RFR (L): 7054512: Compress class pointers after perm gen removal (original) (raw)

Coleen Phillimore coleen.phillimore at oracle.com
Mon Sep 24 07:05:50 PDT 2012

Hi Vladimr,

I can't seem to find this code that I'm talking about below. I was looking for it on Friday.. I thought we added the class_loader oop to the oop section of the nmethod. Maybe Tom implemented it differently or we didn't need this to keep metadata alive.

Coleen

On 9/20/2012 3:49 PM, Vladimir Kozlov wrote:

Thanks, Coleen

Vladimir Coleen Phillimore wrote:

There's code in nmethod.cpp that iterates over these pointers and makes sure that the classloader associated with this metadata is added to the oop section of the nmethod. This is so the class loader isn't unloaded and metadata deallocated. I think that's the main reason we save it when we're generating code in the oopRelocation area. There might be other reasons but I don't know offhand. Coleen On 9/20/2012 3:27 PM, Vladimir Kozlov wrote: This look good.

Coleen, why we need relocation for metadata pointers? Thanks, Vladimir Roland Westrelin wrote: compilers and SA support for compressed klass pointers.

http://cr.openjdk.java.net/~roland/7054512/webrev.00/ Roland.

More information about the hotspot-compiler-dev mailing list