Name: pvR10198 Date: 09/04/2003


---------------------------------------------------
Test : nsk/jvmti/FramePop/framepop001
            : nsk/jvmti/GetStackTrace/getstacktr005
VM : ClientVM, ServerVM
Mode : -Xcomp
Platform : x86
OS : SunOS 5.8, RedHat Linux 8.0, Windows 2000
JDK : 1.5.0-b15 promoted (JVMTI v71)
              1.5.0-b15 w/ Hotspot build 20030820101708.rfield.hotspot3-product (JVMTI v0.2.90)
---------------------------------------------------

The framepop001 test exercises JVMTI event callback function FramePop.
The test starts a new thread, stops it at breakpoint, requests to
generate FramePop events for current and the next stack frames, and
verifies if parameters of received FramePop events contain expected values.
Also the test checks that the location reported by GetFrameLocation
indentifies the executable location in the returning method, immediately
prior to the return.

The getstacktr005 test exercises JVMTI function GetStackTrace via breakpoint
and consequent step. The test starts a new thread, does some nested calls,
stops the thread at breakpoint, does single step and checks if the number
of frames in the thread's stack is as expected and the function returns all
the expected frames.

The framepop001 test fails to get FramePop event from the second stack
frame of the child thread, but gets FramePop event from a stack frame
of the main thread.

The getstacktr005 test steps into the main thread right after breakpoint.

The tests pass in interpreted and mixed modes on x86 and all modes/solsparc.
The tests pass with jdk 1.5.0-b14.

To reproduce the framepop001 test failure run doit1.sh script in
GammaBase/Bugs/<this bug number> directory:

    sh doit1.sh <JAVA_HOME> -Xcomp

To reproduce the getstacktr005 test failure run doit2.sh script in
GammaBase/Bugs/<this bug number> directory:

    sh doit2.sh <JAVA_HOME> -Xcomp

Running in verbose mode framepop001 outputs:

$ sh doit1.sh /export/ld24/java/hotspot/jdk1.5/solx86 -v -Xcomp
framepop001.c:
JVMTITools.c:
java version "1.5.0-beta"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta-b15)
Java HotSpot(TM) Client VM (build 1.5.0-beta-b15, compiled mode)

>>> retrieving frame pop info ...
>>> class: "Lnsk/jvmti/FramePop/framepop001;"
>>> method: "chain()V"
>>> location: 0x000000000
>>> ... done
Wrong number of frame pop events: 1, expected: 2
>>> retrieving frame pop info ...
>>> class: "Lnsk/jvmti/FramePop/framepop001;"
>>> method: "run([Ljava/lang/String;Ljava/io/PrintStream;)I"
>>> location: 0x000000003
>>> ... done
(pop#1) wrong class: "Lnsk/jvmti/FramePop/framepop001;", expected: "Lnsk/jvmti/FramePop/framepop001a;"
(pop#1) wrong method name: "run", expected: "dummy"
(pop#1) wrong method sig: "([Ljava/lang/String;Ljava/io/PrintStream;)I", expected: "()V"
Exit Code: 97

Running getstacktr005 outputs:

$ sh doit2.sh /export/ld24/java/hotspot/jdk1.5/solx86 -Xcomp
framepop001.c:
JVMTITools.c:
java version "1.5.0-beta"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-beta-b15)
Java HotSpot(TM) Server VM (build 1.5.0-beta-b15, compiled mode)

Wrong frame count, expected: 5, actual: 1
(frame#0) wrong class sig: "Ljava/lang/Thread;", expected: "Lnsk/jvmti/GetStackTrace/getstacktr005$TestThread;"
(frame#0) wrong method name: "exit", expected: "chain4"
Exit Code: 97

This looks like the same problem as 4915625:
  TTY: second 'step' command runs incorrectly in "-server -Xcomp" mode on Intel

This bug also affects the following tests:

    nsk/jvmti/GetStackTrace/getstacktr008
    nsk/jvmti/RedefineClasses/redefclass019
    nsk/jvmdi/events/framepop001
    nsk/jvmdi/GetCallerFrame/callframe005
    nsk/jvmdi/GetCallerFrame/callframe008
    nsk/jvmdi/RedefineClasses/redefclass019

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