4 issues for HSDB

[Yasumasa Suenaga]

PING: Have you checked these issues?

I want to know I can file them to JBS and can send webrev.


Yasumasa


On 2017/08/31 23:49, Yasumasa Suenaga wrote:
> Hi all,
> 
> I attached HSDB to jdk.jshell/jdk.internal.jshell.tool.JShellToolProvider . However, I found 4 issues.
> I propose patches for them, but they are not yet filed to JBS. I want to file and to send review request after this discussion.
> 
> 
> 1. Java Stack cannot be shown.
> 
>    - Suggested fix
>      http://cr.openjdk.java.net/~ysuenaga/hsdb-fix/conflict-type/
> 
>    - How to reproduce
>      Choose "main" thread and click "Show Java stack trace" icon on "Java Threads" toolbar.
> 
>    - Stack trace
> ```
> Exception in thread "AWT-EventQueue-0" sun.jvm.hotspot.utilities.AssertionFailure: type check
>      at jdk.hotspot.agent/sun.jvm.hotspot.utilities.Assert.that(Assert.java:32)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.StackValue.getObject(StackValue.java:59)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.StackValueCollection.oopHandleAt(StackValueCollection.java:51)
>      at jdk.hotspot.agent/sun.jvm.hotspot.ui.classbrowser.HTMLGenerator.genHTMLForJavaStackTrace(HTMLGenerator.java:1931)
>      at jdk.hotspot.agent/sun.jvm.hotspot.ui.JavaStackTracePanel.setJavaThread(JavaStackTracePanel.java:78)
>      at jdk.hotspot.agent/sun.jvm.hotspot.HSDB.showJavaStackTrace(HSDB.java:1526)
> ```
> 
>    I guess it is caused by inlined method which is generated by JIT compiler.
>    StackValue has T_CONFLICT value. It might be able to skip.
>    I'm not sure about this. So I need comments.
> 
> 
> 2. ArrayIndexOutOfBoundsException is occurred on Stack Memory window
> 
>    - Suggested fix
>      http://cr.openjdk.java.net/~ysuenaga/hsdb-fix/indy/
> 
>    - How to reproduce
>      Choose "main" thread and click "Stack Memory" icon on "Java Threads" toolbar.
> 
>    - Stack trace
> ```
> Error while performing oopsDo for frame sp: 0x00007f80db89e590, unextendedSP: 0x00007f80db89e598, fp: 0x00007f80db89e5e8, pc: 0x00007f80bb7abf00
> java.lang.ArrayIndexOutOfBoundsException: 384 57
>      at jdk.hotspot.agent/sun.jvm.hotspot.utilities.GenericArray.getIntegerAt(GenericArray.java:82)
>      at jdk.hotspot.agent/sun.jvm.hotspot.utilities.U2Array.at(U2Array.java:59)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ConstantPool.invokedynamicCPCacheIndex(ConstantPool.java:278)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ConstantPool.invokedynamicCPCacheEntryAt(ConstantPool.java:283)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ConstantPool.implNameAndTypeRefIndexAt(ConstantPool.java:292)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ConstantPool.implGetSignatureRefAt(ConstantPool.java:264)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ConstantPool.getSignatureRefAt(ConstantPool.java:256)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.doMethod(GenerateOopMap.java:1731)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.interp1(GenerateOopMap.java:1386)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.interpBB(GenerateOopMap.java:803)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.interpAll(GenerateOopMap.java:1109)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.doInterpretation(GenerateOopMap.java:982)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.GenerateOopMap.computeMap(GenerateOopMap.java:2199)
>      at jdk.hotspot.agent/sun.jvm.hotspot.interpreter.OopMapForCacheEntry.computeMap(OopMapForCacheEntry.java:80)
>      at jdk.hotspot.agent/sun.jvm.hotspot.interpreter.OopMapCacheEntry.fill(OopMapCacheEntry.java:53)
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.Method.getMaskFor(Method.java:261)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.Frame.oopsInterpretedDo(Frame.java:588)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.Frame.oopsDo(Frame.java:440)
>      at jdk.hotspot.agent/sun.jvm.hotspot.HSDB$34.run(HSDB.java:1038)
> ```
>    It seems to relate to JDK-8175816.
>    I guess SA handles incorrect data for ConstantPool for indy. It should be handled like invokedynamic_cp_cache_index().
> 
> 
> 3. UnknownOopException is occurred on Stack Memory window
> 
>    - Suggested fix
>      http://cr.openjdk.java.net/~ysuenaga/hsdb-fix/oopmap/
> 
>    - How to reproduce
>      Choose "output reader" thread and click "Stack Memory" icon on "Java Threads" toolbar.
> 
>    - Stack trace
> ```
> Error while performing oopsDo for frame sp: 0x00007f80986cb630, unextendedSP: 0x00007f80986cb630, fp: null, pc: 0x00007f80c32edfeb
> sun.jvm.hotspot.oops.UnknownOopException
>      at jdk.hotspot.agent/sun.jvm.hotspot.oops.ObjectHeap.newOop(ObjectHeap.java:264)
>      at jdk.hotspot.agent/sun.jvm.hotspot.HSDB$34$1.addAnnotation(HSDB.java:1098)
>      at jdk.hotspot.agent/sun.jvm.hotspot.HSDB$34$1.visitAddress(HSDB.java:1045)
>      at jdk.hotspot.agent/sun.jvm.hotspot.compiler.ImmutableOopMapSet$MyVisitor.visitOopLocation(ImmutableOopMapSet.java:58)
>      at jdk.hotspot.agent/sun.jvm.hotspot.compiler.ImmutableOopMapSet.allDo(ImmutableOopMapSet.java:230)
>      at jdk.hotspot.agent/sun.jvm.hotspot.compiler.ImmutableOopMapSet.oopsDo(ImmutableOopMapSet.java:179)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.Frame.oopsCodeBlobDo(Frame.java:618)
>      at jdk.hotspot.agent/sun.jvm.hotspot.runtime.Frame.oopsDo(Frame.java:444)
>      at jdk.hotspot.agent/sun.jvm.hotspot.HSDB$34.run(HSDB.java:1038)
>      at jdk.hotspot.agent/sun.jvm.hotspot.utilities.WorkerThread$MainLoop.run(WorkerThread.java:70)
>      at java.base/java.lang.Thread.run(Thread.java:844)
> ```
> 
>    According to frame::oopmapreg_to_location(), Frame#oopMapRegToLocation() should be calculated with "VMRegImpl::stack_slot_size". Not address size.
>    I saw similar exception and call stack at reverse pointer calculation.
> 
> 
> 4. [Unknown generation] is shown in Stack Memory for output reader thread
> 
>    - Suggested fix
>      http://cr.openjdk.java.net/~ysuenaga/hsdb-fix/stackmemory-g1/
> 
>    - How to reproduce
>      Choose "output reader" thread and click "Stack Memory" icon on "Java Threads" toolbar.
> 
>    HSDB.java is not handles G1CollectedHeap. So it is not detect the generation of oop.
> 
> 
> Thanks,
> 
> Yasumasa