Review Request (S) 8024345: 'assert(_value != NULL) failed: resolving NULL _value' from VM_RedefineClasses::set_new_constant_pool

[serguei.spitsyn at oracle.com]

On 9/11/13 9:04 PM, David Holmes wrote:
> Hi Serguei,
>
> Typo:
>
> 2489   // scratch_cp is a merged constant po/ ol

Fixed, thanks!

>
>
> In VM_RedefineClasses::set_new_constant_pool do you not need to 
> de-allocate smaller_cp if you return due to an exception here:
>
> 2504   scratch_cp->copy_cp_to(1, scratch_cp_length - 1, smaller_cp, 1, 
> CHECK);

Good catch.
It is better to deallocate it even if it is not very useful.
The only exception possible here is an OOME (Metadata shortage).


Thanks,
Serguei

>
> David
> -----
>
> On 12/09/2013 8:02 AM, serguei.spitsyn at oracle.com wrote:
>> Please, review the fix for:
>>    bug: http://bugs.sun.com/view_bug.do?bug_id=8024345
>>    jbs: https://bugs.openjdk.java.net/browse/JDK-8024345
>>
>>
>> Open webrev:
>> http://cr.openjdk.java.net/~sspitsyn/webrevs/2013/hotspot/8024345-JVMTI-MEM.1 
>>
>>
>>
>> Summary:
>>    The OOME's in the JVMTI merge_cp_and_rewrite and
>> set_new_constant_pool must be handled correctly.
>>
>>
>> Testing:
>>    UTE tests - in progress:  vm.quick-pcl.testlist with limited
>> Metaspace memory,
>>                                          nsk.jvmti.testlist,
>> nsk.jdi.testlist,
>>                                          Jtreg java/lang/instrument
>>
>> Thanks,
>> Serguei