On 08/20/2014 06:43 PM, Andrew Haley wrote: > On 08/20/2014 04:10 PM, Florian Weimer wrote: >> Is there already a way to compute the expression in the subject without >> the ByteBuffer allocation? I saw quite a few equivalent formulations in >> the OpenJDK sources, and perhaps it's time to add a standardized >> solution. > > Isn't this really calling for intrinsification of ByteBuffers? With that, > C2 would know that the ByteBuffer doesn't escape, and could collapse the > lot into a load.
It's a lot of heavy machinery for something that should be quite simple. I also expect quite a few users during VM startup. -- Florian Weimer / Red Hat Product Security
