openjdk team mailing list archive

[Bug Watch Updater <1266004@xxxxxxxxxxxxxxxxxx>]

Launchpad has imported 91 comments from the remote bug at
https://bugs.eclipse.org/bugs/show_bug.cgi?id=400626.

If you reply to an imported comment from within Launchpad, your comment
will be sent to the remote bug automatically. Read more about
Launchpad's inter-bugtracker facilities at
https://help.launchpad.net/InterBugTracking.

------------------------------------------------------------------------
On 2013-02-12T21:07:26+00:00 Frits Jalvingh wrote:

Created attachment 226958
JVM dump

At random, mostly very inopportune times Eclipse 4.3m5 crashes (creating
a JVM dump). The dump will be attached after this. I have no
reproduction scenario.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/0

------------------------------------------------------------------------
On 2013-02-13T18:05:49+00:00 Grant Gayed wrote:

Which Linux distro are you using?

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/1

------------------------------------------------------------------------
On 2013-02-13T18:26:00+00:00 Frits Jalvingh wrote:

I'm using Ubuntu 12.12 64-bit. I also posted a hang that might be
related (bug 400723)- I get the feeling it might be related to the
browser integration which if I remember correctly is used inside hover
popups.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/2

------------------------------------------------------------------------
On 2013-02-19T07:47:37+00:00 Frits Jalvingh wrote:

This bug has made it impossible to use this milestone, sadly enough- it
crashes tens of times a day. Is there a way to disable webkit? Or to
replace it with the old xulrunner?

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/3

------------------------------------------------------------------------
On 2013-02-19T15:41:40+00:00 Grant Gayed wrote:

You can change the Browser's default renderer type by adding the
following vm arg to the end of your eclipse.ini:

-Dorg.eclipse.swt.browser.DefaultType=mozilla

You can then point at a XULRunner to use by adding vm arg:

-Dorg.eclipse.swt.browser.XULRunnerPath=<fullPath>

(note the list of supported XULRunner versions:
http://www.eclipse.org/swt/faq.php#browserlinux )

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/4

------------------------------------------------------------------------
On 2013-03-27T10:44:25+00:00 Frits Jalvingh wrote:

FYI: this same bug now also occurs several times per hour on Eclipse
3.8.2, when it runs on Ubuntu 13.04. Yes, I know it is not yet released,
just adding info.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/5

------------------------------------------------------------------------
On 2013-04-29T20:40:23+00:00 Grant Gayed wrote:

*** Bug 401603 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/6

------------------------------------------------------------------------
On 2013-05-13T19:08:02+00:00 Grant Gayed wrote:

*** Bug 396577 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/7

------------------------------------------------------------------------
On 2013-05-13T19:13:25+00:00 Grant Gayed wrote:

To update, I haven't been able to reproduce this yet (using 3.8.2 on
several different Ubuntu versions, closing the initial Welcome screen by
clicking its 'X').  Am continuing to look into this, but will be
difficult to track down if it's not reproducible.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/8

------------------------------------------------------------------------
On 2013-05-14T14:42:20+00:00 Daniel-megert wrote:

(In reply to comment #8)
> To update, I haven't been able to reproduce this yet (using 3.8.2 on several
> different Ubuntu versions, closing the initial Welcome screen by clicking
> its 'X').  Am continuing to look into this, but will be difficult to track
> down if it's not reproducible.

I've seen it on 3.x only with Ubuntu 12.04, but that was via VNC.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/9

------------------------------------------------------------------------
On 2013-05-15T09:24:33+00:00 Frits Jalvingh wrote:

It does not abort that often on 3.8.2. But 4.3, currently m7, does it an awful lot. The problem remains that there is no certain way to reproduce: sometimes I need to start 4.3m7 several times quickly in succession because as soon as the main window appears and I move the mouse over source code (which would probably show the hover containing source information) Eclipse aborts.
But there are also times that I can work for 3 hours without any problem.

I don't know if it's possible, but the huge problem now is that I can
hardly use 4.3m7 at all. Would it be possible to somehow "guard" against
this exception somewhere in the source code, so that instead of
completely dying the process continues?

In the meantime, if there is anything I can help with to get more
details on this let me know...

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/10

------------------------------------------------------------------------
On 2013-05-15T20:39:18+00:00 Grant Gayed wrote:

(In reply to comment #10)

You should be able to avoid this as described in comment 4 (eg.- point
-Dorg.eclipse.swt.browser.XULRunnerPath at a downloaded XULRunner 10).
If you're fine with running with no browser support at all (Eclipse
works generally fine without it by providing content in alternate ways
if Browser creation fails), you can set this property to /dev/null .

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/11

------------------------------------------------------------------------
On 2013-05-21T07:19:00+00:00 Linus-probert-d wrote:

This bug just appeared for me on ArchLinux. The changes I made before it
appeared was installing FreeNX (and xfce but I don't think that's as
important).

FreeNx has a dependency on nx-common wich installs libjpeg8.0.2
libpng16.2.0

If I'm not mistaken these have been involved in previous crashes
revolving around 'notifications when hovering a variable/type/etc'.

Posting my hs_err dump aswell

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/12

------------------------------------------------------------------------
On 2013-05-21T07:20:11+00:00 Linus-probert-d wrote:

Created attachment 231232
hs_err dump from archlinux with freenx installed (lipng16.2.0 libjpeg8.0.2)

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/13

------------------------------------------------------------------------
On 2013-05-22T10:02:16+00:00 Ablochberger wrote:

Here's another possible solution: http://aleksz-
programming.blogspot.de/2013/03/eclipse-and-webkit-on-ubuntu-64.html

Replace libwebkitgtk-1 with libwebkitgtk-3

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/14

------------------------------------------------------------------------
On 2013-05-22T10:28:50+00:00 Frits Jalvingh wrote:

(In reply to comment #14)
> Replace libwebkitgtk-1 with libwebkitgtk-3
Thanks a million for posting this!
I removed the libwebkitgtk-1.0* files to test. Eclipse starts normally and hovers work, but they do not seem to use a browser window anymore: they look like text input boxes with only rudimentary formatting of the html javadoc.

Looking at what libraries are open (lsof -p) I see that webkitgtk-3* is
not used at all - so clearly eclipse is not automatically picking up
libwebkitgtk-3 but just not using libwebkitgtk at all. Which is so far
fine by me - not dying has priority ;-)

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/15

------------------------------------------------------------------------
On 2013-05-22T14:03:23+00:00 Grant Gayed wrote:

(In reply to comment #14)

Deleting this library just disables the browser altogether.  A better
way of doing the same thing is described in comment 4
(-Dorg.eclipse.swt.browser.XULRunnerPath can be set to a value like
"/dev/null" if you don't have a supported XULRunner release to point
at).

Regarding GTK3, you can run your Eclipse download (as of Eclipse 4.3)
with GTK3 libraries instead of GTK2, which may or may not help with this
case.  To try this, set Linux environment variable SWT_GTK3=1 before
launching eclipse.  Note that the GTK3 port of SWT is currently
considered to be early access.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/16

------------------------------------------------------------------------
On 2013-05-23T12:54:58+00:00 Frits Jalvingh wrote:

(In reply to comment #16)
> (In reply to comment #14)
> Regarding GTK3, you can run your Eclipse download (as of Eclipse 4.3) with
> GTK3 libraries instead of GTK2
Heh, that dies immediately with:

# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007ff2be8a8d98, pid=32672, tid=140683416110848
#
# JRE version: 7.0_21-b02
# Java VM: OpenJDK 64-Bit Server VM (23.7-b01 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# C  [libwebkitgtk-3.0.so.0+0x3fcd98]  DumpRenderTreeSupportGtk::setComposition(_WebKitWebView*, char const*, int, int)+0x8d28

I am surprised I seem to be one of the few people having this.. Since
it's close to 4.3 release I sure hope I'm the exception..

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/17

------------------------------------------------------------------------
On 2013-05-24T14:25:04+00:00 Grant Gayed wrote:

*** Bug 408843 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/18

------------------------------------------------------------------------
On 2013-06-06T21:36:47+00:00 Marc-andre-laperle wrote:

Hi Grant. We've been getting this crash here (3 of us on Ubuntu 13.04).
We all enabled core dumps and installed libwebkitgtk-1.0-0-dbg so
hopefully we can help provide more information.

The first core dump we got doesn't contain
webkitWebViewRegisterForIconNotification so it might be a different bug,
I'm not sure.

<signal handler called>() at 0x7fb7c2ba60b0	
WebKit::core() at /build/buildd/webkit-1.10.2/build-2.0/../Source/WebKit/gtk/webkit/webkitwebview.cpp:5,368 0x7fb7268cf6d5	
WebKit::GtkAdjustmentWatcher::updateAdjustmentsFromScrollbars() at /build/buildd/webkit-1.10.2/build-2.0/../Source/WebKit/gtk/WebCoreSupport/GtkAdjustmentWatcher.cpp:65 0x7fb7268a8108	
WebKit::updateAdjustmentCallback() at /build/buildd/webkit-1.10.2/build-2.0/../Source/WebKit/gtk/WebCoreSupport/GtkAdjustmentWatcher.cpp:76 0x7fb7268a8189	

the line is
WebKitWebViewPrivate* priv = webView->priv;

so I guess webView is wrong. Let me know if you want the core file (1.9
GB!!).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/19

------------------------------------------------------------------------
On 2013-06-07T15:22:43+00:00 Grant Gayed wrote:

*** Bug 400723 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/20

------------------------------------------------------------------------
On 2013-06-07T17:37:16+00:00 Grant Gayed wrote:

Looking at webkitWebViewRegisterForIconNotification(), either webview is
invalid (retrieved from a WebKitWebFramePrivate) or webview->priv is
invalid.  I've tried various things with the content assist and hover
help browsers to try to make this happen but still haven't seen this
crash (Ubuntu 12.10 with webkitgtk 1.10.0).  I'll try with Ubuntu 13
next week, but I don't anticipate this making a difference since its
webkitgtk libraries are approximately the same.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/21

------------------------------------------------------------------------
On 2013-06-07T17:44:29+00:00 francesco wrote:

Hi,

I began with this type of problems when I upgraded to Ubuntu 13.04.

A trick is to reset the view when the problem arise.

Thank you for your work.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/22

------------------------------------------------------------------------
On 2013-06-10T14:42:37+00:00 HENRY Florian wrote:

Created attachment 232182
Another crask log

Ubuntu 13.04 (kernel 3.9.4). Juno 4.2.2, JDT, Egit (from Juno repo)

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/23

------------------------------------------------------------------------
On 2013-06-10T14:43:49+00:00 HENRY Florian wrote:

Hi,
I experience a crash at least 5 times per days. I have to restrat my computer because on next strat eclispe pop up error about PHP debug view  and it is not usable.
I've attach on of the a pid log file

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/24

------------------------------------------------------------------------
On 2013-06-11T03:36:13+00:00 Grant Gayed wrote:

With Ubuntu 13.04 I had a case crashing almost consistently earlier, at
either webkitWebViewRegisterForIconNotification() or
DumpRenderTreeSupportGtk::setComposition() (like comment 23).  It only
happened with openjdk7, never with an IBM JRE.  However this case no
longer crashes for me with any JRE.

Regardless, it's obviously very timing-dependent, which makes me
suspicious of BrowserInformationControl.setVisible()'s invocation of
readAndDispatch(), which appears in the attached dumps right before the
crash.  This can make an event run before it should, which frequently
causes problems like native null ptr's.  I'm going to see if changing
this class' implementation to no longer spin the event loop seems
helpful.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/25

------------------------------------------------------------------------
On 2013-06-11T08:58:36+00:00 HENRY Florian wrote:

Ok, I will try with the standard JRE, but I've already try and I
experience other crash. I'll do that and give you more information.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/26

------------------------------------------------------------------------
On 2013-06-11T09:34:46+00:00 HENRY Florian wrote:

I've got a crash with standard JRE but not pid file just the /var/crash
report (310Mo) impossible to upload on this site. I'll try to cut the
dump and just past the header part.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/27

------------------------------------------------------------------------
On 2013-06-11T09:36:46+00:00 HENRY Florian wrote:

Created attachment 232214
At least the .log file of .metadata

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/28

------------------------------------------------------------------------
On 2013-06-11T09:40:44+00:00 HENRY Florian wrote:

Created attachment 232216
The apport crash report without core Dump

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/29

------------------------------------------------------------------------
On 2013-06-11T10:02:44+00:00 HENRY Florian wrote:

Another crash. I go back with openJDK at least crash appens but not all
5 minutes

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/30

------------------------------------------------------------------------
On 2013-06-11T10:28:11+00:00 francesco wrote:

Using Eclipse Juno 4.2 and Ubuntu 13.04 there are some problems with
libwebkitgtk,  as reported in this bug and as I reporter with error and
log files.

Using Eclipse Juno 4.2 and Ubuntru 13.04 on amd64 architecture, the
system crash without any information. No Java stack and no information
saved in log file.

Eclipse on Ubuntu seemes to work no so well as in past did.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/31

------------------------------------------------------------------------
On 2013-06-14T02:45:04+00:00 Grant Gayed wrote:

Created attachment 232362
patch to disable icons database (apply post-4.3)

Determining the root cause of this has proven to be quite difficult.
Even compiling WebKitGTK from source with instrumentation did not help
because its characteristics did not match the WebKitGTK that's shipped
with Ubuntu.

Anyways, one thing I was able to determine from the instrumented
WebKitGTK is that the path that crashes is never run if WebKit's icons
database is disabled (disabling this is not uncommon).  So the current
proposed fix is to disable the icons database since the Browser does not
use it in any way.  Hopefully this will prove to be adequate.

The patch for this is attached.  It will be put into the 4.4 stream as
soon as it is open, and if it proves to work well over the summer then
it will be back-ported to the 4.3.1 stream as well.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/32

------------------------------------------------------------------------
On 2013-06-19T17:36:31+00:00 Marc-andre-laperle wrote:

(In reply to comment #32)
> Created attachment 232362 [details]
> patch to disable icons database (apply post-4.3)
> 
> Determining the root cause of this has proven to be quite difficult.  Even
> compiling WebKitGTK from source with instrumentation did not help because
> its characteristics did not match the WebKitGTK that's shipped with Ubuntu.

I think you should be able to get the matching source with:
apt-get source libwebkitgtk-1.0-0

> Anyways, one thing I was able to determine from the instrumented WebKitGTK
> is that the path that crashes is never run if WebKit's icons database is
> disabled (disabling this is not uncommon).  So the current proposed fix is
> to disable the icons database since the Browser does not use it in any way. 
> Hopefully this will prove to be adequate.
> 
> The patch for this is attached.  It will be put into the 4.4 stream as soon
> as it is open, and if it proves to work well over the summer then it will be
> back-ported to the 4.3.1 stream as well.

It's open now, I believe.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/33

------------------------------------------------------------------------
On 2013-06-25T15:52:34+00:00 Grant Gayed wrote:

Fixed > 20130625 in the 4.4 stream, commit:
http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=76d7d429442fe73692a90b1854cf0455cb3d83fa
.

If crashes are seen in WebKitGTK-based Browsers in native frames other
than webkitWebViewRegisterForIconNotification() then please log separate
reports.  This report deals specifically with this case.

Leaving report open and settings its Target Milestone to 4.3.1 as a
reminder to include this fix in 4.3.1 if it proves to work well in the
4.4 stream in the interim.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/34

------------------------------------------------------------------------
On 2013-07-05T20:57:35+00:00 Marc-andre-laperle wrote:

Created attachment 233150
Crash after fix

Unfortunately, I got this crash again using 4.4.0-I20130702-1230.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/35

------------------------------------------------------------------------
On 2013-07-08T16:10:32+00:00 Grant Gayed wrote:

(In reply to comment #35)
> Unfortunately, I got this crash again using 4.4.0-I20130702-1230.

I've taken a look and see that the fix is not in there.  It's in the
latest nightly builds, but I-builds require an explicit submit from the
component, and it looks like SWT did not do this for
4.4.0-I20130702-1230.  I'll ensure that we submit for the next I-build.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/36

------------------------------------------------------------------------
On 2013-07-10T11:44:24+00:00 Arunkumar-thondapu-i wrote:

*** Bug 412235 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/37

------------------------------------------------------------------------
On 2013-07-16T15:01:34+00:00 HENRY Florian wrote:

Hi all, I repoduce this bug on 4.3 with SWT_GTK3=1. I'll attacht the pid
error file

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/38

------------------------------------------------------------------------
On 2013-07-16T15:02:14+00:00 HENRY Florian wrote:

Created attachment 233523
pid error with 4.3 and SWT_GTK3=1

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/39

------------------------------------------------------------------------
On 2013-07-16T15:05:52+00:00 HENRY Florian wrote:

Well when I said reporduce... I still didn't find a way to reproduce it
as a test case, eclipse just ramdomly crash with this report. Somethimes
it stick grayed (freeze) and I just force quit and reluch. I also add a
-clean parameters in the first line of eclipse.ini

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/40

------------------------------------------------------------------------
On 2013-07-16T17:44:15+00:00 Grant Gayed wrote:

This is fixed in the 4.4 stream, it is not in the 4.3 release, nor in
the 4.3.1 stream yet (will mark this bug accordingly when it gets
there).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/41

------------------------------------------------------------------------
On 2013-07-16T18:09:16+00:00 Marc-andre-laperle wrote:

Just an update, 3 of us have been running 4.4.0 Build id: I20130709-0800
and this specific crash seems to be gone. However, we've been
experiencing what seems to be a stack overflow in the overlay-scrollbar
library (just by looking at addresses without debugging symbols). Grant,
if that's OK with you, I'd like to look into this crash this week before
the fix gets applied to 4.3.1 to make sure it's unrelated.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/42

------------------------------------------------------------------------
On 2013-07-16T20:29:59+00:00 Grant Gayed wrote:

(In reply to comment #42)
> I'd like to look into this crash this week before the fix gets applied to 4.3.1

Sure, the change can wait.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/43

------------------------------------------------------------------------
On 2013-07-18T21:44:18+00:00 Marc-andre-laperle wrote:

(In reply to comment #43)
> (In reply to comment #42)
> > I'd like to look into this crash this week before the fix gets applied to 4.3.1
> 
> Sure, the change can wait.

Looking at that other crash with debugging symbols, the crash seems
unrelated so I think this is ready to be applied to 4.3 (at least
testing wise...). Thank you for the fix!

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/44

------------------------------------------------------------------------
On 2013-07-19T14:40:33+00:00 Marc-andre-laperle wrote:

Created attachment 233623
Crash with 4.4.0-I20130717-1600

Got the crash again with 4.4.0-I20130717-1600.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/45

------------------------------------------------------------------------
On 2013-07-26T14:46:19+00:00 Paul Bußmann wrote:

Freshly downloaded eclipse-standard-kepler-R-linux-gtk.tar.gz and
unpacked it on Linux Mint 13. Installed AddOn pydev. "Switch to
workbench" crashes Eclipse.

Workaround to put "-Dorg.eclipse.swt.browser.DefaultType=mozilla" at end
of eclipse.ini worked for me.

Any reason for not making Mozilla default for eclipse-standard-kepler-R
-linux-gtk.tar.gz ? Looks like many users trapped on this bug (and its
duplicates).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/46

------------------------------------------------------------------------
On 2013-07-26T14:58:25+00:00 francesco wrote:

unfortunatly the workaround don't work fine for me (and I think for
others users too).

Eclipse seems more robust, but there are many crashes yet.

Is very difficult to debug this problem, as eclipse crash often without
any stack trace or logs..

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/47

------------------------------------------------------------------------
On 2013-08-12T22:45:48+00:00 Marc-andre-laperle wrote:

Interesting, I got the same crash with the same hs_err log containing
webkitWebViewRegisterForIconNotification but when I open the core dump,
the stack is different:

__GI_raise() at raise.c:56 0x7f1293217037	
__GI_abort() at abort.c:90 0x7f129321a698	
os::abort() at os_linux.cpp:1,593 0x7f12925e04a9	
VMError::report_and_die() at vmError.cpp:1,018 0x7f129273ad9f	
JVM_handle_linux_signal() at os_linux_x86.cpp:528 0x7f12925e781d	
<signal handler called>() at 0x7f12932170b0	
WebKit::core() at webkitwebview.cpp:5,368 0x7f11ecd786d5	
WebKit::GtkAdjustmentWatcher::updateAdjustmentsFromScrollbars() at GtkAdjustmentWatcher.cpp:65 0x7f11ecd51108	
WebKit::updateAdjustmentCallback() at GtkAdjustmentWatcher.cpp:76 0x7f11ecd51189	
g_timeout_dispatch() at gmain.c:4,413 0x7f127b9f3b1b	
g_main_dispatch() at gmain.c:3,054 0x7f127b9f2f05	
g_main_context_dispatch() at gmain.c:3,630 0x7f127b9f2f05	
g_main_context_iterate() at gmain.c:3,701 0x7f127b9f3248	
g_main_context_iteration() at gmain.c:3,762 0x7f127b9f3304	
Java_org_eclipse_swt_internal_gtk_OS__1g_1main_1context_1iteration() at 0x7f1249c72606	
0x7f1289820256	
0x7f12939ab220	
0x7f128c006b60	
0xdadf1ad0	
InterpreterRuntime::resolve_invoke() at interpreterRuntime.cpp:744 0x7f129240bd71	
0x7f128983fc84	
0x0

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/48

------------------------------------------------------------------------
On 2013-08-14T18:53:10+00:00 Grant Gayed wrote:

I've traced through the WebKitGTK source and identified all paths that
lead to webkitWebViewRegisterForIconNotification().  Once the icons
database has been disabled these paths all detect this and run away
before making this call.  By running a WebKitGTK instrumented with
fprintf's the only way I can see this function possibly being
encountered is if one of these paths happens before the database has
been disabled.

The Browser's first invocation of webkit_get_favicon_database(), which
is needed in order to disable it, seems like the only place where this
could happen in theory, because it creates a background thread that can
ultimately invoke webkitWebViewRegisterForIconNotification() after re-
joining the main thread.  The timing of this re-join is of course
variable.  In testing I was never able to get this case to actually
happen, but I did see some variance in ordering that occurred as a
result of the thread join happening at different times.

I've changed the fix in the 4.4 stream to only run for the first
WebKitGTK-based Browser since the icons database is shared amongst all
instances.  I'd like to put the change in 4.3.1, and am reasonably sure
that it's not dangerous, but the fact that the crash has been seen in
the 4.4 stream is a bit discouraging.  Does the frequency of this crash
seem at least diminished in 4.4 from what it was in 4.3?

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/49

------------------------------------------------------------------------
On 2013-08-15T19:27:26+00:00 Marc-andre-laperle wrote:

(In reply to comment #49)
> I've traced through the WebKitGTK source and identified all paths that lead
> to webkitWebViewRegisterForIconNotification().  Once the icons database has
> been disabled these paths all detect this and run away before making this
> call.  By running a WebKitGTK instrumented with fprintf's the only way I can
> see this function possibly being encountered is if one of these paths
> happens before the database has been disabled.
> 
> The Browser's first invocation of webkit_get_favicon_database(), which is
> needed in order to disable it, seems like the only place where this could
> happen in theory, because it creates a background thread that can ultimately
> invoke webkitWebViewRegisterForIconNotification() after re-joining the main
> thread.  The timing of this re-join is of course variable.  In testing I was
> never able to get this case to actually happen, but I did see some variance
> in ordering that occurred as a result of the thread join happening at
> different times.

Are you sure it has something to do with
webkitWebViewRegisterForIconNotification? I'm thinking what's being
reported in hs_err is wrong because my core dumps show a different call
stack (see previous comment).

> I've changed the fix in the 4.4 stream to only run for the first
> WebKitGTK-based Browser since the icons database is shared amongst all
> instances.  I'd like to put the change in 4.3.1, and am reasonably sure that
> it's not dangerous, but the fact that the crash has been seen in the 4.4
> stream is a bit discouraging.  Does the frequency of this crash seem at
> least diminished in 4.4 from what it was in 4.3?

It greatly varies. It didn't crash for a few days but I crashed 3 times
the same day, I'm at 6 this week.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/50

------------------------------------------------------------------------
On 2013-08-19T18:42:52+00:00 Pas wrote:

Hello,

I've just got bitten by this bug. A completely fresh download-unpack-
start of Eclipse.

The first time it started, after I OK-ed the workspace path, I clicked
go to workbench (on the still useless welcome screen), and then I double
clicked on the window's titlebar to maximize it (yes, 800x600 and 640K
should be enough for everybody, I know) and it crashed.

Then I restarted Eclipse, was able to create a new Java project, add two
jackson JARs to the library path, create a new class, then I started
typing, I saw the help autocomplete, then it got stuck and I had to
"force close" Eclipse.

Now every time I start it and start typing it instantly crashes.

I'll try with the integration build now.

Okay, it happens on that too. (
http://static.zomg.hu/hs_err_pid18729.log - 4.4)

Ubuntu 13.10, openjdk-7 7u25-2.3.12-4ubuntu1 (jvm 1.7.0_25), yadda-
yadda, see also, http://static.zomg.hu/hs_err_pid18237.log for the 4.3
log.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/51

------------------------------------------------------------------------
On 2013-08-20T15:58:27+00:00 Grant Gayed wrote:

It seems apparent that the change released to the 4.4 stream is not
fixing the problem, so this will not go into 4.3.1.  If the actual issue
is more like what's in comment 48 then it has the appearance of a
WebKitGTK bug (in WebCore::core() either webView or webView->priv is an
invalid ptr).

Will continue to investigate this angle, and am clearing Target
Milestone for now since there is not currently a fix for 4.3.1.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/52

------------------------------------------------------------------------
On 2013-08-20T17:02:23+00:00 Grant Gayed wrote:

(In reply to comment #52)
> If the actual issue is more like what's in comment 48 then it has the
> appearance of a WebKitGTK bug

Apparently fixed a few weeks ago:
https://bugs.webkit.org/show_bug.cgi?id=118396 .

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/53

------------------------------------------------------------------------
On 2013-08-20T17:27:54+00:00 Marc-andre-laperle wrote:

(In reply to comment #53)
> (In reply to comment #52)
> > If the actual issue is more like what's in comment 48 then it has the
> > appearance of a WebKitGTK bug
> 
> Apparently fixed a few weeks ago:
> https://bugs.webkit.org/show_bug.cgi?id=118396 .

Thanks! I pointed to the upstream on the Launchpad bug.

https://bugs.launchpad.net/ubuntu/+source/webkit/+bug/1160411

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/54

------------------------------------------------------------------------
On 2013-08-26T08:33:59+00:00 HENRY Florian wrote:

What I can say, this bugs seems to appear when after a commit and push from git plugin. Also with unbuntu 13.04 with I use ring switcher (from compizconfig rather tahn normal shift switcher), and this bug appens after multiple ALT+TAB applications switch. 
But since I've upgrade to M20130731-0800 it seems that this particular bug do not fire as often as it was before. It also could be partly solve by another update from open-JDK, or webkit, I don't know.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/55

------------------------------------------------------------------------
On 2013-09-20T05:35:20+00:00 Mchallss wrote:

Created attachment 235646
JVM crash logfile

JVM crash logfile

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/56

------------------------------------------------------------------------
On 2013-09-20T05:35:51+00:00 Mchallss wrote:

Created attachment 235647
JVM crash logfile

JVM crash logfile

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/57

------------------------------------------------------------------------
On 2013-09-20T05:36:33+00:00 Mchallss wrote:

I think I have the exact same problem, except it crashes somewhere else
in the webkit library (in GTK parts). I use Eclipse running on 64bit
Linux, via X over SSH to a Windows 7 machine. I'm using latest Cygwin as
X-server.

I also attatched the crash files, one segmentation fault, one illegal
instruction (!) :)

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/58

------------------------------------------------------------------------
On 2013-10-10T09:03:57+00:00 Lorenzo-bettini wrote:

Isn't there a way of forcing the use of webkit3? I have both 1 and 3
versions installed in my system, but I cannot remove version 1 (programs
like Gimp depend on it),

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/59

------------------------------------------------------------------------
On 2013-10-10T13:52:18+00:00 Grant Gayed wrote:

(In reply to Lorenzo Bettini from comment #59)

If you're using swt's GTK3 implementation (export SWT_GTK3=1 before
launching Eclipse) then it should use the webkit3 lib.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/60

------------------------------------------------------------------------
On 2013-10-21T13:24:30+00:00 francesco wrote:

Created attachment 236708
JVM crash file

JVM crash file.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/61

------------------------------------------------------------------------
On 2013-10-29T20:40:21+00:00 Marc-andre-laperle wrote:

(In reply to Grant Gayed from comment #60)
> (In reply to Lorenzo Bettini from comment #59)
> 
> If you're using swt's GTK3 implementation (export SWT_GTK3=1 before
> launching Eclipse) then it should use the webkit3 lib.

I got the crash twice today using GTK3 + webkit3 (4.4.0 I20131028-0800)
with SWT_GTK3=1. I'll apply the fix locally for
https://bugs.webkit.org/show_bug.cgi?id=119003 and test that for a
while.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/62

------------------------------------------------------------------------
On 2013-11-15T15:09:16+00:00 HENRY Florian wrote:

Hi all,
I don't know why but since few days, this bug don't appears anymore, maybe because I remove the -clean option into eclipse.ini or upgrade of GTK from Ubuntu (still 13.04).
I'm on Eclipse Version: Kepler Service Release 1
Build id: 20130919-0819
This comment is not usefull so resolve the bug itself, but at least if other user see it gone too...

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/63

------------------------------------------------------------------------
On 2013-11-19T11:52:08+00:00 francesco wrote:

In ubuntu 13.10, if a set SWT_GTK3=1 as global variable, the menus of
Eclipse disappear.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/64

------------------------------------------------------------------------
On 2013-11-19T14:45:43+00:00 Marc-andre-laperle wrote:

(In reply to Francesco D\'Amore from comment #64)
> In ubuntu 13.10, if a set SWT_GTK3=1 as global variable, the menus of
> Eclipse disappear.

Have you tried undefining UBUNTU_MENUPROXY? like UBUNTU_MENUPROXY=

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/65

------------------------------------------------------------------------
On 2013-11-19T15:08:06+00:00 francesco wrote:

No.
In order to make manu working in Eclipse Kepler and Ubuntu 13.10, a just changed some things in the launcher.

But this work around seems don't work if a set SWT_GT3=1.

All these errors are very annoying for Ubuntu users...   Eclipse appear
very unusable. Today it crashed five times in three hours!


In order to (In reply to Marc-Andre Laperle from comment #65)
> (In reply to Francesco D\'Amore from comment #64)
> > In ubuntu 13.10, if a set SWT_GTK3=1 as global variable, the menus of
> > Eclipse disappear.
> 
> Have you tried undefining UBUNTU_MENUPROXY? like UBUNTU_MENUPROXY=

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/66

------------------------------------------------------------------------
On 2013-11-19T15:43:07+00:00 francesco wrote:

Created attachment 237557
JAVA crash

java crash

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/67

------------------------------------------------------------------------
On 2013-11-19T18:18:02+00:00 Marc-andre-laperle wrote:

(In reply to Francesco D\'Amore from comment #66)
> No.
> In order to make manu working in Eclipse Kepler and Ubuntu 13.10, a just
> changed some things in the launcher.
> 
> But this work around seems don't work if a set SWT_GT3=1.

Can you describe your workaround? Or perhaps a link to it? I might look
into fixing this menu issue soon and any information is useful.

> All these errors are very annoying for Ubuntu users...   Eclipse appear very
> unusable. Today it crashed five times in three hours!

So far, everything as led me to believe that this particular crash has
been fixed in webkit and Ubuntu will get an updated version in Ubuntu
14.04. I have applied the fix locally on my 13.10 and haven't seen the
crash since. I would attach the .deb files here but I don't think this
would sit well with Eclipse policy since they are not EPL (?).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/68

------------------------------------------------------------------------
On 2013-11-19T22:45:58+00:00 Grant Gayed wrote:

(In reply to Marc-Andre Laperle from comment #68)
> I would attach the .deb files here but I don't think this would sit well with

Probably not.  Unfortunately the release date for Ubuntu 14.04 is still
far out.  Perhaps the Fedora 20 release will ship with it.

Are you running with 4.3.1 or a 4.4-stream build?  I'm wondering if the
comment 34 change could be relevant, which is only in the 4.4 stream.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/69

------------------------------------------------------------------------
On 2013-11-19T23:38:00+00:00 Marc-andre-laperle wrote:

(In reply to Grant Gayed from comment #69)
> (In reply to Marc-Andre Laperle from comment #68)
> Are you running with 4.3.1 or a 4.4-stream build?  I'm wondering if the
> comment 34 change could be relevant, which is only in the 4.4 stream.

Unfortunately, I'm running the 4.4 stream and got the crash multiple
times after the SWT fix, although it didn't manifest itself for a few
days. Then I applied the webkit fix and everything was fine.

I also tested without the webkit fix when I upgraded to Ubuntu 13.10 and
at that time SWT was switched to use GTK3 by default and I got the crash
again. I have patched my webkit again and so far so good.

> > I would attach the .deb files here but I don't think this would sit well with
> 
> Probably not.  Unfortunately the release date for Ubuntu 14.04 is still far
> out.  Perhaps the Fedora 20 release will ship with it.

If only this bug was easier to reproduce, the webkit patch could be
verified and it would make a better case for applying the patch to
13.10.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/70

------------------------------------------------------------------------
On 2013-11-20T05:43:12+00:00 Marc-andre-laperle wrote:

Created attachment 237569
Test project

Grant, could you try this test project and see if you are seeing the
crash? I can get it in a couple of seconds. Steps to reproduce:

Using 4.4 I20131119-0800, Ubuntu 13.10 64 bit

1. Open "Test Browser" view
2. Maximize the view (Double-click on the tab)
3. Move around the mouse a bit inside the view (optional?)

Sometimes I get the DumpRenderTreeSupportGtk crash, sometimes the
webkitWebViewRegisterForIconNotification but when I open the core file
with gdb they both show the back trace with the
updateAdjustmentsFromScrollbars call.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/71

------------------------------------------------------------------------
On 2013-11-20T06:01:04+00:00 HENRY Florian wrote:

This is exactly the two bug notification i've got before. Like I said,
with 13.04 and 20130919-0819 I don't get it anymore. It's strange
because I don't think the patch on webKit was apply into ubuntu, but it
seems to be the cause of these crash.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/72

------------------------------------------------------------------------
On 2013-11-20T06:27:39+00:00 Marc-andre-laperle wrote:

(In reply to Florian HENRY from comment #72)
> This is exactly the two bug notification i've got before. Like I said, with
> 13.04 and 20130919-0819 I don't get it anymore. It's strange because I don't
> think the patch on webKit was apply into ubuntu, but it seems to be the
> cause of these crash.

Do you still have the packages libwebkitgtk-1.0-0 and
libwebkitgtk-1.0-common installed? I still get the crash on 13.04,
20130919-0819 when I test with the attached project.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/73

------------------------------------------------------------------------
On 2013-11-20T07:36:34+00:00 HENRY Florian wrote:

I don't understand 
1-Open "Test Browser" view

I import the projet into eclipse, I can see it, it didn't crash. Do you
mean run it as Java application, or Applet ? I try without succes.

I'm sorry, I use Eclipse for PHP, I never develop java application with it so i don't know exactly what to do.
I have SWT_GTK3=1 in my launch script
and
libwebkitgtk-1.0-0				install
libwebkitgtk-1.0-common				install
libwebkitgtk-3.0-0				install
libwebkitgtk-3.0-common				install

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/74

------------------------------------------------------------------------
On 2013-11-20T07:44:44+00:00 Kaloyan Raev wrote:

(In reply to Marc-Andre Laperle from comment #68)
> So far, everything as led me to believe that this particular crash has been
> fixed in webkit and Ubuntu will get an updated version in Ubuntu 14.04. I
> have applied the fix locally on my 13.10 and haven't seen the crash since. I
> would attach the .deb files here but I don't think this would sit well with
> Eclipse policy since they are not EPL (?).

I believe you can just share the .deb files via dropbox or similar
service. In the end, WebKitGTK is an open source project and
redistribution is allowed.

Could you explain in more details about the fix you applied? Is this an
official newer version of WebKitGTK? Or you patched the one in Ubuntu
13.04 and rebuilt it?

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/75

------------------------------------------------------------------------
On 2013-11-20T09:54:34+00:00 francesco wrote:

(In reply to Marc-Andre Laperle from comment #68)
> Can you describe your workaround? Or perhaps a link to it? I might look into
> fixing this menu issue soon and any information is useful.

My eclipse launcher, showd in Unity toolbar, is a file stored in:
/home/<myhome>/.local/share/applications/eclipse.desktop

This is my launcher:
[Desktop Entry]
Encoding=UTF-8
Version=1.0
Type=Application
Name=Java - Eclipse 
Icon=eclipse.png
Path=/home/<myhome>/eclipse
Exec=env UBUNTU_MENUPROXY= /home/<myhome>/eclipse/eclipse
StartupNotify=false
StartupWMClass=Eclipse
OnlyShowIn=Unity;
X-UnityGenerated=true

In order to make menus working, you should add this line:
Exec=env UBUNTU_MENUPROXY= /home/<myhome>/eclipse/eclipse

My Eclipse is installed in /home/<myhome>/eclipse

> So far, everything as led me to believe that this particular crash has been
> fixed in webkit and Ubuntu will get an updated version in Ubuntu 14.04. I
> have applied the fix locally on my 13.10 and haven't seen the crash since. I
> would attach the .deb files here but I don't think this would sit well with
> Eclipse policy since they are not EPL (?).

Can you describe this webkit fix in more details??? Can you provide us
with a link?  Is very important to overcame this issue.... Eclipse in
ubuntu 13.10 is unstable....

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/76

------------------------------------------------------------------------
On 2013-11-20T14:57:19+00:00 Marc-andre-laperle wrote:

(In reply to Florian HENRY from comment #74)
> I don't understand 
> 1-Open "Test Browser" view
> 
> I import the projet into eclipse, I can see it, it didn't crash. Do you mean
> run it as Java application, or Applet ? I try without succes.

Sorry this message was directed to Grant so I didn't add as much detail. More steps:
0. Import the project in Eclipse. It's a plug-in project so you'll need PDE installed (Eclipse SDK, Classic, etc have it).
0.1 Run as Eclipse Application so that you get the view inside the launched Eclipse.

(In reply to Kaloyan Raev from comment #75)
> (In reply to Marc-Andre Laperle from comment #68)
> I believe you can just share the .deb files via dropbox or similar service.
> In the end, WebKitGTK is an open source project and redistribution is
> allowed.

Thanks for the suggestion, I was looking for something like that.

> Could you explain in more details about the fix you applied? Is this an
> official newer version of WebKitGTK? Or you patched the one in Ubuntu 13.04
> and rebuilt it?

I applied the fix locally on top of the Ubuntu 13.04 package then I patched it again when I upgraded to Ubuntu 13.10. The fix is included in later official versions of Webkit. I'm not sure exactly which version will get into 14.04 but it looks like it will be 2.3.1 (proposed) which includes the fix. The webkit bugzilla link has been posted before but I'll give you the direct link to the patch:
https://bug-119003-attachments.webkit.org/attachment.cgi?id=207317

(In reply to Francesco D\'Amore from comment #76)
> In order to make menus working, you should add this line:
> Exec=env UBUNTU_MENUPROXY= /home/<myhome>/eclipse/eclipse

Ah OK, I was referring to the same thing, we just misunderstood each
other a little bit.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/77

------------------------------------------------------------------------
On 2013-11-21T21:42:41+00:00 Marc-Andre Laperle wrote:

(In reply to Kaloyan Raev from comment #75)
> I believe you can just share the .deb files via dropbox or similar service.
> In the end, WebKitGTK is an open source project and redistribution is
> allowed.

Here are the patched .deb files for 13.10 64 bit:
https://www.dropbox.com/l/3fsKx3y53bmVoe80asoVKb

It doesn't include the optional debugging symbols (-dbg packages)
because they are quite big (> 1GB!).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/78

------------------------------------------------------------------------
On 2013-11-22T21:15:58+00:00 Grant Gayed wrote:

(In reply to Marc-Andre Laperle from comment #71)
> Grant, could you try this test project and see if you are seeing the crash?
> 
> 1. Open "Test Browser" view
> 2. Maximize the view (Double-click on the tab)
> 3. Move around the mouse a bit inside the view (optional?)

Your view crashes reliably for me if step 3 is replaced by "close the
view (crash!)".  My Ubuntu is 13.04 and my VM is from IBM, which could
account for the difference in the timing of the crashes (eg.- memory
getting corrupted once the view is maximized but the IBM VM gets lucky
and tolerates it longer than the Oracle VM).

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/79

------------------------------------------------------------------------
On 2013-11-25T15:41:30+00:00 Grant Gayed wrote:

*** Bug 421891 has been marked as a duplicate of this bug. ***

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/80

------------------------------------------------------------------------
On 2013-11-25T21:32:10+00:00 Marc-andre-laperle wrote:

(In reply to Grant Gayed from comment #79)
> Your view crashes reliably for me if step 3 is replaced by "close the view
> (crash!)". 

Thanks for reproducing it!

Here are the patched .deb files for 13.04 64 bit:
https://www.dropbox.com/l/fdlcEeDRbVoHuivn6RNLyd

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/81

------------------------------------------------------------------------
On 2013-11-26T11:35:09+00:00 francesco wrote:

Sorry for trivial question...

I would like to apply the patch for ubuntu 13.10. In the tar file
included in dropbox there are 15 .deb files.

Do I install each files in order to apply the patch?

Thank you...

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/82

------------------------------------------------------------------------
On 2013-11-26T18:30:05+00:00 Marc-andre-laperle wrote:

(In reply to Francesco D\'Amore from comment #82)
> Do I install each files in order to apply the patch?
 
You need to at least install (i.e. upgrade) all the ones you have currently installed otherwise you will most likely have broken dependencies. All the packages included in the tar are built from the same webkit source tree so they all have the same Ubuntu version. If an update comes from Ubuntu they will all get replaced. You can install/upgrade only the ones you have currently installed but I think it's easier to just install all of them:

sudo dpkg -i *.deb

If you want to revert to the original packages from Ubuntu servers, you
can specify /saucy when re-installing the packages with apt-get, for
example:

sudo apt-get install gir1.2-javascriptcoregtk-*/saucy
gir1.2-webkit-*/saucy  libjavascriptcoregtk-*/saucy libwebkit-dev/saucy
libwebkitgtk*/saucy

For 13.04 it would be /raring.

Let me know if you see an improvement. On our end, we tested the fix on
5 different PCs and VMs using the test plug-in I attached and it seemed
to resolved the issue. Keep in mind that there are other possible
crashes so check your crash log.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/83

------------------------------------------------------------------------
On 2013-12-06T12:52:53+00:00 francesco wrote:

Even after the patch the problem arise more and mor, working with
Eclipse.

Unfortunatly there aren't JVM crash files available for logging.

Are you experienced the same issue?

Thank you.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/84

------------------------------------------------------------------------
On 2013-12-06T13:13:06+00:00 Frits Jalvingh wrote:

(In reply to Francesco D\'Amore from comment #84)
> Unfortunatly there aren't JVM crash files available for logging. 
Well, I had that too; I managed to get proper jvm logfiles by using a start script instead of Eclipse's launcher. Create a file "start.sh" in your eclipse home with:

#!/bin/bash
ulimit -c unlimited
/opt/java/jdk1.7.0_25/bin/java -XX:MaxPermSize=512m -Xmx2048m -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null -Dorg.eclipse.swt.browser.DefaultType=mozilla -jar plugins/org.eclipse.equinox.launcher_*.jar -console

Mark it executable (chmod a+x start.sh) and use it to start Eclipse.
This at least produces the jvm logs again.

In addition, for me Eclipse on Ubuntu 13.04 has been *rock solid*, both
4.2 and 4.3, no crashes *at all* with daily use for many hours a day
(yes, lack of real life ;-) by doing the following:

* Disable browser support by adding the 
-Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null 
-Dorg.eclipse.swt.browser.DefaultType=mozilla 
parameters to eclipse.ini

* Apply the workaround for bug# 416869:
sudo apt-get remove overlay-scrollbar

With those things no crashes.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/85

------------------------------------------------------------------------
On 2013-12-06T13:42:40+00:00 francesco wrote:

Thank you Frits...  I will try it.

I hope that in the next releases, of both Eclipse and Ubuntu, these
problems will be fixed.


(In reply to Frits Jalvingh from comment #85)
> (In reply to Francesco D\'Amore from comment #84)
> > Unfortunatly there aren't JVM crash files available for logging. 
> Well, I had that too; I managed to get proper jvm logfiles by using a start
> script instead of Eclipse's launcher. Create a file "start.sh" in your
> eclipse home with:
> 
> #!/bin/bash
> ulimit -c unlimited
> /opt/java/jdk1.7.0_25/bin/java -XX:MaxPermSize=512m -Xmx2048m
> -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null
> -Dorg.eclipse.swt.browser.DefaultType=mozilla -jar
> plugins/org.eclipse.equinox.launcher_*.jar -console
> 
> Mark it executable (chmod a+x start.sh) and use it to start Eclipse. This at
> least produces the jvm logs again.
> 
> In addition, for me Eclipse on Ubuntu 13.04 has been *rock solid*, both 4.2
> and 4.3, no crashes *at all* with daily use for many hours a day (yes, lack
> of real life ;-) by doing the following:
> 
> * Disable browser support by adding the 
> -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null 
> -Dorg.eclipse.swt.browser.DefaultType=mozilla 
> parameters to eclipse.ini
> 
> * Apply the workaround for bug# 416869:
> sudo apt-get remove overlay-scrollbar
> 
> With those things no crashes.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/86

------------------------------------------------------------------------
On 2013-12-06T15:22:17+00:00 Marc-andre-laperle wrote:

(In reply to Francesco D\'Amore from comment #84)
> Even after the patch the problem arise more and mor, working with Eclipse.
> 
> Unfortunatly there aren't JVM crash files available for logging. 
> 
> Are you experienced the same issue?
> 
> Thank you.

Hi Francesco. There are a few other bugs that could be still affecting
you. The other crash that I got really often is the one related to
overlay-scrollbar triggering a stack overflow. I also saw this on a
colleague's 13.04. Could you try still using webkit but disabling
overlay-scrollbar?

sudo apt-get remove overlay-scrollbar
or start Eclipse with LIBOVERLAY_SCROLLBAR=0 environment variable.

I'm trying to get webkit fixed in Ubuntu by backporting the fix so any
help in testing is appreciated.

One other crash is Bug 420550, but it should be unrelated to webkit or
overlay-scrollbar.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/87

------------------------------------------------------------------------
On 2013-12-06T15:28:33+00:00 francesco wrote:

Ok...

I will apply the fix next week.

Thank you for your help.

(In reply to Marc-Andre Laperle from comment #87)
> (In reply to Francesco D\'Amore from comment #84)
> > Even after the patch the problem arise more and mor, working with Eclipse.
> > 
> > Unfortunatly there aren't JVM crash files available for logging. 
> > 
> > Are you experienced the same issue?
> > 
> > Thank you.
> 
> Hi Francesco. There are a few other bugs that could be still affecting you.
> The other crash that I got really often is the one related to
> overlay-scrollbar triggering a stack overflow. I also saw this on a
> colleague's 13.04. Could you try still using webkit but disabling
> overlay-scrollbar?
> 
> sudo apt-get remove overlay-scrollbar
> or start Eclipse with LIBOVERLAY_SCROLLBAR=0 environment variable.
> 
> I'm trying to get webkit fixed in Ubuntu by backporting the fix so any help
> in testing is appreciated.
> 
> One other crash is Bug 420550, but it should be unrelated to webkit or
> overlay-scrollbar.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/88

------------------------------------------------------------------------
On 2013-12-06T19:06:14+00:00 Grant Gayed wrote:

(In reply to Marc-Andre Laperle from comment #87)

I can confirm that the test case stopped crashing for me after applying
the comment 81 packages.

I don't use Eclipse on Ubuntu on a frequent enough basis to
confirm/refute a change in its regular day-to-day use.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/89

------------------------------------------------------------------------
On 2013-12-10T15:14:09+00:00 francesco wrote:

Thank you Frits.

After applying the patch related to webkit and scrollbar, as you
suggested, eclipse seems to be very robust and stable.

Take in consideration that I installed also the .deb files about
webwkit, suggested in #comment 83

After two days coding, eclipse not crash yet.


(In reply to Frits Jalvingh from comment #85)
> (In reply to Francesco D\'Amore from comment #84)
> > Unfortunatly there aren't JVM crash files available for logging. 
> Well, I had that too; I managed to get proper jvm logfiles by using a start
> script instead of Eclipse's launcher. Create a file "start.sh" in your
> eclipse home with:
> 
> #!/bin/bash
> ulimit -c unlimited
> /opt/java/jdk1.7.0_25/bin/java -XX:MaxPermSize=512m -Xmx2048m
> -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null
> -Dorg.eclipse.swt.browser.DefaultType=mozilla -jar
> plugins/org.eclipse.equinox.launcher_*.jar -console
> 
> Mark it executable (chmod a+x start.sh) and use it to start Eclipse. This at
> least produces the jvm logs again.
> 
> In addition, for me Eclipse on Ubuntu 13.04 has been *rock solid*, both 4.2
> and 4.3, no crashes *at all* with daily use for many hours a day (yes, lack
> of real life ;-) by doing the following:
> 
> * Disable browser support by adding the 
> -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null 
> -Dorg.eclipse.swt.browser.DefaultType=mozilla 
> parameters to eclipse.ini
> 
> * Apply the workaround for bug# 416869:
> sudo apt-get remove overlay-scrollbar
> 
> With those things no crashes.

Reply at:
https://bugs.launchpad.net/ubuntu/+source/openjdk-7/+bug/1266004/comments/90


** Changed in: eclipse
       Status: Unknown => Confirmed

** Changed in: eclipse
   Importance: Unknown => Medium

** Bug watch added: bugs.webkit.org/ #118396
   https://bugs.webkit.org/show_bug.cgi?id=118396

** Bug watch added: bugs.webkit.org/ #119003
   https://bugs.webkit.org/show_bug.cgi?id=119003

-- 
You received this bug notification because you are a member of OpenJDK,
which is subscribed to openjdk-7 in Ubuntu.
https://bugs.launchpad.net/bugs/1266004

Title:
  Eclipse crash

Status in Eclipse:
  Confirmed
Status in “openjdk-7” package in Ubuntu:
  New

Bug description:
  No obvious cause, just using the Java editor

  1)  Description:	Ubuntu 13.04 Release:	13.04 64 bit
  2)  Eclipse Java EE IDE for Web Developers.  Version: Kepler Service Release 1  Build id: 20130919-0819
  3)  Modify some Java
  4) Crash!


  Log file attached

To manage notifications about this bug go to:
https://bugs.launchpad.net/eclipse/+bug/1266004/+subscriptions