RFR: 8024176: [macosx] gc/metaspace/ClassMetaspaceSizeInJmapHeap.java failed since jdk8b105, hs25b47

[Stefan Johansson]

Thanks Staffan and Jon for reviewing this.

Cheers,
Stefan

On 2013-09-10 18:38, Jon Masamitsu wrote:
> Change looks good.
>
> Jon
>
> On 9/10/13 6:19 AM, Stefan Johansson wrote:
>> Hi all,
>>
>> Please review this really small fix for a test failure on Mac.
>>
>> Webrev:
>> http://cr.openjdk.java.net/~sjohanss/8024176/webrev.00/
>>
>> Summary:
>> A GC jtreg test fails on Mac, and when looking at the failure one can 
>> see that the SA fails to find the type for an oop. It turns out the 
>> code for reading compressed klass pointers in the sa-agent on Mac 
>> used readCompOopAddress instead of readCompKlassAddress, this is 
>> wrong but has been hidden since compressed oops and compressed 
>> klasses has used the same base address. This was recently changed and 
>> now the SA agent will show this issue when attached to a JVM using 
>> compressed klasses.
>>
>> The patch:
>>
>> --- 
>> old/agent/src/share/classes/sun/jvm/hotspot/debugger/bsd/BsdAddress.java 
>> 2013-09-10 13:52:17.163078372 +0200
>> +++ 
>> new/agent/src/share/classes/sun/jvm/hotspot/debugger/bsd/BsdAddress.java 
>> 2013-09-10 13:52:17.079078971 +0200
>> @@ -81,7 +81,7 @@
>>
>>      public Address getCompKlassAddressAt(long offset)
>>              throws UnalignedAddressException, 
>> UnmappedAddressException {
>> -        return debugger.readCompOopAddress(addr + offset);
>> +        return debugger.readCompKlassAddress(addr + offset);
>>      }
>>
>>      //
>>
>>
>> Thanks,
>> Stefan
>