Request for reviews (XXL): 6893268: additional dynamic language related optimizations in C2
John.Rose at Sun.COM
Fri Dec 18 23:00:54 PST 2009
On Dec 16, 2009, at 2:09 PM, Tom Rodriguez wrote:
> The only comment I'd make is that I don't think you need to the ResourceMark in is_in_package anymore and you might consider adding a VM_ENTRY_MARK at the top so that you only have to do one VM transition. Everything you are calling requires a VM transition and is implemented with GUARDED_VM_ENTRY. Otherwise I'm ok with it.
I made the changes you suggested on top of Christian's. I also removed the 'friend ciKlass' declaration.
But I don't think the package test was strict enough. I added checks to reject the klass if it is in a subpackage of the queried package, or if it is not loaded by the system class loader. I also moved the logic into ciInstanceKlass, where I think it fits better.
Doing the subpackage test requires an "index_of" operation in symbolOop etc.
Here's the patch; let me know what you think. I could hold back this change until I do the "trust finals" option in a later push.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 2419 bytes
Desc: not available
Url : http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20091218/a9c04b7a/attachment.zip
More information about the hotspot-compiler-dev