[9] RFR(M): 7194669: CodeCache::mark_for_deoptimization should avoid verifying dependencies multiple times (original) (raw)

Christian Thalinger christian.thalinger at oracle.com
Tue Jan 14 16:58:17 PST 2014

Looks good.

On Jan 14, 2014, at 7:51 AM, Albert Noll <albert.noll at oracle.com> wrote:

Hi Roland,

thanks again for looking at the patch. Please see comments inline:

On 01/14/2014 01:45 PM, Roland Westrelin wrote: http://cr.openjdk.java.net/~anoll/7194669/webrev.03/ dependencies.cpp

No need to cast twice: 688 return (uintptrt)(oopDesc*)argumentoop(i); If I skip the cast to (oopDesc*), the compiler gives the following error: dependencies.cpp:688:51: error: invalid cast from type ‘oop’ to type ‘uintptrt {aka long unsigned int}’ What is that for? 747 void* DependencySignature::operator new (sizet size) throw() { 748 return NEWRESOURCEARRAY(DependencySignature, 1); 749 } I guess this is not the usual way to allocate objects into a ResourceMark. The updated version derives DependencySignature from : public ResourceObj. I hope this is fine. Shouldn’t we abort if we find something wrong? Printing stuff will easily go unnoticed. I agree. This version fails with an assert if dependency checking fails. Roland. Here is the new webrev: http://cr.openjdk.java.net/~anoll/7194669/webrev.04/ Best, Albert

More information about the hotspot-compiler-dev mailing list