Bug 444672

Summary: OpenJDK crash when HotSpot compiles Eclipse's org.eclipse.core.internal.dtree.DataTreeNode.forwardDeltaWith
Product: [Fedora] Fedora Reporter: Thomas Fitzsimmons <fitzsim>
Component: java-1.6.0-openjdkAssignee: Andrew Haley <aph>
Status: CLOSED CURRENTRELEASE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: low    
Version: 9CC: aph, langel, metal3d, overholt, patrickm, veguilla
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-08-21 13:11:27 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Thomas Fitzsimmons 2008-04-29 20:52:02 UTC
This bug tracks progress getting a fix for the upstream bug:

http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6614100

into Fedora.  This bug has more details and a workaround:

https://bugs.eclipse.org/bugs/show_bug.cgi?id=214092

Description of problem:

HotSpot may crash when it compiles Eclipse's
org.eclipse.core.internal.dtree.DataTreeNode.forwardDeltaWith.

Version-Release number of selected component (if applicable):

java-1.6.0-openjdk-1.6.0.0-0.9.b09.fc9.i386

How reproducible:

Always.

Steps to Reproduce:
1. Run Eclipse without this line in /usr/lib/eclipse/eclipse.ini

  
-XX:CompileCommand=exclude,org/eclipse/core/internal/dtree/DataTreeNode,forwardDeltaWith

Actual results:

HotSpot crashes with a message like:

#
# An unexpected error has been detected by Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00002aaaaacd5fa1, pid=8904, tid=1080056144
#
# Java VM: IcedTea 64-Bit Server VM (1.7.0-b21 mixed mode linux-amd64)
# Problematic frame:
# V  [libjvm.so+0x205fa1]
#
# If you would like to submit a bug report, please visit:
#   http://java.sun.com/webapps/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

...

Current CompileTask:
C2:1219     
org.eclipse.core.internal.dtree.DataTreeNode.forwardDeltaWith([Lorg/eclipse/core/internal/dtree/AbstractDataTreeNode;[Lorg/eclipse/core/internal/dtree/AbstractDataTreeNode;Lorg/eclipse/core/internal/dtree/IComparator;)[Lorg/eclipse/core/internal/dtree/AbstractDataTreeNode;
(469 bytes)

Expected results:

No crash.

Additional info:

Filing under x86-64 since that seems to be the most commonly-reported case.

Comment 1 Thomas Fitzsimmons 2008-04-29 20:53:54 UTC
eclipse-3.3.2-11.fc9 contains a workaround but HotSpot needs to be fixed.

Comment 2 Thomas Fitzsimmons 2008-04-29 20:54:38 UTC
*** Bug 441403 has been marked as a duplicate of this bug. ***

Comment 3 Andrew Haley 2008-05-01 16:18:01 UTC
I have eclipse-platform-3.3.2-9.fc9.x86_64 installed.

java-1.6.0-openjdk-devel-1.6.0.0-0.7.b08.fc9.x86_64

 $ cat /usr/lib64/eclipse/eclipse.ini
-showsplash
org.eclipse.platform
-vmargs
-Xms40m
-Xmx256m
-XX:MaxPermSize=128m

I see no crash.  Is this new with b09 ?


Comment 5 Thomas Fitzsimmons 2008-05-01 17:43:05 UTC
I haven't tried reproducing this yet.  I filed this bug to summarize 441403 and
give the issue a searchable summary.  /usr/lib/eclipse/eclipse.ini should be
/usr/lib64/eclipse/eclipse.ini, and "How reproducible" should be "?" -- sorry.

Is there a way to enable HotSpot logging to confirm that
org.eclipse.core.internal.dtree.DataTreeNode.forwardDeltaWith is being compiled
in your runs?  If it is being compiled then either the upstream bug has been
fixed (but the bug report left open for some reason) or the failure is
intermittent or dependent on hardware specifics.

If the bug has been fixed upstream then it would be nice to know what the fix was.

Comment 6 Bug Zapper 2008-05-14 10:24:07 UTC
Changing version to '9' as part of upcoming Fedora 9 GA.
More information and reason for this action is here:
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

Comment 7 Red Hat Bugzilla 2008-07-25 01:04:24 UTC
Adding patrickm to the cc list as the manager of the disabled user fitzsim who reported this bug

Comment 8 Ricardo Veguilla 2008-08-12 17:07:12 UTC
I had the same problem and filed a bug in the icedtea bugzilla. See: 

http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=180  

According to the comments, the bug was fixed after openjdk b010. I personally tested openJDK 1.6.0_10-rc-b28 and it working correctly.

Comment 9 Andrew Overholt 2008-08-12 17:15:57 UTC
Note that we're still carrying the workaround in %{_libdir}/eclipse/eclipse.ini for a few of these crashes:

-XX:CompileCommand=exclude,org/eclipse/core/internal/dtree/DataTreeNode,forwardDeltaWith
-XX:CompileCommand=exclude,org/eclipse/jdt/internal/compiler/lookup/ParameterizedMethodBinding,<init>
-XX:CompileCommand=exclude,org/eclipse/cdt/internal/core/dom/parser/cpp/semantics/CPPTemplates,instantiateTemplate
-XX:CompileCommand=exclude,org/eclipse/cdt/internal/core/pdom/dom/cpp/PDOMCPPLinkage,addBinding
-XX:CompileCommand=exclude,org/eclipse/osgi/framework/internal/core/FilterImpl$Parser,parse_filter

That last one I just hit this morning but the syntax on it is incorrect for the inner class; anyone know how to fix it?

Comment 10 Andrew Haley 2008-08-21 13:11:27 UTC
This is now fixed.  Closing.

Comment 11 Andrew Overholt 2008-08-21 13:17:27 UTC
So we should remove the CompileCommand=exclude for the DataTreeNode class?  What about the others?