(Preliminary) RFC 7038914: VM could throw uncaught OOME in ReferenceHandler thread

[Peter Levart]

On 05/04/2013 07:38 PM, Vitaly Davidovich wrote:
>
> Oops, that was my mistake - I thought the lock here was a j.u.c.Lock 
> which of course doesn't even make sense given we're talking about 
> ObjectMonitor.  So disregard that bit.
>
> Ignoring OOM and continuing just seems very fragile as it means you 
> somehow know that all state is still consistent.  Most java code isn't 
> async exception safe, so it's hard to reason about state after OOM.  
> Maybe Reference Handler is OK in that regard though.
>

I think it is safe to ignore OOME here, since this is the only place 
where heap allocation happens and it is known what provokes it. 
Otherwise ReferenceHandler just shuffles existing pointers in existing 
objects...

Regards, Peter

> Cheers
>
> Thanks
>
> Sent from my phone
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20130504/8dad0b4c/attachment.htm>