JDK-4288475 : assert(test != 0, "Unexpected object header as 0"); -Xcomp
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 2.0
  • Priority: P5
  • Status: Resolved
  • Resolution: Fixed
  • OS: windows_nt
  • CPU: x86
  • Submitted: 1999-11-05
  • Updated: 2000-02-17
  • Resolved: 2000-02-17
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.
Other
2.0 fcsFixed
Related Reports
Relates :  
Relates :  
Relates :  
Description
HS 2.0-rc1B intermittently brings the whole WinNT system to crash down while 
executing the JCK-stress test 'jck12a017'. However, sometimes crash is not so 
severe -- HS sometimes crashes by exactly the same way as is described in the 
bug:
    4288182 (P4/S5) RC1 : Build B : 2 JCK VM tests crash with assert in os_win32.cpp. 729

I've executed the test 'jck12a017' manually 9 times. Twice, I've observed WinNT 
"blue screens of death", displaying something like: 
    *** STOP: 0x0000001E (0xC0000005,0x801399CA,0x00000001,0x0EAF8437)
    KMODE_EXCEPTION_NOT_HANDLED*** Address 801399ca has base at 80100000 - ntoskrnl.exe
    . . . 
And once the following crash message appeared:
    #
    # HotSpot Virtual Machine Error, EXCEPTION_ACCESS_VIOLATION
    #
    # Error ID: 4F533F57494E13120E43505002D9
    #
(i.e.: ErrorID = "os_win32.cpp, 729") 

Note, that the test 'jck12a017' involves JCK 1.2a tests from API-tests sub-suite, 
not VM-tests. So, the problem described in the bug #4288182 affects not only VM 
tests from JCK testbase. 

The stress test 'jck12a017' belongs to the 'testbase_nsk' testbase, and its 
sources could be found in the directory: 
    /net/sqesvr/vsn/testbase/testbase_nsk/src/nsk/stress/jck12a/jck12a017

To execute this test follow these instructions: 

    1. Create some temporary directory, to say C:\TEMP\jck12a017, 
       and make this directory your current directory:
           cd C:\TEMP\jck12a017

    2. Copy the test's sources into this directory. You need to copy 
       only the file 'jck12a017.java', and the files 'jck12a017.cfg' 
       and 'jck12a017.README' are optional.

    3. Also copy into the current directory the stress-wrapper needed to 
       execute the test. The wrapper is the 'StressTest.java' file found 
       in the directory:
           /net/sqesvr/vsn/testbase/testbase_nsk/src/stress/share

    4. Create 'classes' subdirectory for the wrapper's class-files you will 
       need to compile:
           mkdir classes

    5. Mount to the drive L: the NFS directory:
        /usr/local/java
       (it is available as \\grinder\local-java for NT machines)

    6. Setup the CLASSPATH to provide the test with JavaTest 2.0 tool 
       and JCK 1.2a pre-compiled API tests:
           set CLASSPATH=.;.\classes;L:\sqe-tools2.0\javatest.jar;L:\jck1.2\JCK-runtime-api-12a\classes

    7. Compile the wrapper with JDK 1.2, JDK 1.2.2 or JDK 1.3 compiler:
           javac -d .\classes StressTest.java

    8. Compile the test with JDK 1.2, JDK 1.2.2 or JDK 1.3 compiler:
           javac jck12a017.java

    9. Execute the test several times to reproduce the failure:
           java jck12a017 2>log

OOPS! -- I've fogotten to mention, that the problem only arises in -Xcomp mode! 
So, please execute the test with -Xcomp keyword:
    java -Xcomp jck12a017

Comments
CONVERTED DATA BugTraq+ Release Management Values COMMIT TO FIX: generic FIXED IN: 2.0_fcs INTEGRATED IN: 2.0_fcs
14-06-2004

EVALUATION michael.paleczny@eng 1999-11-07 Tested using classic, interpreted, compiled classic: finished several graphics tests then threw OutOfMemoryError interp: finished full suite, but failed ten tests, several NullPointerException compile: finished full suite, threw many NullPointerExceptions Did not observe blue screen during 4 evaluation runs. michael.paleczny@eng 2000-02-08 Crashed with current release build, not blue-screen. Debug build hit assert. christopher.vick@eng 2000-02-08 With JDK1.2.2, this hits a bad pointer in the loop code at method swing.LookAndFeel.getsetTests::LookAndFeel2008. With JDK1.3 this produces inconsistant crashes in vtable lookup or monitor code. Handing over to Cliff to fix the loop problems. ====== All obvious bugs now fixed. App hangs with -Xint, -Xcomp, -Xmixed and classic after running awhile. Throws out of memory exceptions on classic occasionally. Consumes a lot of memory while running (80Meg+). Appears to be the correct behavior of a buggy app. In mixed mode I inspected the state of a few hundred java threads. 1 thread was stuck in the AWT event loop native code, all the rest in Object::wait (except for the scattered VM threads stuck in their usual wait locations). cliff.click@eng 2000-02-16
16-02-2000