Review request: 8029106: JVM crashes in Metachunk::Metachunk during parallel class redefinition (PrivateMLetController, anonymous-simple_copy_1)

[Stefan Karlsson]

On 2013-12-09 17:11, Coleen Phillimore wrote:
> On 12/9/2013 4:23 AM, Stefan Karlsson wrote:
>> http://cr.openjdk.java.net/~stefank/8029106
>>
>> Please, review this bug fix for:
>>
>> 8029106: JVM crashes in Metachunk::Metachunk during parallel class 
>> redefinition (PrivateMLetController, anonymous-simple_copy_1)
>> Summary: Fixed overflow bug in VirtualSpaceNode::is_available
>>
>> thanks,
>> StefanK
>
> The bug fix looks good.  When do these tests run?

Your run them by adding -XX:+ExecuteInternalVMTests to the command line 
of a debug build. JPRT exercises these tests.

> Do they allocate 3 new virtualspace nodes and leave them around until 
> exit?

They allocate VirtualSpaceNodes but never link them to the virtual space 
lists. The memory is released in the VirtualSpaceNode destructor, which 
is called when vsn goes out of scope.

>   Is this another leak if the JVM is embedded?  What happens if there 
> isn't enough memory to allocate these chunks?

It's very unlikely to happen, but if it does we'll hit this assert:
     assert(vsn.initialize(), "Failed to setup VirtualSpaceNode");

thanks,
StefanK

>
> thanks,
> Coleen