Nice! Thank you! On Wed, Dec 18, 2024 at 5:57 AM Martin Balao <mba...@redhat.com> wrote:
> On 12/18/24 05:25, Xuelei Fan wrote:> Hm, that would be nice if JIT > could optimize it. Did you have a chance > > to verify it? > > > > Yes. For example, to verify we can do the following: > > 1) Debug Signature::getInstance throughout the path that calls > Signature::getInstanceRSA and ProvidersFilter::isAllowed to make sure > that we are hitting the filter code (i.e. the fast-path for a disabled > filter). > > 2) Dump C2's Signature::getInstance. > > 3) Recompile the JDK removing the call to ProvidersFilter::isAllowed. > > 4) Debug again to make sure that the call was removed (i.e. the > classfile recompiled) > > 5) Dump C2's Signature::getInstance. > > 6) Compare the two Signature::getInstance dumps. > > After following the previous steps, we can see how the difference > between the methods are just addresses. Find attached to this email > Main.java to reproducer yourself (run with > '-XX:CompileCommand=compileonly,java.security.Signature::getInstance > -XX:CompileCommand=print,java.security.Signature::getInstance > -XX:-BackgroundCompilation -XX:-TieredCompilation'), and my two > ProvidersFilter::isAllowed dumps.
