RFR: 8220682: runtime/ErrorHandling/TestHeapDumpOnOutOfMemoryErrorInMetaspace.java fails with JDK-8214712

[Jiangli Zhou]

+1

Thanks,
Jiangli

On Fri, Mar 15, 2019 at 7:56 AM Jean Christophe Beyler <jcbeyler at google.com>
wrote:

> Hi Claes,
>
> Looks good to me. For those who would want to see the patch change done in
> the serviceability-agent:
> http://hg.openjdk.java.net/jdk/jdk/rev/91f06b86c0da
>
> Thanks,
> Jc
>
> On Fri, Mar 15, 2019 at 5:10 AM Claes Redestad <claes.redestad at oracle.com>
> wrote:
>
>> Hi,
>>
>> to allow for archiving objects in CDS that might not be loaded (e.g.,
>> JDK-8214712), we need to ignore objects at heap dump time that haven't
>> actually been loaded. This patch fixes the VM heapDumper implementation,
>> and is a mirror of a fix already done in the serviceability-agent heap
>> dumper[1].
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8220682
>> Patch:
>> diff -r 082cd1e3c441 src/hotspot/share/services/heapDumper.cpp
>> --- a/src/hotspot/share/services/heapDumper.cpp Mon Dec 03 16:25:27 2018
>> +0100
>> +++ b/src/hotspot/share/services/heapDumper.cpp Fri Mar 15 13:11:28 2019
>> +0100
>> @@ -958,6 +958,11 @@
>>   // creates HPROF_GC_INSTANCE_DUMP record for the given object
>>   void DumperSupport::dump_instance(DumpWriter* writer, oop o) {
>>     Klass* k = o->klass();
>> +  if (k->java_mirror() == NULL) {
>> +    // Ignoring this object since the corresponding java mirror is not
>> loaded.
>> +    // Might be a dormant archive object.
>> +    return;
>> +  }
>>
>>     writer->write_u1(HPROF_GC_INSTANCE_DUMP);
>>     writer->write_objectID(o);
>>
>>
>> Thanks!
>>
>> /Claes
>>
>> [1] https://bugs.openjdk.java.net/browse/JDK-8214756
>>
>
>
> --
>
> Thanks,
> Jc
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20190315/88e61b79/attachment.html>