JDK-6614100 : EXCEPTION_ACCESS_VIOLATION while running Eclipse with 1.6.0_05-ea
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 6u5,6u7-rev,7
  • Priority: P3
  • Status: Resolved
  • Resolution: Fixed
  • OS: generic,linux,solaris_10,windows_xp
  • CPU: generic,x86,sparc
  • Submitted: 2007-10-08
  • Updated: 2011-02-16
  • Resolved: 2008-06-12
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 6 JDK 7 Other
6u10Fixed 7Fixed OpenJDK6,hs10Fixed
Related Reports
Duplicate :  
Duplicate :  
Duplicate :  
Relates :  
Relates :  
Relates :  
Relates :  
Description
FULL PRODUCT VERSION :
C:\jdk1.6.0_05\bin>java -version
java version "1.6.0_05-ea"
Java(TM) SE Runtime Environment (build 1.6.0_05-ea-b04)
Java HotSpot(TM) Client VM (build 1.6.0_05-ea-b04, mixed mode, sharing)

C:\jdk1.6.0_05\bin>java -server -version
java version "1.6.0_05-ea"
Java(TM) SE Runtime Environment (build 1.6.0_05-ea-b04)
Java HotSpot(TM) Tiered VM (build 1.6.0_05-ea-b04, mixed mode)

FULL OS VERSION :
Microsoft Windows XP [Version 5.1.2600]

A DESCRIPTION OF THE PROBLEM :
While running Eclipse with 1.6.0_05-ea, the following
First reported in http://forums.java.net/jive/thread.jspa?threadID=31465


I first hit this error on my 32-bit Windows XP machine at work.
However, I will not have time while at work to investigate further.

Luckily, in doing some more testing at home on my Windows XP 64-bit machine I was able to trigger the exact same bug with the 64-bit server VM running Eclipse-3.4-M2 for Windows x86-64.
I am attaching that new error log for completeness.

The Eclipse code in question is available here: http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.core.resources/src/org/eclipse/core/internal/dtree/DataTreeNode.java?revision=1.20&view=markup

As of yet, I have been unable to produce a standalone test case... still, in reviewing DataTreeNode.forwardDeltaWith, I would hope it might be more evident to you how that code (with all the looping, array allocation and System.arrayCopy calls) might stress the hotspot server compiler.

FYI, I have been running Eclipse (misc. versions) with 1.6.0_02-b05 since it was released, so this is *definitely* a recent regression introduced in 1.6.0_05-ea.


THE PROBLEM WAS REPRODUCIBLE WITH -Xint FLAG: Did not try

THE PROBLEM WAS REPRODUCIBLE WITH -server FLAG: Did not try

ERROR MESSAGES/STACK TRACES THAT OCCUR :
Attached seperatly


REPRODUCIBILITY :
This bug can be reproduced occasionally.

Comments
EVALUATION http://hg.openjdk.java.net/jdk7/hotspot-comp/hotspot/rev/65fe2bd88839
10-06-2008

EVALUATION The assertion failures are very similar to 6659207 but it is in fact a different bug. I've reproduced it and narrowed it down the change in jdk 7 b13. 6552204 attempts to undo useless Phi of AddP but in the process it seems to expose a bug in stretch_base_pointer_live_ranges that creates a phi that merges a real value with an ideal ConP of NULL. Bad things happen since the ideal node doesn't actually produce a value and doesn't have an LRG. The simplest fix for this is to undo 6552204 and do a real fix for stretch_base_pointer_live_ranges in a later putback. Addressing that will be somewhat complicated but thankfully the bug there has never triggered until the change for 6552204.
23-05-2008

EVALUATION This is likely a duplicate of 6659207 given all the symptoms. It's possible that it wasn't showing up prior to u5 because other optimizations added in u5 exposed the latent bug for your particular code sequence. It's a fairly picky bug as far as code shape goes. given all the symptoms. It's possible that it wasn't showing up prior to u5 because other optimizations added in u5 exposed the latent bug for your particular code sequence. It's a fairly picky bug as far as code shape goes.
07-03-2008