Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-5041802

JDWP agent dumps core running with 32-bit VM on solaris-sparcv9

XMLWordPrintable

    • sparc
    • solaris_8

      Name: ipR10196 Date: 05/04/2004


      JDK solaris-sparcv9 bundle provides both 32-bit and 64-bit VM
      (32-bit VM is launched by default), both VM are running well in b49:

        $ .../jdk1.5.0-b49/solaris-sparcv9/bin/java -d64 -version
        java version "1.5.0-beta2"
        Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta2-b49)
        Java HotSpot(TM) 64-Bit Server VM (build 1.5.0-beta2-b49, mixed mode)
       
        $ .../jdk1.5.0-b49/solaris-sparcv9/bin/java -version
        java version "1.5.0-beta2"
        Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta2-b49)
        Java HotSpot(TM) Client VM (build 1.5.0-beta2-b49, mixed mode)

      JDWP agent runs fine with 64-bit VM, but dumps core running
      with 32-bit VM:

        $ .../jdk1.5.0-b49/solaris-sparcv9/bin/java -d64 -agentlib:jdwp=transport=dt_socket,server=y,suspend=n -version
        Listening for transport dt_socket at address: 65236
        java version "1.5.0-beta2"
        Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta2-b49)
        Java HotSpot(TM) 64-Bit Server VM (build 1.5.0-beta2-b49, mixed mode)

        $ .../jdk1.5.0-b49/solaris-sparcv9/bin/java -agentlib:jdwp=transport=dt_socket,server=y,suspend=n -version
        #
        # An unexpected error has been detected by HotSpot Virtual Machine:
        #
        # SIGSEGV (0xb) at pc=0xdbf1b6ec, pid=1580, tid=1
        #
        # Java VM: Java HotSpot(TM) Client VM (1.5.0-beta2-b49 mixed mode, sharing)
        # Problematic frame:
        #
        [error occurred during error reporting, step 60, id 0xb]

        # An error report file with more information is saved as hs_err_pid1580.log
        #
        # If you would like to submit a bug report, please visit:
        # http://java.sun.com/webapps/bugreport/crash.jsp
        #

      This crash is reproduced for any JDWP agent mode (i.e, server=y|n,suspend=y|n).

      Running JDWP agent with 32-bit VM from solaris-sparc bundle
      shows no crash:

        $ .../jdk1.5.0-b49/solaris-sparc/bin/java -agentlib:jdwp=transport=dt_socket,server=y,suspend=n -version
        Listening for transport dt_socket at address: 65243
        java version "1.5.0-beta2"
        Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta2-b49)
        Java HotSpot(TM) Client VM (build 1.5.0-beta2-b49, mixed mode)

      With the previous JDK build b48 this crash also does not occur,
      though error messages are printed in some modes of JDWP agent:

        $ .../jdk1.5.0-b48/solaris-sparcv9/bin/java -agentlib:jdwp=transport=dt_socket,server=y,suspend=y -version
        Listening for transport dt_socket at address: 65248
        <^C does not kill this>

        $ .../jdk1.5.0-b48/solaris-sparcv9/bin/java -agentlib:jdwp=transport=dt_socket,server=y,suspend=n -version
        Listening for transport dt_socket at address: 65244
        java version "1.5.0-beta2"
        Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta2-b48)
        Java HotSpot(TM) Client VM (build 1.5.0-beta2-b48, mixed mode)
        JDWP exit error JVMTI_ERROR_INVALID_MONITOR(50): on raw monitor enter
        FATAL ERROR in native method:
        JDWP on raw monitor enter, jvmtiError=JVMTI_ERROR_INVALID_MONITOR(50)

      Generated hs_err_pid1580.log includes empty stack trace,
      but dbx shows that crash occurs in Agent_OnLoad (see Comments).

      This bug affects the following JDI tests of testbase_vm:

        nsk/jdi/VirtualMachineManager/createVirtualMachine/createVM002
        nsk/jdi/VirtualMachineManager/createVirtualMachine/createVM003
        nsk/jdi/VirtualMachineManager/createVirtualMachine/createVM004
        nsk/jdi/VirtualMachineManager/createVirtualMachine/createVM005

      These tests ignore Java options and launch debuggee VM in default
      32-bit mode on solaris-sparcv9 platformm, that leads to crash.
      There is RFE for these tests to not to ignore Java options:

        4943437 TEST_RFE: createVM002 does not honor DEBUGEE_VM_* setting

      ======================================================================

      Name: ipR10196 Date: 05/04/2004


      The same crash occur just running JDB from solaris-sparcv9 bundle
      with any application, e.g., HelloWorld.java:

      $ .../solaris-sparcv9/bin/jdb HelloWorld
      Initializing jdb ...
      > run
      run HelloLoop
      VM start exception: VM initialization failed for: .../solaris-sparcv9/jre/bin/java -Xdebug
      -Xrunjdwp:transport=dt_socket,address=novo153:65431,suspend=y HelloWorld

      #
      # An unexpected error has been detected by HotSpot Virtual Machine:
      #
      # SIGSEGV (0xb) at pc=0xdbf1b6ec, pid=1741, tid=1
      #
      # Java VM: Java HotSpot(TM) Client VM (1.5.0-beta2-b49 mixed mode, sharing)
      # Problematic frame:
      #
      [error occurred during error reporting, step 60, id 0xb]

      # An error report file with more information is saved as hs_err_pid1741.log
      #
      # If you would like to submit a bug report, please visit:
      # http://java.sun.com/webapps/bugreport/crash.jsp
      #

      Fatal error:
      Target VM failed to initialize.


      ======================================================================

            ohair Kelly Ohair (Inactive)
            popovsunw Popov Popov (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            0 Start watching this issue

              Created:
              Updated:
              Resolved:
              Imported:
              Indexed: