Community
Participate
Working Groups
Eclipse crashes when displaying JavaDoc which requires a scroll bar. This applies to in-editor pop-ups as well as to the 'Javadoc' view. I am unable to get a stack trace ( nothing in the logs ) and there is nothing printed on the command line except 'Aborted'. I have run the program with export LIBXCB_ALLOW_SLOPPY_LOCK=1 but it had no effect. OS: OpenSUSE 10.3 Beta1 Kernel: Linux rm 2.6.22.1-16-default #1 SMP 2007/08/06 15:19:41 UTC i686 i686 i386 GNU/Linux Eclipse: I20070621-1340 GTK: gtk2-2.11.6-7 Browser-related packages: mozilla-xulrunner181-1.8.1.4-6 MozillaFirefox-2.0.0.5-9
For completeness' sake, the crash also happens when opening the 'Internal Web Browser' view.
Was previously cross-posted at https://bugzilla.novell.com/show_bug.cgi?id=293439
Is this always reproducible? Try running with ./eclipse -consoleLog to see if there is anything in the error stream.
I can always reproduce this. Shortest path is to switch on the internal browser view. Log output using -consoleLog only displays !SESSION 2007-08-14 16:48:12.161 ----------------------------------------------- eclipse.buildId=I20070625-1500 java.version=1.5.0_12 java.vendor=Sun Microsystems Inc. BootLoader constants: OS=linux, ARCH=x86, WS=gtk, NL=en_US Command-line arguments: -os linux -ws gtk -arch x86 -consoleLog !ENTRY csdl.jupiter 1 0 2007-08-14 16:48:18.820 !MESSAGE Jupiter Review Plugin is up. Aborted
Can you try out the 3.3 release by itself (ie. no additional plugins) and see if you still have the problem?
Using a fresh download of 'Eclipse IDE for Java Developers' from http://www.eclipse.org/downloads/download.php?file=/technology/epp/downloads/release/20070702/eclipse-java-europa-linux-gtk.tar.gz&r=1&protocol=http and a new workspace, the first run crashes after briefly showing the welcome screen. The second run is successful, but on opening the internal web browser view, it fails again. Strangely, running with -consoleLog displays no output except 'Aborted' and there are no .log files in the .metadata directory
If you do a search for hs_err files does anything come up? (not all .log flies are written to the .metadata). (Do something like 'find . -name hs*' making sure to include both your workspace and eclipse install directories).
Created attachment 76049 [details] Error file (mistakenly attached) This is the only error file I could find. I think it was one of the first crashes. After that no hs_err_ file is generated.
Unfortunately, the trace doesn't give us much to go on. Grant, any suggestions on how to proceed?
I am the one who originally reported the novell bug report. running with -consolelog I get the following. # /usr/share/eclipse/eclipse.bin -debug -consolelog -startup /usr/share/eclipse/plugins/org.eclipse.equinox.launcher_1.0.0.v20070606.jar Start VM: -Djava.class.path=/usr/share/eclipse/plugins/org.eclipse.equinox.launcher_1.0.0.v20070606.jar -os linux -ws gtk -arch x86 -showsplash -launcher /usr/lib/eclipse/eclipse -name Eclipse --launcher.library /usr/lib/eclipse/plugins/org.eclipse.equinox.launcher.gtk.linux.x86_1.0.0.v20070606/eclipse_1017a.so -startup /usr/share/eclipse/plugins/org.eclipse.equinox.launcher_1.0.0.v20070606.jar -debug -consolelog -vm /usr/lib/jvm/java-1.6.0.u1-sun-1.6.0.u1/jre/bin/../lib/i386/client/libjvm.so -vmargs -Djava.class.path=/usr/share/eclipse/plugins/org.eclipse.equinox.launcher_1.0.0.v20070606.jar Install location: file:/usr/share/eclipse/ Configuration file: file:/usr/share/eclipse/configuration/config.ini loaded Configuration location: file:/home/fm/.eclipse/org.eclipse.platform_3.3.1_1543616141/configuration/ Configuration file: file:/home/fm/.eclipse/org.eclipse.platform_3.3.1_1543616141/configuration/config.ini not found or not read Shared configuration location: file:/usr/share/eclipse/configuration/ Framework located: file:/usr/share/eclipse/plugins/org.eclipse.osgi_3.3.0.v20070530.jar Framework classpath: file:/usr/share/eclipse/plugins/org.eclipse.osgi_3.3.0.v20070530.jar Splash location: /usr/share/eclipse/plugins/org.eclipse.platform_3.3.0.v20070613/splash.bmp Debug options: file:/home/fm/.options not found Time to load bundles: 4 Starting application: 1127 Application Started: 4838 It has been crashing for weeks now and I have reproduced it at least 50 times.
I am not going to pretend I understand too much of the mechanics involved, but here goes: a) During an unrelated development session I tried to use selenium to invoke firefox from java. It failed with /usr/lib/firefox/firefox-bin: symbol lookup error: /usr/lib/firefox/firefox-bin: undefined symbol: JS_SaveFrameChain b) I tried to gather information about JS_SaveFrameChain, and found out that it is contained in the Firefox source code (declared in js/src/api.h, downloaded from ftp.mozilla.org) c) Looking at the source code from the provided xulrunner ( mozilla-xulrunner181-1.8.1.4-9 ) I found that the js/jsapi.h file did not include such a declaration. It seems to be that this _might_ cause a capabilities mismatch between what eclipse expects and what mozilla offers.
Robert is the xulrunner-1.8.1.4-6 on your system in /usr/lib/xulrunner-1.8.1b2? I ask because the names don't match up, but /usr/lib/xulrunner-1.8.1b2 is on your LD_LIBRARY_PATH in your attached vm dump, and appears throughout the dump. And do you know how it got on your LD_LIBRARY_PATH (did you add it)?
(In reply to comment #12) > Robert is the xulrunner-1.8.1.4-6 on your system in /usr/lib/xulrunner-1.8.1b2? No, it is actually in /usr/lib/xulrunner-1.8.1.4. That directory does not exist on my computer. It might be that it was an older error file, since I am not able to reproduce the error now, and I have mistakenly attached it. lrwxrwxrwx 1 root root 17 2007-08-12 00:08 xulrunner-1.8.1 -> xulrunner-1.8.1.4 drwxr-xr-x 11 root root 840 2007-08-12 00:08 xulrunner-1.8.1.4 > I ask because the names don't match up, but /usr/lib/xulrunner-1.8.1b2 is on > your LD_LIBRARY_PATH in your attached vm dump, and appears throughout the dump. > And do you know how it got on your LD_LIBRARY_PATH (did you add it)? > I did not add it manually, it is probably the distribution. I have tried using the BEA JVM, but the result is still the same, and I get no error output.
Works fine with the ibm sdk, all cases of browser/javadoc do not cause errors: robert@rm:~> /opt/ibm-java2-i386-50/bin/java -version java version "1.5.0" Java(TM) 2 Runtime Environment, Standard Edition (build pxi32devifx-20070806 (SR5a)) IBM J9 VM (build 2.3, J2RE 1.5.0 IBM J9 2.3 Linux x86-32 j9vmxi3223-20070426 (JIT enabled) J9VM - 20070420_12448_lHdSMR JIT - 20070419_1806_r8 GC - 200704_19) JCL - 20070725
xulrunner 1.8.1.4 is not available as a binary download from mozilla.org, so I've compiled it on my RHEL4 box and confirmed that it works for me (using a sun jre). It's surprising that switching to an ibm jre makes this work for you. In the scenario that fails for you (using a sun jre) can you ensure that there are no entries on your LD_LIBRARY_PATH that point at any mozilla/firefox/xulrunner installs on your machine?
Created attachment 76932 [details] here are all the environment variables of a programm started from eclipse The attachment is generated by: public class environment { public static void main (String[] arg) { java.util.Map<String,String> map = System.getenv(); for (java.util.Map.Entry<String, String> entry : map.entrySet()) { System.out.println(entry.toString()); } } }
Created attachment 76938 [details] i had some old symlinks arround, this should be the more correct environment
The following snippet will print the XULRunner that's being found and used on your machine. Can you run it, just to ensure that another native renderer is not being found? public static void main (String [] args) { Device.DEBUG = true; final Display display = new Display (); final Shell shell = new Shell (display); shell.setBounds(10,10,200,200); Browser browser = new Browser(shell, SWT.NONE); browser.setBounds(10,10,150,150); shell.open (); while (!shell.isDisposed ()) { if (!display.readAndDispatch ()) display.sleep (); } display.dispose (); }
(In reply to comment #18) > The following snippet will print the XULRunner that's being found and used on > your machine. Can you run it, just to ensure that another native renderer is > not being found? Your code runs after I added: import org.eclipse.swt.SWT; import org.eclipse.swt.widgets.*; import org.eclipse.swt.browser.*; import org.eclipse.swt.graphics.*; And downloaded http://sunsite.informatik.rwth-aachen.de/eclipse/downloads/drops/R-3.3-200706251500/swt-3.3-gtk-linux-x86.zip which I imported into my project. Your program outputs ... XULRunner path: /usr/lib/xulrunner-1.8.1.4/libxpcom.so ... and on closing the window crashes with ... # # An unexpected error has been detected by Java Runtime Environment: # # SIGSEGV (0xb) at pc=0xacabd3c0, pid=12363, tid=3084921744 # # Java VM: Java HotSpot(TM) Client VM (1.6.0_01-b06 mixed mode) # Problematic frame: # C [libxmlextras.so+0x13c0] # # An error report file with more information is saved as hs_err_pid12363.log # # If you would like to submit a bug report, please visit: # http://java.sun.com/webapps/bugreport/crash.jsp # openSUSE delivers a "libswt3-gtk2-3.3-22" which contains ... # rpm -ql libswt3-gtk2-3.3-22 /usr/lib/eclipse /usr/lib/eclipse/plugins /usr/lib/eclipse/plugins/org.eclipse.swt.gtk.linux.x86_3.3.0.v3346.jar /usr/share/eclipse /usr/share/eclipse/plugins /usr/share/eclipse/plugins/org.eclipse.swt.gtk.linux.x86_3.3.0.v3346.jar /usr/share/java/swt-gtk-3.3.jar ... but I could not really find out what they are good for.
Created attachment 76962 [details] java crash log I hope this is interesting for you. ;)
I installed http://ftp-stud.fht-esslingen.de/pub/Mirrors/eclipse/eclipse/downloads/drops/R-3.2.2-200702121330/eclipse-SDK-3.2.2-linux-gtk.tar.gz on the same machine and it is working without a problem. So it really seems to be caused by eclipse. Are there any other version I should try?
To confirm, does your 3.2.2 download work for you if you set MOZILLA_FIVE_HOME and LD_LIBRARY_PATH to /usr/lib/xulrunner-1.8.1.4 before running eclipse? The GRE detection changed between 3.2.2 and 3.3 so it needs to be verified that the working and non-working cases are using the same GRE.
MOZILLA_FIVE_HOME=/usr/lib/xulrunner-1.8.1.4/ LD_LIBRARY_PATH=/usr/lib/xulrunner-1.8.1.4/:$LD_LIBRARY_PATH Desktop/eclipse/eclipse does not crash displaying the "Help -> Welcome" and the "JavaDoc Tooltips" although it does crash at random places too :-(.
The core-dump inspected with gdb gives the following. Program terminated with signal 6, Aborted. #0 0xffffe410 in __kernel_vsyscall () (gdb) bt #0 0xffffe410 in __kernel_vsyscall () #1 0xb79d18f5 in *__GI_raise (sig=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64 #2 0xb79d31e1 in *__GI_abort () at abort.c:88 #3 0xb7a0e4db in malloc_printerr (action=2, str=0xb7abbe7b "free(): invalid pointer", ptr=0x1b26) at malloc.c:5892 #4 0xb7a10125 in *__GI___libc_free (mem=0x80756d8) at malloc.c:3586 #5 0xb73de665 in �� () from /usr/lib/eclipse/plugins/org.eclipse.equinox.launcher.gtk.linux.x86_1.0.0.v20070606/eclipse_1017a.so #6 0x080492e7 in ?? () #7 0x00000003 in ?? () #8 0xbfe20d54 in ?? () #9 0x00000000 in ?? ()
To my knowledge eclipse 3.2.2 and 3.3 do not have these problems (random crashes and intro page crashes, respectively) on opensuse 10.2, so something relevant has changed underneath. I was going to suggest rebuilding swt's libraries on your end, but I see in comment 19 that (presumably) this has already been done in the jars in the rpm. Does running with one of these jars rather than the downloaded one behave any differently? The only other possibility that comes to mind is that there's something strange with the shipped xulrunner.
Created attachment 77939 [details] screenshot of eclipse after removing xulrunner As suggested in the novell bug I removed xulrunner with # rpm -e mozilla-xulrunner181 --nodeps Now eclipse does not crash anymore. But the background image of help->about is missing?! (see attached screenshot) What is the exact relation between xulrunner and eclipse? The openSUSE Eclipse RPM does not even require xulrunner ...
Hm, a tad more information here: Program terminated with signal 6, Aborted. #0 0xffffe410 in __kernel_vsyscall () (gdb) bt #0 0xffffe410 in __kernel_vsyscall () #1 0xb79298f5 in *__GI_raise (sig=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64 #2 0xb792b1e1 in *__GI_abort () at abort.c:88 #3 0xb79664db in malloc_printerr (action=2, str=0xb7a13e7b "free(): invalid pointer", ptr=0x4325) at malloc.c:5892 #4 0xb7968125 in *__GI___libc_free (mem=0x8c93990) at malloc.c:3586 #5 0x0622d1e9 in jni_ReleaseBooleanArrayElements () from /usr/lib/jvm/java-1.6.0.u1-sun-1.6.0.u1/jre/bin/../lib/i386/client/libjv m.so #6 0xac69b5a2 in Java_org_eclipse_swt_internal_mozilla_XPCOM_VtblCall__II_3Z () from /home/skh/.eclipse/org.eclipse.platform_3.3.1_1543616141/configuration/o rg.eclipse.osgi/bundles/83/1/.cp/libswt-xulrunner-gtk-3346.so #7 0xb528a67e in ?? () #8 0x080798e8 in ?? () #9 0xbfce3450 in ?? () #10 0x00000016 in ?? () #11 0x08cce178 in ?? () #12 0xbfce3458 in ?? () #13 0x08079e20 in ?? () #14 0xbfce3428 in ?? () #15 0xb25da764 in ?? () ---Type <return> to continue, or q <return> to quit--- #16 0xbfce3460 in ?? () #17 0xb25decc0 in ?? () #18 0x00000000 in ?? () (also openSUSE 10.3, beta3) How can I build, or where can I download, libswt with debug information? We are indeed rebuilding them in the openSUSE packages, a hint how to tweak the build process to keep debuginfo in would be helpful. Thanks.
re: comment 26 The Welcome page with the white background indicates that a native browser was not found, which makes sense since XULRunner was uninstalled. If there's another browser on the system (eg.- a dynamically-linked firefox, a seamonkey, etc.) then setting MOZILLA_FIVE_HOME and LD_LIBRARY_PATH to point at it will help the Browser find it. As of eclipse 3.3 the Browser's startup goes: - try to find an installed-and-registered XULRunner: this is done with staticly-linked XULRunner code which I think basically looks for entries in /etc/gre.d/ - if one could not be found then see if environment variable MOZILLA_FIVE_HOME is set, and if so then try to use it (typically used to point at a native browser that is not meant to be registered the way XULRunner is, such as a dynamically-linked firefox, a seamonkey, etc.) - if this fails too then return an error Clients of the Browser (eg.- the Welcome page) are responsible for catching the error and handling it some other way (eg.- in this case showing a faked Welcome page with a white background). So XULRunner is not an absolute dependency for eclipse, but for the Browser to work it needs to find a native browser to use, and XULRunner has become the "proper" native renderer to use. Alternatives like a dynamically-linked firefox, a seamonkey, etc. should still continue to work. re: comment 27 Defining NO_STRIP when compiling should do this (see make_linux.mak).
*** Bug 202666 has been marked as a duplicate of this bug. ***
those were the last words when I was running the JVM with -verbose:jni [Dynamic-linking native method org.eclipse.swt.internal.gtk.OS._gtk_tooltips_data_get ... JNI] [Dynamic-linking native method org.eclipse.swt.internal.gtk.OS.GTK_TOOLTIPS_SET_ACTIVE ... JNI] [Dynamic-linking native method org.eclipse.swt.internal.gtk.OS._g_signal_handler_disconnect ... JNI] [Dynamic-linking native method org.eclipse.swt.internal.mozilla.XPCOM.VtblCall ... JNI] [Dynamic-linking native method org.eclipse.swt.internal.mozilla.XPCOM.VtblCall ... JNI]
And those are the two classes that are getting loaded in the time of death [Loaded org.eclipse.swt.internal.mozilla.nsIDOMUIEvent from file:/usr/local/share/eclipse/3.3.1-x86_64/plugins/org.eclipse.swt.gtk.linux.x86_64_3.3.1.v3346h.jar] [Loaded org.eclipse.swt.internal.mozilla.nsIDOMMouseEvent from file:/usr/local/share/eclipse/3.3.1-x86_64/plugins/org.eclipse.swt.gtk.linux.x86_64_3.3.1.v3346h.jar]
On upgrading to the latest 10.3 pre-release, the browser view works, as well as the other components. Not sure what really made it work, here are the updated versions: OS: openSUSE 10.3 (i586) Beta3plus Kernel: Linux rm 2.6.22.5-18-default #1 SMP 2007/09/13 11:40:13 UTC i686 i686 i386 GNU/Linux Eclipse: I20070621-1340 GTK: gtk2-2.11.6-26 Browser-related packages: mozilla-xulrunner181-1.8.1.4-26 MozillaFirefox-2.0.0.6-20 Java is vendor-supplied: java version "1.6.0_01"
Ok I have sad news to report this bug popped up again with current factory. Versions: mozilla-xulrunner181-1.8.1.4-47 eclipse-platform-3.3-63 libswt3-gtk2-3.3-61 gtk2-2.12.0-15
This bug has appeared for me under eclipse 3.3 and eclipse 3.4M5. I am running KDE4 (4.0.2) over a Mandriva 2008 installation. While the program runs correctly in KDE 3.5.9 it fails whenever the mouse floats over the javadoc windows (including the blue Ctrl-Space javadoc window). Adding export MOZILLA_FIVE_HOME=/usr/bin/firefox Resolves the problem. I think this relates to KDE4 handling default browsers differently to KDE 3.5 Adding -consoleLog gives no information about the failure except the following line which occurs in 3.4M5 when the failure occurs. DOUBLE-CLICK: 250 --> -1 THRESHOLD: 8 --> -1 Aborted OS: Mandriva 2008 Kernel: 2.6.22.18-desktop-1mdv #1 SMP Mon Feb 11 13:53:50 EST 2008 i686 GTK: gtk+2.0-2.12.1-2.1mdv2008.0 mozilla-firefox-2.0.0.12-1.1mdv2008.0 libmozilla-firefox2.0.0.12-2.0.0.12-1.1mdv2008.0 Note: xulrunner is not installed (I do not believe it is an installable package in Mandriva)
Actually setting the following environmental variable works too: export MOZILLA_FIVE_HOME= Looking at the bug description more carefully the export should have been export MOZILLA_FIVE_HOME=/usr/lib/firefox-2.0.0.12 As it is the home of the browser. Interestingly using this export eclipse crashes. For me it appears that if eclipse CAN find the browser it crashes.
If you're seeing the crash happen when you mouse over a browser instance (eg.- javadoc view, welcome page, etc.) then see bug 209428.
Comment 31 gives me some hope that this may be fixed as a side effect of the fix for bug 209428. Can someone try the nightly build at http://download.eclipse.org/eclipse/downloads/drops/N20080306-0100/eclipse-SDK-N20080306-0100-linux-gtk.tar.gz and follow up here whether it helped or not?
All is quiet, so marking report as a duplicate. If this problem still happens with build http://download.eclipse.org/eclipse/downloads/drops/S-3.4M7-200805020100/index.php then please reopen this report, thanks. *** This bug has been marked as a duplicate of bug 209428 ***
