hs25 review request (2nd round): 8008511 JSR 292: MemberName vmtarget refs to methods must be updated at class redefinition
David Holmes
david.holmes at oracle.com
Mon Apr 15 16:54:26 PDT 2013
On 16/04/2013 9:48 AM, David Holmes wrote:
> On 16/04/2013 3:59 AM, serguei.spitsyn at oracle.com wrote:
>> On 4/14/13 10:12 PM, David Holmes wrote:
>>> Not a full review ...
>>>
>>> In instanceKlass.cpp:
>>>
>>> + MemberNameTable* mnt = member_names();
>>> + if (mnt != NULL) {
>>> + delete mnt;
>>> + }
>>>
>>> a) do we need to call set_member_names(null) ?
>>
>> Fixed, thanks.
>>
>>
>>> b) do we need to use the MemberNameTable_lock to guard this?
>>
>> I don't think so.
>> This deallocation must be done at a safepoint.
>> We can't grab locks at safepoints.
>
> Okay but that is only one side of things. We also have to verify that
> any code that does grab the lock can't enter a safepoint while holding
> the lock.
That was an over generalization. Presumably we would never be in a
position of doing the delete if the table could actually still be in use.
David
> David
> ------
More information about the serviceability-dev
mailing list