Request for review (S): 8008382: Remove redundant use of Atomic::add(jlong, jlong *) in create_new_gc_id()

[Bengt Rutisson]

Hi again,

Erik Helin pointed out a potential race in CMS where the concurrent 
cycle may be trying to increment the GC id in parallel with a young 
collection. This is scary for other reasons as well but removing the 
atomic call exposes it. We concluded that the safe change is to move the 
call to register_gc_start() in to the CMS VM operation. This way we get 
the VM operation synchronization and can be sure that two threads are 
never incrementing the GC id in parallel.

Updated webrev here:
http://cr.openjdk.java.net/~brutisso/8008382/webrev.01/

Bengt

On 2/18/13 2:21 PM, Bengt Rutisson wrote:
>
> Hi all,
>
> Could I have a couple of reviews for this change?
> http://cr.openjdk.java.net/~brutisso/8008382/webrev.00/
>
> There is no need to use atomics in create_new_gc_id() since it is not 
> called by multiple threads in parallel. Also, Atomic::add(jlong, jlong 
> *) is broken for ARM.
>
> We should remove the use of Atomic::add in create_new_gc_id() for now. 
> If we need this to be called by multiple threads in the future we have 
> to reconsider how this should be done. Either adding a lock or doing 
> some kind of 32 bit atomic work.
>
> Thanks,
> Bengt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20130219/1f35773d/attachment.htm>