linuxx86_64 when tested on CentOS 6.3.

The GTK workaround (System Settings/Application Appearance/GTK/Gtk2 theme Raleigh) works for my application, after a reboot of the VM. But I do not like this as the look and feel is somewhat old fashioned and not in line with the rest of the system.

However, and this is my contribution to this discussion, when I run the application from the command line as root, with sudo <myapp>, the problem disappears (has anybody tried?). I then thought that it may have been caused by some missing file access permissions and temporarily gave full access to the whole workspace directory but the problem reappeared, so it is not connected to permissions on the workspace. Also for me the hs_errxxx.pid point to libgobject-2.0.s0.0+0x16bfe __float128+0x1e

[Tom B (0-tom-4)]

Unfortunately running Eclipse via sudo does not fix the issue for me and I get the exact same crash:

Stack: [0x00007f252c760000,0x00007f252c861000], sp=0x00007f252c85d040, free space=1012k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libgobject-2.0.so.0+0x19a48] g_object_get_qdata+0x18

[Tom B (0-tom-4)]

This may also be relevant: It doesn't seem to be an Ubuntu/Debian based distro specific bug. I also have an Arch Linux installation and it happens on that as well.

[David Matějček (dmatej)]

Reproduced also with the JDK6, OpenJDK7, and finally JDK8:
Java(TM) SE Runtime Environment (build 1.8.0-b132)
Java HotSpot(TM) 64-Bit Server VM (build 25.0-b70, mixed mode)

Problematic frames occurs in two variants:
1) # C [libgobject-2.0.so.0+0x31b3c] g_type_check_instance_is_a+0x3c
2) # C [libgtk-x11-2.0.so.0+0x24492a] gtk_widget_queue_draw+0x1a

The Eclipse crashes sometimes after 5 minutes after start, sometimes after 6 hours, sometimes hangs and crashes after few seconds or even ten minutes. Always with one of these variants.

[David Matějček (dmatej)]

Reproduced also with the JDK6, OpenJDK7, and finally JDK8:
Java(TM) SE Runtime Environment (build 1.8.0-b132)
Java HotSpot(TM) 64-Bit Server VM (build 25.0-b70, mixed mode)

Problematic frames occurs in two variants:
1) # C [libgobject-2.0.so.0+0x31b3c] g_type_check_instance_is_a+0x3c
2) # C [libgtk-x11-2.0.so.0+0x24492a] gtk_widget_queue_draw+0x1a

The Eclipse crashes sometimes after 5 minutes after start, sometimes after 6 hours, sometimes hangs and crashes after few seconds or even ten minutes. Always with one of these variants.

[David Matějček (dmatej)]

I have an idea, theoretical scenario: isn't it somehow similar problem as with the volatile and synchronization in Java?
GTK makes many updates and redrawing at the same time.
Even Eclipse have parallel threads for redrawing.
I got 2core/4thread CPU Intel i5.

What if the GTK tries to redraw something, but the CPU context switching occurs and then it has invalid references?
All code then seems correct, but it is not.

I see these crashes for maybe two years (perhaps since Kubuntu 12.10), some were closed as incomplete, one even as fixed (bug in intel driver), but this problem is still the same.

Maybe I will try some experiments with the affinity and I will see if it will make some difference:
http://karuppuswamy.com/wordpress/2010/09/23/how-to-launch-your-application-in-a-specific-cpu-cpu-affinity-in-linux/

[Tom B (0-tom-4)]

It's probably worth mentioning that with the latest Eclipse (Kepler) this bug does not occur. Perhaps looking at the differences between what happens on "close" in the find dialog in Kepler and Juno will shed some light on what is causing the crash.

[Claude Durocher (claude-d)]

The problem is fixed with the latest trusty build (at least the menus behavior) in Kepler 4.3.2 : it was present in previous versions of Kepler.

[David Matějček (dmatej)]

Workaround:
1) Start the Eclipse
2) find pid of the process
3) taskset -p 0x00000001 <pid>

The problem is something like with volatile and synchronized; with this I used two Eclipse instances for 12 hours without a crash!
After restart I forgot to set it again and Eclipse crashed after 5 minutes ...

Now again with taskset after 2 hours no problem.

[David Matějček (dmatej)]

I use Eclipse Kepler 4.3.2 and this bug occurs more often then ever before!
Eclipse also is not the only application crashing with this bug - today I have seen bugreport also for Gimp.

Workaround:
1) Start the Eclipse
2) find pid of the process
3) taskset -p 0x00000001 <pid>

The problem is something like with volatile and synchronized; with this I used two Eclipse instances for 12 hours without a crash!
After restart I forgot to set it again and Eclipse crashed after 5 minutes ...

Now again with taskset after 2 hours no problem.

[auspex (auspex)]

Nice idea, David but no luck for me.

I start Eclipse with:
    GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc
./adt-bundle-linux-x86_64-20131030/eclipse/eclipse -data .

and do Ctrl-F, and click on Find, Find, Close. Eclipse aborts at
g_object_get_qdata+0x18

I restart Eclipse, find the PID, and do:
    $ taskset -p 0x00000001 32060
    pid 32060's current affinity mask: f
    pid 32060's new affinity mask: 1
so, it appears to have worked. I execute the same sequence and Eclipse
aborts at g_object_get_qdata+0x18. Also note:
    # An error report file with more information is saved as:
    # /home/derek/.../hs_err_pid32060.log
so I had the right PID.

On Thu, Apr 10, 2014 at 10:28 PM, David Matějček <<email address hidden>
> wrote:

> I use Eclipse Kepler 4.3.2 and this bug occurs more often then ever before!
> Eclipse also is not the only application crashing with this bug - today I
> have seen bugreport also for Gimp.
>
> Workaround:
> 1) Start the Eclipse
> 2) find pid of the process
> 3) taskset -p 0x00000001 <pid>
>
> The problem is something like with volatile and synchronized; with this I
> used two Eclipse instances for 12 hours without a crash!
> After restart I forgot to set it again and Eclipse crashed after 5 minutes
> ...
>
> Now again with taskset after 2 hours no problem.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1241101
>
> Title:
> Java crash in libglib-2.0 after upgrade from 13.04 to 13.10
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/eclipse/+bug/1241101/+subscriptions
>

--
derek

[auspex (auspex)]

I'll add that my test was not using Kepler, but David's suggesting Kepler doesn't change things.

[Tom B (0-tom-4)]

I've also tested this using Juno (the version included in Zend Studio 10) and unfortunately share auspex's experience, it doesn't solve the issue for me.

I also tried explicitly launching the application with an affinity:

taskset 0x00000001 zend-studio

with the same reslult.

[David Matějček (dmatej)]

Yes, today Eclipse crashes again, taskset does not help. :(
This is really very bad, hardly reproducible bug. It seems the faster machine and newer JDK and Linux, the more occurences I have ...
Maybe the fastest method to fix it is to dig into the code of GTK ...?

[David Matějček (dmatej)]

EDIT: workaround does not help. And one more note: Eclipse crashes earlier when showing variables in debug mode.

[David Matějček (dmatej)]

It seems it is fixed in newer GTK, but after my recent experiences I am not so optimistic. But after upgrade to 1.3.5 and also GTK 3.10 and restarting the system Eclipse still did not crash (10 hours yesterday, 5 hours today, switching git branches on massive project, debugging, refactoring).

Can you try it someone too? (package is big, you need fast connection)
sudo add-apt-repository ppa:gnome3-team/gnome3
sudo apt-get install libgtk-3-0
Download 1.3.5: https://launchpad.net/ubuntu/+source/oxygen-gtk3
Unpack and see instructions in the file INSTALL.

I hope Eclipse will be stable on Kubuntu 14.04 without these experiments ...

[auspex (auspex)]

Fingers crossed, David! Simply installing libgtk-3-0 worked for the
trivial case I described above, where I could always make it fail (I
haven't upgraded oxygen, yet, and probably won't if this works without
it—I'm still using the override
GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc).

On Wed, Apr 16, 2014 at 8:49 AM, David Matějček
<email address hidden>wrote:

> It seems it is fixed in newer GTK, but after my recent experiences I am
> not so optimistic. But after upgrade to 1.3.5 and also GTK 3.10 and
> restarting the system Eclipse still did not crash (10 hours yesterday, 5
> hours today, switching git branches on massive project, debugging,
> refactoring).
>
> Can you try it someone too? (package is big, you need fast connection)
> sudo add-apt-repository ppa:gnome3-team/gnome3
> sudo apt-get install libgtk-3-0
> Download 1.3.5: https://launchpad.net/ubuntu/+source/oxygen-gtk3
> Unpack and see instructions in the file INSTALL.
>
> I hope Eclipse will be stable on Kubuntu 14.04 without these experiments
> ...
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1241101
>
> Title:
> Java crash in libglib-2.0 after upgrade from 13.04 to 13.10
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/eclipse/+bug/1241101/+subscriptions
>

--
derek

[auspex (auspex)]

I did have to upgrade oxygen as specified by David, too, as even with the Raleigh theme I was getting failures (much less frequent or reproducible than before upgrading libgtk, though). After upgrading oxygen, I didn't experience any further failures.

Of course, now I'm in the process of upgrading to ubuntu 14.04, so the whole process is likely to begin again...

[Hkaiserl (hkaiserl)]

same issue here with ubuntu 13.10 x64

[José Tomás Atria (jtatria)]

This also occurs here, with Ubuntu 14.04 x64.

[José Tomás Atria (jtatria)]

I've been using the Raleigh override, and I have experienced no crashes.

I also tried David's suggestions, though the libgtk-3-0 version in the Trusty repos seems to be the same as the one in the gnome3 ppa (Package libgtk3 is already the newest version etc.). I compiled oxygen-gtk locally, but the problem persists.

All of the above on 14.04.

[kecsap (csaba-kertesz)]

Guys, I am on 13.10 and nothing fixed this issue except downloading Eclipse Kepler from eclipse.org instead of the non-working Eclipse in the Ubuntu repos

[aanno (thomas-pasch)]

Dear kecsap,

well, the only problem is that this bug was never limited to eclipse from ubuntu repos. It has also stroked me with an eclipse from eclipse.org...

[Fabien-toral (fabien-toral)]

Created attachment 244042
crash log

(In reply to David Matějček from comment #22)
> Problematic frames occurs in two variants:
> 1) # C [libgobject-2.0.so.0+0x31b3c] g_type_check_instance_is_a+0x3c
> 2) # C [libgtk-x11-2.0.so.0+0x24492a] gtk_widget_queue_draw+0x1a
>

I've a third variant in the crash log provided in attachment :
# C [libgdk-x11-2.0.so.0+0x5173f] gdk_display_open+0x3f

The crash occurs directly at startup while the splash screen is displayed.

My system is a fresh Debian Wheezy 7.5, the gtk version found is :
libgtk-3-common 3.4.2-7

I'll try to find some time to check with a newer GTK as suggested by your comment #25

[Federico Tello Gentile (federicotg)]

NetBeans 8 also crashes the first time you launch it and works ok the second time. Problem is in

C [libgobject-2.0.so.0+0x125be0] _fini+0xecb18

[xor (xor)]

Can confirm the issue with Kubuntu 14.04 amd64, Eclipse crashes when closing the search window:

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f487db99718, pid=13246, tid=139949693536000
#
# JRE version: OpenJDK Runtime Environment (7.0_55-b14) (build 1.7.0_55-b14)
# Java VM: OpenJDK 64-Bit Server VM (24.51-b03 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x19718] g_object_get_qdata+0x18
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# If you would like to submit a bug report, please include
# instructions on how to reproduce the bug and visit:
# http://icedtea.classpath.org/bugzilla
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

[xor (xor)]

Tried the following workarounds on Kubuntu 14.04 amd64 with Eclipse from package management, all did not help
- Opening the search menu (CTRL+F) and closing it with ESC only takes 2-3 attempts to crash Eclipse:

- Setting affinity (taskset -c 1 eclipse)
- Setting theme to Raleigh in systemsettings
- GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc eclipse
- UBUNTU_MENUPROXY= eclipse
- Using all of the above at once ("export GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc ; export UBUNTU_MENUPROXY= ; taskset -c 1 eclipse" - and setting Raleigh in systemsettings)

[Fnfhspwanv (fnfhspwanv)]

Can confirm the issue with:
- Kubuntu 14.04 amd64
- Eclipse 3.8.1
- java -version = java version "1.7.0_55" OpenJDK Runtime Environment (IcedTea 2.4.7, (7u55-2.4.7-1ubuntu1), OpenJDK 64-Bit Server VM (build 24.51-b03, mixed mode)

Tried the following workarounds, all did not help
- Opening the search menu (CTRL+F) and closing it with ESC only takes 2-3 attempts to crash Eclipse:

- Setting affinity (taskset -c 1 eclipse)
- Setting theme to Raleigh in systemsettings
- GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc eclipse
- UBUNTU_MENUPROXY= eclipse
- Using all of the above at once ("export GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc ; export UBUNTU_MENUPROXY= ; taskset -c 1 eclipse" - and setting Raleigh in systemsettings)

Typical hs_err file:

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f487db99718, pid=13246, tid=139949693536000
#
# JRE version: OpenJDK Runtime Environment (7.0_55-b14) (build 1.7.0_55-b14)
# Java VM: OpenJDK 64-Bit Server VM (24.51-b03 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x19718] g_object_get_qdata+0x18
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# If you would like to submit a bug report, please include
# instructions on how to reproduce the bug and visit:
# http://icedtea.classpath.org/bugzilla
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

[Fabien-toral (fabien-toral)]

So, after few tries in Virtual Machines to test different Debian/gtk versions, and other researches on the net, I found a workaround to make Luna work on my Debian laptop :

export SWT_GTK3=0

That aims to fallback to the GTK2 SWT implementation and bring my Eclipse back!

I was not on the right bug report, and found my way with a comment on bug #430736 https://bugs.eclipse.org/bugs/show_bug.cgi?id=430736#c26

With that, it comes to me that the only solution, as i don't want to upgrade my glibc, is to fallback to GTK2...

And thanks to http://www.eclipse.org/swt/R4_4/new_and_noteworthy.html#m3, the information was there...

[mathew (meta23)]

Same problem here with IBM Sametime using the embedded IBM JRE and Eclipse Expeditor. So if it's an Eclipse bug, it's one you'll find in commercial closed-source products.

Workaround of setting GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc works.

[Robert Muil (robertmuil)]

I'm also seeing instability with SIGSEVs in the native code when running eclipse.
I tried the SWT_GTK3=0 workaround, but got the following crash:

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f5f9d789d6c, pid=7849, tid=140048942749440
#
# JRE version: Java(TM) SE Runtime Environment (7.0_60-b19) (build 1.7.0_60-b19)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (24.60-b09 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x31d6c] g_type_check_instance_is_a+0x3c
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /home/rmuil/hs_err_pid7849.log
Compiled method (nm) 87459 1185 n org.eclipse.swt.internal.gtk.OS::_g_object_get_qdata (native)
 total in heap [0x00007f5fa517f290,0x00007f5fa517f610] = 896
 relocation [0x00007f5fa517f3b0,0x00007f5fa517f410] = 96
 main code [0x00007f5fa517f420,0x00007f5fa517f610] = 496
#
# If you would like to submit a bug report, please visit:
# http://bugreport.sun.com/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

[Berserq-k-mail (berserq-k-mail)]

Created attachment 246108
full log

Hello, with Eclipse 4.2.2 with ADT get this bug.
Core dump https://yadi.sk/d/UTGqkxJmZz6e7

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007fe3b0b716b8, pid=6629, tid=140615974131456
#
# JRE version: Java(TM) SE Runtime Environment (7.0_67-b01) (build 1.7.0_67-b01)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (24.65-b04 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x1a6b8] g_object_get_qdata+0x18
#
# Core dump written. Default location: /home/lberserq/java/adt-bundle-linux-x86_64-20140702/eclipse/core or core.6629
#
# An error report file with more information is saved as:
# /home/lberserq/java/adt-bundle-linux-x86_64-20140702/eclipse/hs_err_pid6629.log
Compiled method (nm) 139197 719 n org.eclipse.swt.internal.gtk.OS::_g_object_get_qdata (native)
 total in heap [0x00007fe3a9103390,0x00007fe3a9103710] = 896
 relocation [0x00007fe3a91034b0,0x00007fe3a9103510] = 96
 main code [0x00007fe3a9103520,0x00007fe3a9103710] = 496

[Arunkumar-thondapu-i (arunkumar-thondapu-i)]

(In reply to Kamil Khamitov from comment #31)
> Created attachment 246108 [details]
> full log
>
> Hello, with Eclipse 4.2.2 with ADT get this bug.
> Core dump https://yadi.sk/d/UTGqkxJmZz6e7
>
> #
> # A fatal error has been detected by the Java Runtime Environment:
> #
> # SIGSEGV (0xb) at pc=0x00007fe3b0b716b8, pid=6629, tid=140615974131456
> #
> # JRE version: Java(TM) SE Runtime Environment (7.0_67-b01) (build
> 1.7.0_67-b01)
> # Java VM: Java HotSpot(TM) 64-Bit Server VM (24.65-b04 mixed mode
> linux-amd64 compressed oops)
> # Problematic frame:
> # C [libgobject-2.0.so.0+0x1a6b8] g_object_get_qdata+0x18

This crash is bug 372560 which is fixed since Eclipse 4.3.

[jstammi (jstammi)]

Concerning switching the oxygen theme please consider https://bugs.launchpad.net/ubuntu/+source/java-common/+bug/1205452/comments/20.

IMHO there are multiple reasons at the moment causing java applications to SIGSEGV the reported way. This link shows AFAIS one of them.

[Simone (tigerjack89)]

Don't know if it is strongly related, but I think so.
I'm using Modelio 3.1 on Ubuntu 64 bit. Sometimes, apparently without a specific sequence pattern, Modelio closes; other times, it doesn't allow to make any type of operations (i.e. when I right click on an element, it doesn't show anything). I don't know what the problem is. If I launch it from the terminal, it always returns error message like this

>Modelio 3:10699): GLib-GObject-CRITICAL **: g_closure_unref: assertion 'closure->ref_count > 0' failed
or this

>(Modelio 3:10699): GLib-CRITICAL **: Source ID 20104 was not found when attempting to remove it

The error message returned in the log file is
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f803bba4d6c, pid=27247, tid=140190697162496
#
# JRE version: Java(TM) SE Runtime Environment (7.0_45-b18) (build 1.7.0_45-b18)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (24.45-b08 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x31d6c] g_type_check_instance_is_a+0x3c
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# If you would like to submit a bug report, please visit:
# http://bugreport.sun.com/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

If it could be of help, I can attach the various log files it generates.

[tdeering (tomdeering7)]

Created attachment 247831
Additional crash log

Eclipse 4.3.2 crashes *constantly* (several times per day) with this problem on Ubuntu 14.04. The UI goes grey and unresponsive, and after force-quitting the application I get an error log with a stack trace about a seg fault in native code invoked from org.eclipse.swt.internal.gtk.OS._gtk_widget_show().

I've heard rumors that GTK on Ubuntu is not entirely sound. Eclipse may actually be a victim here rather than the culprit.

[tdeering (tomdeering7)]

This bug in GTK causes Eclipse to crash *constantly* (several times per day) on Ubuntu 14.04. I've attached an error log that gets left behind after the crash. The incriminating frame is:

C [libglib-2.0.so.0+0x38fe0] g_str_hash+0x0

I really hope this gets fixed! Stability bugs are the worst.

[Lukasz Skalski (l-skalski)]

Hi Simeone,

Your problem with Modelio and removing non-existent sources:

>(Modelio 3:10699): GLib-CRITICAL **: Source ID 20104 was not found when attempting to remove it

is connected with some changes in GLib library. GLib 2.40.0 introduced the following change:

"[..] g_source_remove() will now throw a critical in the case that you try to remove a non-existent source. We expect that there is some code in the wild that will fall afoul of this new critical but considering that we now reuse source IDs, this code is already broken and should probably be fixed."

[JPT (j-p-t)]

Hi,

I simply don't understand why nobody fixes this bug.
This was already discussed a thousand times.
See https://bugs.eclipse.org/bugs/show_bug.cgi?id=430736

The problem occurs in any eclipse versions I tried (beginning with eclipse-3.8 from Ubuntu repo).

workarounds are either:

export SWT_GTK3=0

for me also worked
export SWT_GTK3=1

the other bug says put into eclipse.ini:
--launcher.GTK_version
2

If you don't fix it, why don't you deliver any of the workarounds?

[tdeering (tomdeering7)]

I'm still getting segmentation faults out of libglib-2.0 with Java 1.8.0_25-b17, Ubuntu 14.10, and Eclipse. However, after upgrading to Java 8, the seg faults I'm seeing have moved to from g_str_hash() to g_type_check_instance_is_a()

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f5bcdd41d6c, pid=12791, tid=140033135843072
#
# JRE version: Java(TM) SE Runtime Environment (8.0_25-b17) (build 1.8.0_25-b17)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.25-b02 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libgobject-2.0.so.0+0x31d6c] g_type_check_instance_is_a+0x3c

...

Stack: [0x00007f5c01419000,0x00007f5c0151a000], sp=0x00007f5c015167b8, free space=1013k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libgobject-2.0.so.0+0x31d6c] g_type_check_instance_is_a+0x3c
J 20947 C2 org.eclipse.swt.widgets.MenuItem.setText(Ljava/lang/String;)V (259 bytes) @ 0x00007f5bea2fa368 [0x00007f5bea2f9780+0xbe8]

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
J 8151 org.eclipse.swt.internal.gtk.OS._gtk_bin_get_child(J)J (0 bytes) @ 0x00007f5bea1fa241 [0x00007f5bea1fa200+0x41]
J 20947 C2 org.eclipse.swt.widgets.MenuItem.setText(Ljava/lang/String;)V (259 bytes) @ 0x00007f5bea2fa368 [0x00007f5bea2f9780+0xbe8]
...

[Mrull (mrull)]

same problem for an old Eclipse 3.6.2. the workaround worked for me: http://askubuntu.com/a/530468

[Lfarkas (lfarkas)]

it's still happened on rhel/centos-7.0 with:
- eclipse-swt-4.4.0
- glib2-2.36.3
is there any solution to this?

[Aaron Curtis (substantialnoninfringinguser)]

I can confirm this in Comsol Multiphysics, and various versions of eclipse and liclipse, on Kubuntu 14.04 64-bit. Really serious bug. The Eclipses tend to crash during debugging but it seems random. Comsol used to run for a while before crashing but now crashes as soon as you create a new blank model.

[Aaron Curtis (substantialnoninfringinguser)]

Ok, changing the combobox appearance as detailed at http://askubuntu.com/questions/513471/kubuntu-14-04eclipse-adt-crashes-at-button-ok-from-project-properties/530468 fixed it for me.

[Marc-andre-laperle (marc-andre-laperle)]

I can reproduce the crash using Thomas Singer's example code (comment 9), on Ubuntu 13.10 and 14.04 when UBUNTU_MENUPROXY is enabled. If I disable UBUNTU_MENUPROXY it works. I think it's the same issue as bug 461311: radio button in the menu. Can we mark one as duplicate of the other?

[Arunkumar-thondapu-i (arunkumar-thondapu-i)]

*** Bug 466326 has been marked as a duplicate of this bug. ***

[Simone (tigerjack89)]

For me, no one of the suggested workarounds work. I tried with export SWT_GTK3=0 or export SWT_GTK3=1, but every time I move the mouse pointer on Project>Build Working Set Eclipse crashes with the following
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007fa334a1efe0, pid=30089, tid=140339583497984
#
# JRE version: Java(TM) SE Runtime Environment (8.0_45-b14) (build 1.8.0_45-b14)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.45-b02 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C [libglib-2.0.so.0+0x38fe0] g_str_hash+0x0
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again

(Note that it always names g_str_hash+0x0).
If you like, I can attach the full log file.

[Marc-andre-laperle (marc-andre-laperle)]

(In reply to Simone Perriello from comment #39)
> For me, no one of the suggested workarounds work. I tried with export
> SWT_GTK3=0 or export SWT_GTK3=1, but every time I move the mouse pointer on
> Project>Build Working Set Eclipse crashes with the following

Have you tried UBUNTU_MENUPROXY=0 ?

[Simone (tigerjack89)]

(In reply to Marc-Andre Laperle from comment #40)
> (In reply to Simone Perriello from comment #39)
> > For me, no one of the suggested workarounds work. I tried with export
> > SWT_GTK3=0 or export SWT_GTK3=1, but every time I move the mouse pointer on
> > Project>Build Working Set Eclipse crashes with the following
>
> Have you tried UBUNTU_MENUPROXY=0 ?

Tried, and it always crash every single time.

[Cristiano Gavião (cvgaviao)]

(In reply to Simone Perriello from comment #41)
> Tried, and it always crash every single time.

Have you tried this? to create a desktop configuration file and set its command as:

env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0 /thepathToEclipse/eclipse

I've being using this for months with many ubuntu machines and is working great.

[Simone (tigerjack89)]

(In reply to Cristiano Gaviao from comment #42)
> (In reply to Simone Perriello from comment #41)
> > Tried, and it always crash every single time.
>
> Have you tried this? to create a desktop configuration file and set its
> command as:
>
> env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0 /thepathToEclipse/eclipse
>
> I've being using this for months with many ubuntu machines and is working
> great.

Do you mean creating a .desktop file? If so, where I have to put the commands in the exec, right? Can you provide a template of a .desktop file? I've always launched Eclipse from terminal.

[Simone (tigerjack89)]

(In reply to Cristiano Gaviao from comment #42)
> (In reply to Simone Perriello from comment #41)
> > Tried, and it always crash every single time.
>
> Have you tried this? to create a desktop configuration file and set its
> command as:
>
> env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0 /thepathToEclipse/eclipse
>
> I've being using this for months with many ubuntu machines and is working
> great.

It doesn't work anyway.
I've used this .desktop file

[Desktop Entry]
Version=4.4.2
Name=Eclipse-jee
GenericName=IDE
Comment=Programming IDE
MimeType=
Exec=env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0 /opt/eclipse_all/eclipse-jee/eclipse
TryExec=eclipse-jee
Icon=/opt/eclipse_all/eclipse-jee/icon.xpm
Type=Application
Terminal=false
Categories=IDE;Development;

Eclipsee starts well, but crashes every time I try to move the pointer on Project->Build Working Set. It also crashes other times, but I haven't identified a pattern yet.

[Cristiano Gavião (cvgaviao)]

(In reply to Simone Perriello from comment #43)
> Do you mean creating a .desktop file? If so, where I have to put the
> commands in the exec, right? Can you provide a template of a .desktop file?
> I've always launched Eclipse from terminal.

Yep, its a .desktop file that I've put in /usr/share/applications.

[Desktop Entry]
Version=4.5
Comment=Integrated Development Environment
Exec=env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0 /opt/eclipse-ide/java-mars/eclipse/eclipse
Type=Application
Icon=eclipse-luna.xpm
Terminal=false
NoDisplay=false
Categories=Development;IDE;Java
StartupWMClass=Eclipse
Name[en_US]=Eclipse Mars Java

[Simone (tigerjack89)]

(In reply to Cristiano Gaviao from comment #45)
> (In reply to Simone Perriello from comment #43)
> > Do you mean creating a .desktop file? If so, where I have to put the
> > commands in the exec, right? Can you provide a template of a .desktop file?
> > I've always launched Eclipse from terminal.
>
> Yep, its a .desktop file that I've put in /usr/share/applications.
>
> [Desktop Entry]
> Version=4.5
> Comment=Integrated Development Environment
> Exec=env LIBOVERLAY_SCROLLBAR=0 SWT_GTK3=0
> /opt/eclipse-ide/java-mars/eclipse/eclipse
> Type=Application
> Icon=eclipse-luna.xpm
> Terminal=false
> NoDisplay=false
> Categories=Development;IDE;Java
> StartupWMClass=Eclipse
> Name[en_US]=Eclipse Mars Java

More or less, it's the same .desktop file that I've used (see my comment above). However, as said, it doesn't work at all.

[Simone (tigerjack89)]

Don't know if it can be useful, but often after the crash the bash shell from which I've launched Eclipse doesn't respond to keyboard commands.

[Sxenos (sxenos)]

Created attachment 253846
Short SWT snippet that reproduces the problem

[Eclipse-p (eclipse-p)]

Created attachment 254676
crash log

[Sxenos (sxenos)]

Note that there was some further discussion of this bug here:

https://bugs.eclipse.org/bugs/show_bug.cgi?id=461311

In addition to causing crashes when opening the Window menu, it also can cause "reset perspective" to crash.

This only occurs under Ubuntu, and seems related to Ubuntu's menuproxy feature (the thing where it moves the menus to the top of the screen).

In case anyone missed it, I've attached an SWT-only code snippet that demonstrates the crash in about a page of code.

[Sxenos (sxenos)]

You can also reproduce this in Eclipse by clicking Run... > Breakpoint Types

[Eclipse-p (eclipse-p)]

A user reported this problem for the latest SmartGit, too, though we already set a couple of environment variables in the launcher script to prevent it. He runs Linux Mint 17.2 Cinnamon and had the "Global Application Menu" applet installed. After disabling that, SmartGit did not crash any more.

[Thkwak (thkwak)]

Created attachment 256390
Patch

Hi. I have looked into this problem and came up with the following workaround patch that still uses Unity and avoids the crash.

I found that the crash occurs because a new Radio menu is created with *empty label*
for which unity_gtk_action_group_get_state_name() returns null and this ends up with a null-pointer dereference at g_str_hash().

My suggestion is to modify _1gtk_1radio_1menu_1item_1new_1with_1label() in os.c to call gtk_radio_menu_item_new_with_label() with an arbitrary, non-empty string if the
given string for label is for empty string (2nd parameter). In this patch, I used the string with one space (" ") as an arbitrary text.

I confirmed that, with this patch, the test given by Thomas Singer passes and all tests in Test_org_eclipse_swt_widgets_MenuItem.java pass as well.
Please let me know if you have any question or suggestion for this patch. Thanks.

--- "a/bundles/org.eclipse.swt/Eclipse SWT PI/gtk/library/os.c" 2013-05-14 11:50:23.000000000 +0900
+++ "b/bundles/org.eclipse.swt/Eclipse SWT PI/gtk/library/os.c" 2015-09-04 19:49:40.768009600 +0900
@@ -13868,7 +13868,10 @@ JNIEXPORT jintLong JNICALL OS_NATIVE(_1gtk_1radio_1menu_1item_1new_1with_1label)
    jintLong rc = 0;
    OS_NATIVE_ENTER(env, that, _1gtk_1radio_1menu_1item_1new_1with_1label_FUNC);
    if (arg1) if ((lparg1 = (*env)->GetByteArrayElements(env, arg1, NULL)) == NULL) goto fail;
- rc = (jintLong)gtk_radio_menu_item_new_with_label((GSList *)arg0, (const gchar *)lparg1);
+ if (lparg1[0] == '\0')
+ rc = (jintLong)gtk_radio_menu_item_new_with_label((GSList *)arg0, (const gchar *)" ");
+ else
+ rc = (jintLong)gtk_radio_menu_item_new_with_label((GSList *)arg0, (const gchar *)lparg1);
 fail:
    if (arg1 && lparg1) (*env)->ReleaseByteArrayElements(env, arg1, lparg1, 0);
    OS_NATIVE_EXIT(env, that, _1gtk_1radio_1menu_1item_1new_1with_1label_FUNC);

[Marc-andre-laperle (marc-andre-laperle)]

(In reply to Taehoon Kwak from comment #53)
> I found that the crash occurs because a new Radio menu is created with
> *empty label*
> for which unity_gtk_action_group_get_state_name() returns null and this ends
> up with a null-pointer dereference at g_str_hash().

Would it be a good idea to fix unity_gtk_action_group_get_state_name to that it returns an empty string instead of null? I'm thinking it would be good to fix the root cause for other programs and earlier versions of Eclipse.

[Thkwak (thkwak)]

(In reply to Marc-Andre Laperle from comment #54)
> (In reply to Taehoon Kwak from comment #53)
> > I found that the crash occurs because a new Radio menu is created with
> > *empty label*
> > for which unity_gtk_action_group_get_state_name() returns null and this ends
> > up with a null-pointer dereference at g_str_hash().
>
> Would it be a good idea to fix unity_gtk_action_group_get_state_name to that
> it returns an empty string instead of null? I'm thinking it would be good to
> fix the root cause for other programs and earlier versions of Eclipse.

My patch is a workaround that changes uses of Unity. Patching Unity seems a nice idea, yet I am not sure of any side effect

[Martin Oberhuber (martin-oberhuber)]

Many thanks for the analysis so far, the explanation with "empty radio label" makes a lot of sense! - Does anyone know if a ticket already exists with Ubuntu for getting this fixed ?

We also do see the crash occasionally with Mars on Ubuntu 14.04 64-bit.
The suggested workarounds SWT_GTK3=0 UBUNTU_MENUPROXY=0 have been applied.
A fix in either Mars.2 or an Ubuntu 14.04 update would be highly appreciated !

CQ:WIND00-WB4-6132

[Marc-andre-laperle (marc-andre-laperle)]

*** Bug 449031 has been marked as a duplicate of this bug. ***

[Marc-andre-laperle (marc-andre-laperle)]

I'll assign to myself for now to investigate a fix in Ubuntu. Otherwise we can try to apply the patch in SWT.

[Marc-andre-laperle (marc-andre-laperle)]

The issue is fixed in Ubuntu >= 14.10. I proposed a patch to back-port the fix to 14.04, see https://bugs.launchpad.net/ubuntu/+source/unity-gtk-module/+bug/1427866/comments/5

[Kalyan-prasad (kalyan-prasad)]

*** Bug 457512 has been marked as a duplicate of this bug. ***

[Markus-keller (markus-keller)]

*** Bug 462002 has been marked as a duplicate of this bug. ***

[Markus-keller (markus-keller)]

There's no point in delaying a workaround in SWT for this crash. Attachment 253846 still reproduces on Ubuntu 14.04.

AFAICS, this is the place where the fix in Ubuntu could eventually show up:
http://bazaar.launchpad.net/~indicator-applet-developers/unity-gtk-module/trunk.14.04/view/head:/lib/unity-gtk-action-group.c#L848

Released a simple workaround to master: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=534ed490b51e7b64f0afdbd1595dc7098177d066

[Markus-keller (markus-keller)]

*** Bug 465485 has been marked as a duplicate of this bug. ***

[Marc-andre-laperle (marc-andre-laperle)]

Thanks Markus!