Review request (S) 8003553: NPG: metaspace objects should be zeroed in constructors (original) (raw)

Jon Masamitsu jon.masamitsu at oracle.com
Tue Mar 5 21:26:36 PST 2013

http://cr.openjdk.java.net/~coleenp/8003553/src/share/vm/oops/cpCache.cpp.udiff.html

- ConstantPoolCache ConstantPoolCache::allocate(ClassLoaderData* loader_data, int length, TRAPS) {* + int length, + const intStack& index_map, *+ const intStack& invokedynamic_map, TRAPS) { *

Why did you not move the TRAP parameter to a new line?

http://cr.openjdk.java.net/~coleenp/8003553/src/share/vm/oops/methodData.cpp.udiff.html

Did you remove the TieredCompilation test

*if (TieredCompilation) { *

for consistency of initialization?

http://cr.openjdk.java.net/~coleenp/8003553/src/share/vm/oops/klass.hpp.udiff.html

Not concerned about redundant verifies anymore?

Rest looks good.

Jon

On 3/5/2013 11:30 AM, Coleen Phillimore wrote:

Adding hotspot-dev to get some more potential reviewers from maybe the GC team, hint... Coleen On 03/04/2013 03:02 PM, Coleen Phillimore wrote: Summary: Zero metadata in constructors, not in allocation (and some in constructors)

This seems like a good first step in passing initial values into constructors and no initializing metadata types by the callers (although a lot more parameters will have to be passed for some). Tested with runThese jck, NSK vm.quick.testlist, lang and vm jck8, java/lang/annotation jtreg tests and java/lang/invoke jtreg tests. open webrev at http://cr.openjdk.java.net/~coleenp/8003553/ bug link at http://bugs.sun.com/viewbug.do?bugid=8003553 Thanks, Coleen

More information about the hotspot-dev mailing list